Re: Announcing Redshift - Biased GPU Renderer
not AFAIK. There is a =alpha version that you can download from the forums, which an individual is developing. That would be Stefan Woermann http://vimeo.com/user2509578 That's why I was going to do the comparison in standalone. On Wed, Mar 27, 2013 at 3:24 PM, Daryl Dunlap twinsnakes...@gmail.comwrote: Ed, did Octane ever release their SI plugin? On Wed, Mar 27, 2013 at 3:23 PM, Ed Manning etmth...@gmail.com wrote: In what spare time I have I'm setting up a shootout between Octane standalone and redshift in SI. -- --- Stefan Kubicek --- keyvis digital imagery Alfred Feierfeilstraße 3 A-2380 Perchtoldsdorf bei Wien Phone:+43/699/12614231 www.keyvis.at ste...@keyvis.at -- This email and its attachments are -- --confidential and for the recipient only--
Re: Announcing Redshift - Biased GPU Renderer
Haha, thanx Christopher. That's an old keyboard i dismantled a long time ago. By the way, that site and reel are more than 3 years old now. Have just finished the new reel and while working on the new site was actually thinking whether or not to dump the keyboard thing. On Fri, Apr 5, 2013 at 3:13 AM, Christopher christop...@thecreativesheep.ca wrote: I like your keyboard graphic on your web site, very appealing :) Christopher Octavian Ureche okt...@gmail.com Thursday, April 04, 2013 11:58 AM Can't say anything about Octane because i just toyed with it. But i really enjoyed Keyshot. Even if you can't really do animation rendering with it other than it's built in srt sytem, it's a very fast CPU based raytracer. Very HDRI oriented. Can't remember if it has lights at all. For all i know i always rendered using ibl. But apart from its obvious drawbacks, it's a very simple and effective choice for product designers and the like. Of course, Reshift blows all of that to dust given it's tight app integration and gpu rendering speed. -- visual | stuff www.okto.ro Tim Crowson tim.crow...@magneticdreams.com Thursday, April 04, 2013 9:39 AM Well, if I understand correctly, RS does Distributed Monte Carlo, which is a bit different from Octane's Pathtracing. So you're not going to get an apples-to-apples comparison between the two. Now, I have Octane as well, and in my opinion, RS beats it soundly on modest hardware, both in performance and workflow. As others have said, the mere fact that RS is so well-integrated into Softimage is a BIG DEAL. I can't wait for multi-GPU support. -Tim On 4/4/2013 4:12 AM, Doeke Wartena wrote: Doeke Wartena clankil...@gmail.com Thursday, April 04, 2013 5:12 AM How is redshift compared to octane? James De Colling james.decoll...@gmail.com Thursday, April 04, 2013 3:52 AM Welcome to the pro card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price olivier jeannel olivier.jean...@noos.fr Thursday, April 04, 2013 3:26 AM I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... -- visual | stuff www.okto.ro compose-unknown-contact.jpg
Re: Announcing Redshift - Biased GPU Renderer
I'd keep the keyboard, try something new with it on the new site :) Christopher Octavian Ureche Friday, April 05, 2013 3:08 AM Haha, thanx Christopher. That's an old keyboard i dismantled a long time ago.By the way, that site and reel are more than 3 years old now.Have just finished the new reel and while working on the new site was actually thinking whether or not to dump the keyboard thing. -- visual | stuffwww.okto.ro Christopher Thursday, April 04, 2013 8:13 PM I like your keyboard graphic on your web site, very appealing :) Christopher Octavian Ureche Thursday, April 04, 2013 11:58 AM Can't say anything about Octane because i just toyed with it. But i really enjoyed Keyshot.Even if you can't really do animation rendering with it other than it's built in srt sytem, it's a very fast CPU based raytracer. Very HDRI oriented. Can't remember if it has lights at all. For all i know i always rendered using ibl.But apart from its obvious drawbacks, it's a very simple and effective choice for product designers and the like. Of course, Reshift blows all of that to dust given it's tight app integration and gpu rendering speed.-- visual | stuffwww.okto.ro Tim Crowson Thursday, April 04, 2013 9:39 AM Well, if I understand correctly, RS does Distributed Monte Carlo, which is a bit different from Octane's Pathtracing. So you're not going to get an apples-to-apples comparison between the two. Now, I have Octane as well, and in my opinion, RS beats it soundly on modest hardware, both in performance and workflow. As others have said, the mere fact that RS is so well-integrated into Softimage is a BIG DEAL. I can't wait for multi-GPU support. -Tim On 4/4/2013 4:12 AM, Doeke Wartena wrote: Doeke Wartena Thursday, April 04, 2013 5:12 AM How is redshift compared to octane?
RE: Announcing Redshift - Biased GPU Renderer
It really rocks on a dell T5500 with a quadro4000 card in it. That gets blown away by the mac book pro with the 650m on it. It really has the potential to be a total game changer for a lot of folks. You can get virtually finished look and feel at slightly above what you would currently get for a quick previz render before. Makes a massive difference in your workflow. From: Maxime Philippon [mphilippon.mailingl...@gmail.com] Sent: 04 April 2013 03:58 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer hey Guys, I'm new to this topic, I read a bit of this conversation, This Redshift GPU renderer look really awesome! I wanted to know if Redshift use the mantal ray's materials and lights or did he have his own materials and lights set up, like Arnold? And as a student, can I be an alpha tester? or is this only for professionals and studios? Thanks On Wed, Apr 3, 2013 at 8:37 PM, Christopher christop...@thecreativesheep.camailto:christop...@thecreativesheep.ca wrote: Ahhh the renderers. [cid:part1.00050907.05020202@thecreativesheep.ca] Cristobal Infantemailto:cgc...@gmail.com Wednesday, April 03, 2013 7:37 PM I just started playing with Redshift and have to say I am really impressed. Playing with on my laptop, with a GT 425M, and It still does the trick!. The combo GI and progressive rendering really is a nice combo. Can't wait to try this on a real workstation, with a full on GPU! [cid:part1.00050907.05020202@thecreativesheep.ca] Octavian Urechemailto:okt...@gmail.com Wednesday, April 03, 2013 3:44 PM Haha, we have the exact same video card. To be honest, if things keep going like this, i'll be getting another one used and put in sli. Multi-gpu support is on its way. -- visual | stuff www.okto.rohttp://www.okto.ro [cid:part1.00050907.05020202@thecreativesheep.ca] Tim Crowsonmailto:tim.crow...@magneticdreams.com Wednesday, April 03, 2013 3:30 PM I feel the same way! The only other place I've had this much fun lighting and rendering is with modo (Preview is awesome!). To be able to iterate over high-quality renders in a matter of minutes with RS is just liberating. And I'm using a lowly GTX 470! -Tim [cid:part1.00050907.05020202@thecreativesheep.ca] Octavian Urechemailto:okt...@gmail.com Wednesday, April 03, 2013 2:59 PM https://dl.dropbox.com/u/2109634/classroom_sunsky_animation.mov So here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun. Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i get some more time. But so far, i am pleased with the results, and by looking at the first and last couple of frames in the animation, you can notice the solution is stable. Also changed some settings and managed to get 2:30 min/frame on this one. Can't remember when was the last time i had so much fun rendering. [cid:part1.00050907.05020202@thecreativesheep.ca] Andreas Bystrommailto:andreas.byst...@gmail.com Wednesday, April 03, 2013 12:39 AM another thing I'm curious about with redshift is if you will get the exact same image using different hardware and possibly drivers as well? if you have a gpu renderfarm and decide to expand it a bit later you wont be able to get the same exact hardware in the new boxes, so curious to know if that would cause problems. -- Andreas Byström Weta Digital -- Max table width=100% border=0 cellspacing=0 cellpadding=0 style=width:100%; tr td align=left style=text-align:justify;font face=arial,sans-serif size=1 color=#99span style=font-size:11px;This communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorised signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary. /span/font/td /tr /table inline: compose-unknown-contact.jpg
Re: Announcing Redshift - Biased GPU Renderer
Hi Octavian, Would you share your RedShift scene ? Le 03/04/2013 20:59, Octavian Ureche a écrit : https://dl.dropbox.com/u/2109634/classroom_sunsky_animation.mov So here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun. Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i get some more time. But so far, i am pleased with the results, and by looking at the first and last couple of frames in the animation, you can notice the solution is stable. Also changed some settings and managed to get 2:30 min/frame on this one. Can't remember when was the last time i had so much fun rendering.
Re: Announcing Redshift - Biased GPU Renderer
Ooups, sorry, saw you shared it on ResdShift forum. Thank's a lot ! I just want to know where I am performance whise with that Quadro 4000 which costed me an arm... Le 04/04/2013 08:54, olivier jeannel a écrit : Hi Octavian, Would you share your RedShift scene ? Le 03/04/2013 20:59, Octavian Ureche a écrit : https://dl.dropbox.com/u/2109634/classroom_sunsky_animation.mov So here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun. Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i get some more time. But so far, i am pleased with the results, and by looking at the first and last couple of frames in the animation, you can notice the solution is stable. Also changed some settings and managed to get 2:30 min/frame on this one. Can't remember when was the last time i had so much fun rendering.
Re: Announcing Redshift - Biased GPU Renderer
Hey Olivier, Already did that. You can find both versions in the WIP section of the redshift forum, under the topic Animated Classroom with Dof and Moblur. Have fun, O On Thu, Apr 4, 2013 at 9:54 AM, olivier jeannel olivier.jean...@noos.frwrote: Hi Octavian, Would you share your RedShift scene ? Le 03/04/2013 20:59, Octavian Ureche a écrit : https://dl.dropbox.com/u/**2109634/classroom_sunsky_**animation.movhttps://dl.dropbox.com/u/2109634/classroom_sunsky_animation.mov So here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun. Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i get some more time. But so far, i am pleased with the results, and by looking at the first and last couple of frames in the animation, you can notice the solution is stable. Also changed some settings and managed to get 2:30 min/frame on this one. Can't remember when was the last time i had so much fun rendering. -- visual | stuff www.okto.ro
Re: Announcing Redshift - Biased GPU Renderer
Here you go: http://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me.
Re: Announcing Redshift - Biased GPU Renderer
I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... Le 04/04/2013 09:13, Octavian Ureche a écrit : Here you go: http://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me.
Re: Announcing Redshift - Biased GPU Renderer
Welcome to the pro card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price On Apr 4, 2013 4:28 PM, olivier jeannel olivier.jean...@noos.fr wrote: I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... Le 04/04/2013 09:13, Octavian Ureche a écrit : Here you go: http://www.nvidia.com/object/**product-quadro-4000-us.htmlhttp://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/**hardware/desktop-gpus/geforce-** gtx-470/specificationshttp://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me.
Re: Announcing Redshift - Biased GPU Renderer
How is redshift compared to octane? 2013/4/4 James De Colling james.decoll...@gmail.com Welcome to the pro card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price On Apr 4, 2013 4:28 PM, olivier jeannel olivier.jean...@noos.fr wrote: I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... Le 04/04/2013 09:13, Octavian Ureche a écrit : Here you go: http://www.nvidia.com/object/**product-quadro-4000-us.htmlhttp://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/**hardware/desktop-gpus/geforce-** gtx-470/specificationshttp://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me.
RE: Announcing Redshift - Biased GPU Renderer
And how does it compare to Luxion's Keyshothttp://www.keyshot.com/how-its-different/ or now Lagoahttps://lagoa.com/? From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Doeke Wartena Sent: 4 avril 2013 05:13 To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer How is redshift compared to octane? 2013/4/4 James De Colling james.decoll...@gmail.commailto:james.decoll...@gmail.com Welcome to the pro card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price On Apr 4, 2013 4:28 PM, olivier jeannel olivier.jean...@noos.frmailto:olivier.jean...@noos.fr wrote: I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... Le 04/04/2013 09:13, Octavian Ureche a écrit : Here you go: http://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me.
Re: Announcing Redshift - Biased GPU Renderer
for me it's all about integration with softimage, workflow first always ;) Redshift has got the edge in the sense, but I understand there is a softimage octane plugin coming soon.. On 4 April 2013 13:13, Marc-Andre Carbonneau marc-andre.carbonn...@ubisoft.com wrote: And how does it compare to Luxion’s Keyshothttp://www.keyshot.com/how-its-different/or now Lagoa https://lagoa.com/? ** ** ** ** *From:* softimage-boun...@listproc.autodesk.com [mailto: softimage-boun...@listproc.autodesk.com] *On Behalf Of *Doeke Wartena *Sent:* 4 avril 2013 05:13 *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer ** ** How is redshift compared to octane? ** ** 2013/4/4 James De Colling james.decoll...@gmail.com Welcome to the pro card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price On Apr 4, 2013 4:28 PM, olivier jeannel olivier.jean...@noos.fr wrote: I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... Le 04/04/2013 09:13, Octavian Ureche a écrit : Here you go: http://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me. ** ** ** **
Re: Announcing Redshift - Biased GPU Renderer
Title: Signature Yes, you can use standard Softimage lights, although RS has its own light primitives that may be more optimized. Haven't fully tested that yet. For a list of compatible shaders see the following two pages in the RS documentation: http://docs.redshift3d.com/Default.html#I/Supported Shaders Softimage.html http://docs.redshift3d.com/Default.html#I/Supported Shaders Mental Ray.html -Tim C. On 4/3/2013 8:58 PM, Maxime Philippon wrote: hey Guys, I'm new to this topic, I read a bit of this conversation, This Redshift GPU renderer look really awesome! I wanted to know if Redshift use the mantal ray's materials and lights or did he have his own materials and lights set up, like Arnold? And as a student, can I be an "alpha" tester? or is this only for professionals and studios? Thanks On Wed, Apr 3, 2013 at 8:37 PM, Christopher christop...@thecreativesheep.ca wrote: Ahhh the renderers. Cristobal Infante Wednesday, April 03, 2013 7:37 PM I just started playing with Redshift and have to say I am really impressed. Playing with on my laptop, with a GT 425M, and It still does the trick!. The combo GI and progressive rendering really is a nice combo. Can't wait to try this on a real workstation, with a full on GPU! Octavian Ureche Wednesday, April 03, 2013 3:44 PM Haha, we have the exact same video card. To be honest, if things keep going like this, i'll be getting another one used and put in sli. Multi-gpu support is on its way. -- visual | stuff www.okto.ro Tim Crowson Wednesday, April 03, 2013 3:30 PM I feel the same way! The only other place I've had this much fun lighting and rendering is with modo (Preview is awesome!). To be able to iterate over high-quality renders in a matter of minutes with RS is just liberating. And I'm using a lowly GTX 470! -Tim Octavian Ureche Wednesday, April 03, 2013 2:59 PM https://dl.dropbox.com/u/2109634/classroom_sunsky_animation.mov So here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun. Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i
Re: Announcing Redshift - Biased GPU Renderer
Well, if I understand correctly, RS does Distributed Monte Carlo, which is a bit different from Octane's Pathtracing. So you're not going to get an apples-to-apples comparison between the two. Now, I have Octane as well, and in my opinion, RS beats it soundly on modest hardware, both in performance and workflow. As others have said, the mere fact that RS is so well-integrated into Softimage is a BIG DEAL. I can't wait for multi-GPU support. -Tim On 4/4/2013 4:12 AM, Doeke Wartena wrote: How is redshift compared to octane? 2013/4/4 James De Colling james.decoll...@gmail.com mailto:james.decoll...@gmail.com Welcome to the pro card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price On Apr 4, 2013 4:28 PM, olivier jeannel olivier.jean...@noos.fr mailto:olivier.jean...@noos.fr wrote: I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... Le 04/04/2013 09:13, Octavian Ureche a écrit : Here you go: http://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me. -- Signature *Tim Crowson */Lead CG Artist/ *Magnetic Dreams, Inc. *2525 Lebanon Pike, Building C. Nashville, TN 37214 *Ph* 615.885.6801 | *Fax* 615.889.4768 | www.magneticdreams.com tim.crow...@magneticdreams.com /Confidentiality Notice: This email, including attachments, is confidential and should not be used by anyone who is not the original intended recipient(s). If you have received this e-mail in error please inform the sender and delete it from your mailbox or any other storage mechanism. Magnetic Dreams, Inc cannot accept liability for any statements made which are clearly the sender's own and not expressly made on behalf of Magnetic Dreams, Inc or one of its agents./
Re: Announcing Redshift - Biased GPU Renderer
Can't say anything about Octane because i just toyed with it. But i really enjoyed Keyshot. Even if you can't really do animation rendering with it other than it's built in srt sytem, it's a very fast CPU based raytracer. Very HDRI oriented. Can't remember if it has lights at all. For all i know i always rendered using ibl. But apart from its obvious drawbacks, it's a very simple and effective choice for product designers and the like. Of course, Reshift blows all of that to dust given it's tight app integration and gpu rendering speed. On Thu, Apr 4, 2013 at 4:39 PM, Tim Crowson tim.crow...@magneticdreams.comwrote: Well, if I understand correctly, RS does Distributed Monte Carlo, which is a bit different from Octane's Pathtracing. So you're not going to get an apples-to-apples comparison between the two. Now, I have Octane as well, and in my opinion, RS beats it soundly on modest hardware, both in performance and workflow. As others have said, the mere fact that RS is so well-integrated into Softimage is a BIG DEAL. I can't wait for multi-GPU support. -Tim On 4/4/2013 4:12 AM, Doeke Wartena wrote: How is redshift compared to octane? 2013/4/4 James De Colling james.decoll...@gmail.com Welcome to the pro card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price On Apr 4, 2013 4:28 PM, olivier jeannel olivier.jean...@noos.fr wrote: I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though... Le 04/04/2013 09:13, Octavian Ureche a écrit : Here you go: http://www.nvidia.com/object/product-quadro-4000-us.html http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-470/specifications The last one is what i currently have. as you can see, the quadro's memory bandwith, cuda cores and memory interface are below the gtx. But you have bigger vram which means you can cram more into the scenes. Speed wise, given the differences it might be slower at the actual rendering, but we're talking gpu rendering here so i'm not sure if it's going to be that much noticeable. Do a render with both scenes and post your times in the forum. Then we'll know better how hardware affects the performance. To be honest, i always found quadros to be extremely overpriced, but maybe that's just me. -- *Tim Crowson **Lead CG Artist* *Magnetic Dreams, Inc. *2525 Lebanon Pike, Building C. Nashville, TN 37214 *Ph* 615.885.6801 | *Fax* 615.889.4768 | www.magneticdreams.com tim.crow...@magneticdreams.com *Confidentiality Notice: This email, including attachments, is confidential and should not be used by anyone who is not the original intended recipient(s). If you have received this e-mail in error please inform the sender and delete it from your mailbox or any other storage mechanism. Magnetic Dreams, Inc cannot accept liability for any statements made which are clearly the sender's own and not expressly made on behalf of Magnetic Dreams, Inc or one of its agents.* -- visual | stuff www.okto.ro
Re: Announcing Redshift - Biased GPU Renderer
I like your keyboard graphic on your web site, very appealing :) Christopher Octavian Ureche Thursday, April 04, 2013 11:58 AM Can't say anything about Octane because i just toyed with it. But i really enjoyed Keyshot.Even if you can't really do animation rendering with it other than it's built in srt sytem, it's a very fast CPU based raytracer. Very HDRI oriented. Can't remember if it has lights at all. For all i know i always rendered using ibl.But apart from its obvious drawbacks, it's a very simple and effective choice for product designers and the like. Of course, Reshift blows all of that to dust given it's tight app integration and gpu rendering speed.-- visual | stuffwww.okto.ro Tim Crowson Thursday, April 04, 2013 9:39 AM Well, if I understand correctly, RS does Distributed Monte Carlo, which is a bit different from Octane's Pathtracing. So you're not going to get an apples-to-apples comparison between the two. Now, I have Octane as well, and in my opinion, RS beats it soundly on modest hardware, both in performance and workflow. As others have said, the mere fact that RS is so well-integrated into Softimage is a BIG DEAL. I can't wait for multi-GPU support. -Tim On 4/4/2013 4:12 AM, Doeke Wartena wrote: Doeke Wartena Thursday, April 04, 2013 5:12 AM How is redshift compared to octane? James De Colling Thursday, April 04, 2013 3:52 AM Welcome to the "pro" card market... I only use quadros because that's what the sells we use ship with... Long gone are the days when people cards were worth their sticker price olivier jeannel Thursday, April 04, 2013 3:26 AM I should be out of the office, but will test asap. For the quadro, well it was bundled with the workstation (HP Z620). It's no problem if the quadro is more expensive and produce better performance. It becomes a problem if they are really bellow game cards. Your gtx has more than 400 cores while the 4000 has 256... They are supposed to be stronger when working though...
Re: Announcing Redshift - Biased GPU Renderer
https://dl.dropbox.com/u/2109634/classroom_sunsky_animation.mov So here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun. Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i get some more time. But so far, i am pleased with the results, and by looking at the first and last couple of frames in the animation, you can notice the solution is stable. Also changed some settings and managed to get 2:30 min/frame on this one. Can't remember when was the last time i had so much fun rendering.
Re: Announcing Redshift - Biased GPU Renderer
I feel the same way! The only other place I've had this much fun lighting and rendering is with modo (Preview is awesome!). To be able to iterate over high-quality renders in a matter of minutes with RS is just liberating. And I'm using a lowly GTX 470! -Tim On 4/3/2013 1:59 PM, Octavian Ureche wrote: Can't remember when was the last time i had so much fun rendering. -- Signat
Re: Announcing Redshift - Biased GPU Renderer
Haha, we have the exact same video card. To be honest, if things keep going like this, i'll be getting another one used and put in sli. Multi-gpu support is on its way. On Wed, Apr 3, 2013 at 10:30 PM, Tim Crowson tim.crow...@magneticdreams.com wrote: I feel the same way! The only other place I've had this much fun lighting and rendering is with modo (Preview is awesome!). To be able to iterate over high-quality renders in a matter of minutes with RS is just liberating. And I'm using a lowly GTX 470! -Tim On 4/3/2013 1:59 PM, Octavian Ureche wrote: Can't remember when was the last time i had so much fun rendering. -- Signat -- visual | stuff www.okto.ro
Re: Announcing Redshift - Biased GPU Renderer
I just started playing with Redshift and have to say I am really impressed. Playing with on my laptop, with a GT 425M, and It still does the trick!. The combo GI and progressive rendering really is a nice combo. Can't wait to try this on a real workstation, with a full on GPU! On 3 April 2013 20:44, Octavian Ureche okt...@gmail.com wrote: Haha, we have the exact same video card. To be honest, if things keep going like this, i'll be getting another one used and put in sli. Multi-gpu support is on its way. On Wed, Apr 3, 2013 at 10:30 PM, Tim Crowson tim.crow...@magneticdreams.com wrote: I feel the same way! The only other place I've had this much fun lighting and rendering is with modo (Preview is awesome!). To be able to iterate over high-quality renders in a matter of minutes with RS is just liberating. And I'm using a lowly GTX 470! -Tim On 4/3/2013 1:59 PM, Octavian Ureche wrote: Can't remember when was the last time i had so much fun rendering. -- Signat -- visual | stuff www.okto.ro
Re: Announcing Redshift - Biased GPU Renderer
Ahhh the renderers. Cristobal Infante Wednesday, April 03, 2013 7:37 PM I just started playing with Redshift and have to say I am really impressed.Playing with on my laptop, with a GT 425M, and It still does the trick!. The combo GI and progressive rendering really is a nice combo. Can't wait to try this on a real workstation, with a full on GPU! Octavian Ureche Wednesday, April 03, 2013 3:44 PM Haha, we have the exact same video card.To be honest, if things keep going like this, i'll be getting another one used and put in sli.Multi-gpu support is on its way. -- visual | stuffwww.okto.ro Tim Crowson Wednesday, April 03, 2013 3:30 PM I feel the same way! The only other place I've had this much fun lighting and rendering is with modo (Preview is awesome!). To be able to iterate over high-quality renders in a matter of minutes with RS is just liberating. And I'm using a lowly GTX 470! -Tim Octavian Ureche Wednesday, April 03, 2013 2:59 PM https://dl.dropbox.com/u/2109634/classroom_sunsky_animation.movSo here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun.Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i get some more time.But so far, i am pleased with the results, and by looking at the first and last couple of frames in the animation, you can notice the solution is stable. Also changed some settings and managed to get 2:30 min/frame on this one.Can't remember when was the last time i had so much fun rendering. Andreas Bystrom Wednesday, April 03, 2013 12:39 AM another thing I'm curious about with redshift is if you will get the exact same image using different hardware and possibly drivers as well?if you have a gpu renderfarm and decide to expand it a bit later you wont be able to get the same exact hardware in the new boxes, so curious to know if that would cause problems. -- Andreas BystrmWeta Digital
Re: Announcing Redshift - Biased GPU Renderer
hey Guys, I'm new to this topic, I read a bit of this conversation, This Redshift GPU renderer look really awesome! I wanted to know if Redshift use the mantal ray's materials and lights or did he have his own materials and lights set up, like Arnold? And as a student, can I be an alpha tester? or is this only for professionals and studios? Thanks On Wed, Apr 3, 2013 at 8:37 PM, Christopher christop...@thecreativesheep.ca wrote: Ahhh the renderers. Cristobal Infante cgc...@gmail.com Wednesday, April 03, 2013 7:37 PM I just started playing with Redshift and have to say I am really impressed. Playing with on my laptop, with a GT 425M, and It still does the trick!. The combo GI and progressive rendering really is a nice combo. Can't wait to try this on a real workstation, with a full on GPU! Octavian Ureche okt...@gmail.com Wednesday, April 03, 2013 3:44 PM Haha, we have the exact same video card. To be honest, if things keep going like this, i'll be getting another one used and put in sli. Multi-gpu support is on its way. -- visual | stuff www.okto.ro Tim Crowson tim.crow...@magneticdreams.com Wednesday, April 03, 2013 3:30 PM I feel the same way! The only other place I've had this much fun lighting and rendering is with modo (Preview is awesome!). To be able to iterate over high-quality renders in a matter of minutes with RS is just liberating. And I'm using a lowly GTX 470! -Tim Octavian Ureche okt...@gmail.com Wednesday, April 03, 2013 2:59 PM https://dl.dropbox.com/u/2109634/classroom_sunsky_animation.mov So here's another test with that classroom scene. This time without dof and moblur but with an abruptly animated physical sun. Looking at the overall render, i think it looks good. I know some will jump and say it's too fast, which is why i'll probably render it again with a slower motion of the light when i get some more time. But so far, i am pleased with the results, and by looking at the first and last couple of frames in the animation, you can notice the solution is stable. Also changed some settings and managed to get 2:30 min/frame on this one. Can't remember when was the last time i had so much fun rendering. Andreas Bystrom andreas.byst...@gmail.com Wednesday, April 03, 2013 12:39 AM another thing I'm curious about with redshift is if you will get the exact same image using different hardware and possibly drivers as well? if you have a gpu renderfarm and decide to expand it a bit later you wont be able to get the same exact hardware in the new boxes, so curious to know if that would cause problems. -- Andreas Byström Weta Digital -- Max compose-unknown-contact.jpg
Re: Announcing Redshift - Biased GPU Renderer
CPU-based renderers can render scenes that fit into RAM, while GPU based ones traditionally only have fast access to VRAM, which usually is not more than 2Gb on most cards. Redshift have overcome this limitation, they do most of the processing on the GPU and can still render scenes that need more RAM than what fits into the cards VRAM (Out of core is becoming an increasingly popular paradigm, it's been mentioned a lot in recent talks and papers from Nvidia too). What I havent understood yet is in how far Redshifts lead in this area is based on their own inovation and to what extent it is influenced by NVIDIA middleware (it's CUDA only after all atm). The more of the latter might allow other renderers to follow qickly. Also, the whole memory bottle neck problem would not be one if GPUs had direct access to CPU RAM. Coincidentally, AMD is making the APUs for the new PS4 with exactly such functionality (they even allow the GPU to directly access memory in the 386 adress space directly). It will be interesting to see what happens when such Silicon becomes available to mainstream computing. On 02.04.2013, at 01:01, Doeke Wartena clankil...@gmail.com wrote: in what way is it moreeffective? 2013/4/1 Octavian Ureche okt...@gmail.com yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.com wrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.com wrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise From: okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. mime-attachment.jpg -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- Andreas Byström Weta Digital -- visual | stuff www.okto.ro
Re: Announcing Redshift - Biased GPU Renderer
Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.comwrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.comwrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise -- From: okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. [image: Inline image 1] -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- Andreas Byström Weta Digital
Re: Announcing Redshift - Biased GPU Renderer
Speaking of the wolf Was just getting ready to post it. So here it is: https://dl.dropbox.com/u/2109634/classroom_dof_moblur_animation_v02.mov A couple of notes on it though. It had around 3 min / frame (some frames i saw 2:40 min). The thing is, i'm using brute force for the primary rays, since i'm still trying to understand the engine, and it's the slowest approach of all. Also i doubled the rays since the still image to make sure it looks neat (someone mentioned noise for that one), so now it's 1024 rays. Another thing i did was to lower the screen radius to 8 on the IPC and raise the samples per pixel to 64. Kept a pretty low setting on the dof (128 samples), and put a higher sampling on the moblur (512). That's why, if you look frame by frame, you will see some noise in the dof. All in all, given that, with proper knowledge of the engine and a different primary ray approach like IC, one could surely take the rendertime down, i'm still impressed by a noiseless brute force solution that does dof and moblur in under 3 mins/frame. Oh, and i have a 3 year old gtx470 with 1 gb vram. And i just started using redshift yesterday :) Cheers, Octav On Tue, Apr 2, 2013 at 9:23 PM, Tim Leydecker bauero...@gmx.de wrote: Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.com**wrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.com wrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise -- From: okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.**comsoftimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. [image: Inline image 1] -- __**___ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- Andreas Byström Weta Digital -- visual | stuff www.okto.ro
Re: Announcing Redshift - Biased GPU Renderer
Very nice. I want to use Redshift3d now, too. mental ray has already wasted too much of my life time. Cheers, tim On 02.04.2013 20:37, Octavian Ureche wrote: Speaking of the wolf Was just getting ready to post it. So here it is: https://dl.dropbox.com/u/2109634/classroom_dof_moblur_animation_v02.mov A couple of notes on it though. It had around 3 min / frame (some frames i saw 2:40 min). The thing is, i'm using brute force for the primary rays, since i'm still trying to understand the engine, and it's the slowest approach of all. Also i doubled the rays since the still image to make sure it looks neat (someone mentioned noise for that one), so now it's 1024 rays. Another thing i did was to lower the screen radius to 8 on the IPC and raise the samples per pixel to 64. Kept a pretty low setting on the dof (128 samples), and put a higher sampling on the moblur (512). That's why, if you look frame by frame, you will see some noise in the dof. All in all, given that, with proper knowledge of the engine and a different primary ray approach like IC, one could surely take the rendertime down, i'm still impressed by a noiseless brute force solution that does dof and moblur in under 3 mins/frame. Oh, and i have a 3 year old gtx470 with 1 gb vram. And i just started using redshift yesterday :) Cheers, Octav On Tue, Apr 2, 2013 at 9:23 PM, Tim Leydecker bauero...@gmx.de wrote: Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.com**wrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.com wrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise -- From: okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.**comsoftimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. [image: Inline image 1] -- __**___ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- Andreas Byström Weta Digital
Re: Announcing Redshift - Biased GPU Renderer
Just to put some salt on that wound...did i mention i was archiving a project and reading my mails while this was rendering? It's the weirdest feeling in the world to render something, and then look at the processor threads and see them all on idle. On Tue, Apr 2, 2013 at 9:46 PM, Tim Leydecker bauero...@gmx.de wrote: Very nice. I want to use Redshift3d now, too. mental ray has already wasted too much of my life time. Cheers, tim On 02.04.2013 20:37, Octavian Ureche wrote: Speaking of the wolf Was just getting ready to post it. So here it is: https://dl.dropbox.com/u/**2109634/classroom_dof_moblur_** animation_v02.movhttps://dl.dropbox.com/u/2109634/classroom_dof_moblur_animation_v02.mov A couple of notes on it though. It had around 3 min / frame (some frames i saw 2:40 min). The thing is, i'm using brute force for the primary rays, since i'm still trying to understand the engine, and it's the slowest approach of all. Also i doubled the rays since the still image to make sure it looks neat (someone mentioned noise for that one), so now it's 1024 rays. Another thing i did was to lower the screen radius to 8 on the IPC and raise the samples per pixel to 64. Kept a pretty low setting on the dof (128 samples), and put a higher sampling on the moblur (512). That's why, if you look frame by frame, you will see some noise in the dof. All in all, given that, with proper knowledge of the engine and a different primary ray approach like IC, one could surely take the rendertime down, i'm still impressed by a noiseless brute force solution that does dof and moblur in under 3 mins/frame. Oh, and i have a 3 year old gtx470 with 1 gb vram. And i just started using redshift yesterday :) Cheers, Octav On Tue, Apr 2, 2013 at 9:23 PM, Tim Leydecker bauero...@gmx.de wrote: Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.comwrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.com wrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise -- From: okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.comsoftimage@listproc.** autodesk.com softimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. [image: Inline image 1] -- _____ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- Andreas Byström Weta Digital -- visual | stuff www.okto.ro
RE: Announcing Redshift - Biased GPU Renderer
Appreciated! But the amount of mb and the rapid camera pan in that movie makes any flicker unnoticable even it was there ;) Could you render it without mb, just *slow* camera animation and moving lights so we can see possible flickering (or lack thereof)? Thanks man, sven From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Octavian Ureche Sent: Tuesday, April 02, 2013 20:37 To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Speaking of the wolf Was just getting ready to post it. So here it is: https://dl.dropbox.com/u/2109634/classroom_dof_moblur_animation_v02.mov A couple of notes on it though. It had around 3 min / frame (some frames i saw 2:40 min). The thing is, i'm using brute force for the primary rays, since i'm still trying to understand the engine, and it's the slowest approach of all. Also i doubled the rays since the still image to make sure it looks neat (someone mentioned noise for that one), so now it's 1024 rays. Another thing i did was to lower the screen radius to 8 on the IPC and raise the samples per pixel to 64. Kept a pretty low setting on the dof (128 samples), and put a higher sampling on the moblur (512). That's why, if you look frame by frame, you will see some noise in the dof. All in all, given that, with proper knowledge of the engine and a different primary ray approach like IC, one could surely take the rendertime down, i'm still impressed by a noiseless brute force solution that does dof and moblur in under 3 mins/frame. Oh, and i have a 3 year old gtx470 with 1 gb vram. And i just started using redshift yesterday :) Cheers, Octav On Tue, Apr 2, 2013 at 9:23 PM, Tim Leydecker bauero...@gmx.de wrote: Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.comwrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.comwrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise -- From: okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. [image: Inline image 1] -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- Andreas Byström Weta Digital -- visual | stuff www.okto.ro
Re: Announcing Redshift - Biased GPU Renderer
I agree that it makes it difficult to spot any flickering with a fast moving camera, but it was enough to see that if there was any, that it would be minimal if there was at any at all.. Especially that brute force was used.. meaning flickering should be a non-issue anyways no? We should actually be looking for noise in this case, and there doesn't seem to be too much of that either.. Would still like to see a slow moving shot with perhaps a moving sun, using the aproximation methods that is.. but 3min using brute force :o Thanks for that though! On 02/04/2013 5:06 PM, Sven Constable wrote: Appreciated! But the amount of mb and the rapid camera pan in that movie makes any flicker unnoticable even it was there ;) Could you render it without mb, just *slow* camera animation and moving lights so we can see possible flickering (or lack thereof)? Thanks man, sven *From:* softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] *On Behalf Of *Octavian Ureche *Sent:* Tuesday, April 02, 2013 20:37 *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Speaking of the wolf Was just getting ready to post it. So here it is: https://dl.dropbox.com/u/2109634/classroom_dof_moblur_animation_v02.mov A couple of notes on it though. It had around 3 min / frame (some frames i saw 2:40 min). The thing is, i'm using brute force for the primary rays, since i'm still trying to understand the engine, and it's the slowest approach of all. Also i doubled the rays since the still image to make sure it looks neat (someone mentioned noise for that one), so now it's 1024 rays. Another thing i did was to lower the screen radius to 8 on the IPC and raise the samples per pixel to 64. Kept a pretty low setting on the dof (128 samples), and put a higher sampling on the moblur (512). That's why, if you look frame by frame, you will see some noise in the dof. All in all, given that, with proper knowledge of the engine and a different primary ray approach like IC, one could surely take the rendertime down, i'm still impressed by a noiseless brute force solution that does dof and moblur in under 3 mins/frame. Oh, and i have a 3 year old gtx470 with 1 gb vram. And i just started using redshift yesterday :) Cheers, Octav On Tue, Apr 2, 2013 at 9:23 PM, Tim Leydecker bauero...@gmx.de mailto:bauero...@gmx.de wrote: Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.com mailto:andreas.byst...@gmail.comwrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.com mailto:clankil...@gmail.comwrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca mailto:l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise -- From: okt...@gmail.com mailto:okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.com mailto:softimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav
Re: Announcing Redshift - Biased GPU Renderer
Right primary rays were brute force sorry, Moving lights making small bright lit patches lighting the rest of the room is what makes flickering the most prone. For instance, a slowly moving sun (and camera), with small light portals (perhaps a punched grid running across windows?) just as the sun starts to peirce through after high noon, would be a very tough flicker test. Still cant get over those numbers for something with (all very wide) DOF, MB glossy reflections. On 02/04/2013 5:47 PM, Sven Constable wrote: Its not bruce force, its biased. So flickering can be an issue. To see it, we need a movie that predestines a scenario critical to flickering with biased renderers. *From:* softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] *On Behalf Of *Jason S *Sent:* Tuesday, April 02, 2013 23:21 *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer I agree that it makes it difficult to spot any flickering with a fast moving camera, but it was enough to see that if there was any, that it would be minimal if there was at any at all.. Especially that brute force was used.. meaning flickering should be a non-issue anyways no? We should actually be looking for noise in this case, and there doesn't seem to be too much of that either.. Would still like to see a slow moving shot with perhaps a moving sun, using the aproximation methods that is.. but 3min using brute force :o Thanks for that though! On 02/04/2013 5:06 PM, Sven Constable wrote: Appreciated! But the amount of mb and the rapid camera pan in that movie makes any flicker unnoticable even it was there ;) Could you render it without mb, just *slow* camera animation and moving lights so we can see possible flickering (or lack thereof)? Thanks man, sven *From:* softimage-boun...@listproc.autodesk.com mailto:softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] *On Behalf Of *Octavian Ureche *Sent:* Tuesday, April 02, 2013 20:37 *To:* softimage@listproc.autodesk.com mailto:softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Speaking of the wolf Was just getting ready to post it. So here it is: https://dl.dropbox.com/u/2109634/classroom_dof_moblur_animation_v02.mov A couple of notes on it though. It had around 3 min / frame (some frames i saw 2:40 min). The thing is, i'm using brute force for the primary rays, since i'm still trying to understand the engine, and it's the slowest approach of all. Also i doubled the rays since the still image to make sure it looks neat (someone mentioned noise for that one), so now it's 1024 rays. Another thing i did was to lower the screen radius to 8 on the IPC and raise the samples per pixel to 64. Kept a pretty low setting on the dof (128 samples), and put a higher sampling on the moblur (512). That's why, if you look frame by frame, you will see some noise in the dof. All in all, given that, with proper knowledge of the engine and a different primary ray approach like IC, one could surely take the rendertime down, i'm still impressed by a noiseless brute force solution that does dof and moblur in under 3 mins/frame. Oh, and i have a 3 year old gtx470 with 1 gb vram. And i just started using redshift yesterday :) Cheers, Octav On Tue, Apr 2, 2013 at 9:23 PM, Tim Leydecker bauero...@gmx.de mailto:bauero...@gmx.de wrote: Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.com mailto:andreas.byst...@gmail.comwrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.com mailto:clankil...@gmail.comwrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca mailto:l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I
Re: Announcing Redshift - Biased GPU Renderer
I got drunk with a couple of Icelandish friends instead. That helped quite a bit. Still rendered a F-stop animation of the classroom scene with DOF just to find out that the DOF is jittering badly when I came home. I used mental ray. I´m totally fed up with this now. Thanks for sharing your testsequence. At least I can switch to an Arnold Pass for comparison and have really nice, believable lightdistribution, even if it is currently still a tough punch to the rendertimes. Am looking forward to the slow-mo test but am already deeply impressed. The Redshift guys didn´t reply to my alpha request, so I guess I´ll have to wait that one out. Cheers, tim On 02.04.2013 20:59, Octavian Ureche wrote: Just to put some salt on that wound...did i mention i was archiving a project and reading my mails while this was rendering? It's the weirdest feeling in the world to render something, and then look at the processor threads and see them all on idle. On Tue, Apr 2, 2013 at 9:46 PM, Tim Leydecker bauero...@gmx.de wrote: Very nice. I want to use Redshift3d now, too. mental ray has already wasted too much of my life time. Cheers, tim On 02.04.2013 20:37, Octavian Ureche wrote: Speaking of the wolf Was just getting ready to post it. So here it is: https://dl.dropbox.com/u/**2109634/classroom_dof_moblur_** animation_v02.movhttps://dl.dropbox.com/u/2109634/classroom_dof_moblur_animation_v02.mov A couple of notes on it though. It had around 3 min / frame (some frames i saw 2:40 min). The thing is, i'm using brute force for the primary rays, since i'm still trying to understand the engine, and it's the slowest approach of all. Also i doubled the rays since the still image to make sure it looks neat (someone mentioned noise for that one), so now it's 1024 rays. Another thing i did was to lower the screen radius to 8 on the IPC and raise the samples per pixel to 64. Kept a pretty low setting on the dof (128 samples), and put a higher sampling on the moblur (512). That's why, if you look frame by frame, you will see some noise in the dof. All in all, given that, with proper knowledge of the engine and a different primary ray approach like IC, one could surely take the rendertime down, i'm still impressed by a noiseless brute force solution that does dof and moblur in under 3 mins/frame. Oh, and i have a 3 year old gtx470 with 1 gb vram. And i just started using redshift yesterday :) Cheers, Octav On Tue, Apr 2, 2013 at 9:23 PM, Tim Leydecker bauero...@gmx.de wrote: Hi Octavian, is an update/sequence render of the (animated) classroom scene available already? Would be really interesting how the DOF/MoB and GI play together with animation and how long it takes to get the results smooth across frames. Cheers, tim On 01.04.2013 23:37, Octavian Ureche wrote: yap, i have some time to kill tomorrow so i'll give it a go. see know how it turns out On Tue, Apr 2, 2013 at 12:29 AM, Andreas Bystrom andreas.byst...@gmail.comwrote: octavian, could you render a small animation with that exact setup? with say a camera move and some animated objects inside the room? On Tue, Apr 2, 2013 at 8:11 AM, Doeke Wartena clankil...@gmail.com wrote: Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. 2013/4/1 Len Krenzler l...@creativecontrol.ca It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times! But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise -- From: okt...@gmail.com Date: Mon, 1 Apr 2013 19:17:32 +0300 Subject: Re: Announcing Redshift - Biased GPU Renderer To: softimage@listproc.autodesk.comsoftimage@listproc.** autodesk.com softimage@listproc.autodesk.com Crossposting and a little OT but i just had to share this. Took some time today and finally fiddled a bit with redshift. 1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer. I really dig it so far. Cheers, Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. [image: Inline image 1] -- _____ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- Andreas Byström Weta Digital
Re: Announcing Redshift - Biased GPU Renderer
CPU render is more effective over GPU, even though it can be slower. An example is look at the render farms for recent CG films, there huge mostly CPU based. Doeke Wartena Monday, April 01, 2013 3:11 PM Can someone tell me why so many renderers are CPU based? And what is the up and downside apart from speed. Len Krenzler Monday, April 01, 2013 2:55 PM It is a fantastic render engine. That grain can easily be removed by a little tweaking and not much more render time. - Len On 4/1/2013 12:49 PM, Andres Stephens wrote: -- _Len Krenzler - Creative Control Media ProductionsPhone: 780.463.3126www.creativecontrol.ca - l...@creativecontrol.ca Andres Stephens Monday, April 01, 2013 2:49 PM Wow, I got access to the Alpha, and I'm really digging it also! But I haven't got a sample scene to benchmark yet. But I like what you've got there, and great times!But.. are you happy with the grain in the image? Thanks for sharing the image. =) -Draise Octavian Ureche Monday, April 01, 2013 12:17 PM Crossposting and a little OT but i just had to share this.Took some time today and finally fiddled a bit with redshift.1:41 mins on a gtx470 with the old classroom scene (10 min for material setup, 1 hr to figure out the settings). Dof and motionblur straight from the renderer.I really dig it so far.Cheers,Octav PS.and i managed to finish the vray displacement test scene which i have to cleanup and share later today. Raffaele Fragapane Wednesday, March 27, 2013 7:44 PM While I'm not a huge fan of Anand, they do occasionally have a good article out.http://www.anandtech.com/show/6774/nvidias-geforce-gtx-titan-part-2-titans-performance-unveiled/3 This sheds some light on what you might be asking about, and on why some times you hear that the 580s are doing better than the 680s and why the latter is considered a crippled card for professional use.It does omit the fact that the (factory) OCed premium 680s, especially with the memory clocked higher, actually go up a fair chunk, and that if you have a 680 that hits 1400 then some of those tests, especially short span ones where Titan's turbo doesn't have the time to kick in, will actually see the 680 taking the lead over the titan in both numbers and power usage. Only benching I've done was CUDA and number crunching related because I've taken an interest in it a while ago and still toy with it on and off, and that includes the generic GEMM and FFT tests.I don't bother with game benchmarks or 3DMark or cinebench, but single precision the 680 stock cooled but OCed was constantly bang-on on par with the titan for a lower power draw. Double precision even OCed it (680) will fall back a fair chunk, and water cooled OCed 580s actually take the lead in bang for buck by a mile, but have horrible (high) power draw.You can consider the k5000 somewhat closer to the titan than to the 680. -- Our users will know fear and cower before our software! Ship it! Ship it and let them flee like the dogs they are!
Re: Announcing Redshift - Biased GPU Renderer
The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.comwrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA for a while in number crunching, the main problem is the drivers are still a coin toss chance, and that OCL isn't anywhere as popular as CUDA. With win7 or 8 and recent versions of Soft/Maya they can do well. nVIDIA didn't help with the crippling of the 6xx for professional use, and pissing off Linus. They are still ahead by a slight margin, for now, but I wouldn't discount AMD wholesale anymore. If the next generation is as disappointing as Kepler is, and AMD gets both Linux support AND decent (and properly OSS) drivers out, I'm moving time come for the next upgrade. For now I recently bought a 680 because it was kind of mandatory to not go insane with Mari and Mudbox, and because I like CUDA and I toy with it at home. On Wed, Mar 20, 2013 at 9:58 PM, Dan Yargici danyarg...@gmail.comwrote: Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... Really? I don't remember anyone ever suggesting ATI was anything other than shit! DAN
Re: Announcing Redshift - Biased GPU Renderer
On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.com**wrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA for a while in number crunching, the main problem is the drivers are still a coin toss chance, and that OCL isn't anywhere as popular as CUDA. With win7 or 8 and recent versions of Soft/Maya they can do well. nVIDIA didn't help with the crippling of the 6xx for professional use, and pissing off Linus. They are still ahead by a slight margin, for now, but I wouldn't discount AMD wholesale anymore. If the next generation is as disappointing as Kepler is, and AMD gets both Linux support AND decent (and properly OSS) drivers out, I'm moving time come for the next upgrade. For now I recently bought a 680 because it was kind of mandatory to not go insane with Mari and Mudbox, and because I like CUDA and I toy with it at home. On Wed, Mar 20, 2013 at 9:58 PM, Dan Yargici danyarg...@gmail.com wrote: Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... Really? I don't
Re: Announcing Redshift - Biased GPU Renderer
Personally, I´m hesistant to using two or more cards with SLI because of micro stuttering: http://en.wikipedia.org/wiki/Micro_stuttering If there would be a solution to that, I´d go with two GTX670 w/4GB VRAM, as they are the same GK104´s with a 915MHz chipspeed instead of a 1006Mhz chipspeed as in the reference design GTX680. That could save another 15-35% percent of investment compared to two single chip GTX 680 cards or one GTX Titan. Overclocked versions may use slightly different chip/shader speeds. In any case, as much VRAM as available, as that always helps in many progamms like Mudbox, Redshift and isn´t much of an added cost (comparing 2GB vs 4GB). At a company I worked Mari 1.5.x behaved bitchy unless it was given a Quadro or forced to ignore the actual card´s game heritage. But that may have been solved with 2.0... Cheers, tim On 27.03.2013 08:59, Mirko Jankovic wrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.com**wrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA for a while in number crunching, the main problem is the drivers are still a coin toss chance, and that OCL isn't
Re: Announcing Redshift - Biased GPU Renderer
That's exactly what I'm eager for, having multiple cards in (linked by sli or not) participating in the render. Huge bang for buck potential. Ben -- Benjamin Clifford Davis www.moondog-animation.com office: +33 9 50 04 76 15 mobile: +33 6 88 48 54 50 6 bis avenue des Iles 74000 Annecy FRANCE On Wed, Mar 27, 2013 at 8:59 AM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.com**wrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA for a while in number crunching, the main problem is the drivers are still a coin toss chance, and that OCL isn't anywhere as popular as CUDA. With win7 or 8 and recent versions of Soft/Maya they can do well. nVIDIA didn't help with the crippling of the 6xx for professional use, and pissing off Linus. They are still ahead by a slight margin, for now, but I wouldn't discount AMD wholesale anymore. If the next generation is as disappointing as Kepler is, and AMD gets both Linux support AND decent (and properly OSS) drivers out, I'm moving
Re: Announcing Redshift - Biased GPU Renderer
I don't think micro stuttering would be a terrible issue as far as GPU rendering goes, it's mostly a frustrating drawback as far as framerates being slightly crippled in gameplay, no? -- Benjamin Clifford Davis www.moondog-animation.com office: +33 9 50 04 76 15 mobile: +33 6 88 48 54 50 6 bis avenue des Iles 74000 Annecy FRANCE On Wed, Mar 27, 2013 at 9:26 AM, Tim Leydecker bauero...@gmx.de wrote: Personally, I´m hesistant to using two or more cards with SLI because of micro stuttering: http://en.wikipedia.org/wiki/** Micro_stuttering http://en.wikipedia.org/wiki/Micro_stuttering If there would be a solution to that, I´d go with two GTX670 w/4GB VRAM, as they are the same GK104´s with a 915MHz chipspeed instead of a 1006Mhz chipspeed as in the reference design GTX680. That could save another 15-35% percent of investment compared to two single chip GTX 680 cards or one GTX Titan. Overclocked versions may use slightly different chip/shader speeds. In any case, as much VRAM as available, as that always helps in many progamms like Mudbox, Redshift and isn´t much of an added cost (comparing 2GB vs 4GB). At a company I worked Mari 1.5.x behaved bitchy unless it was given a Quadro or forced to ignore the actual card´s game heritage. But that may have been solved with 2.0... Cheers, tim On 27.03.2013 08:59, Mirko Jankovic wrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps
Re: Announcing Redshift - Biased GPU Renderer
I don´t know how the setup of two or more cards would be best done for GPU rendering purposes but I would at least try to enable SLI to get the best framerate/redraw performance in general applications and games to better justify the investment. But maybe, if I don´t have to bother about it and just see the GPU renderer pick up all the available cards I´d be just as happy with the increased renderspeed and possibilities this gives. In general I´m most likely hesistant as I´ve been burnt by things like mR unified sampling messing up framebuffers (in xsi2012/mR 3.9.x) or not correctly supporting satellite rendering (in xsi2012/mR 3.9.x) and seing those flaws eat up the initial benefit I had hoped for to some extend. Anyone using VRay RT on Maya or mentalray´s iRay here and able to supply info? Cheers, tim On 27.03.2013 09:34, Ben Davis wrote: I don't think micro stuttering would be a terrible issue as far as GPU rendering goes, it's mostly a frustrating drawback as far as framerates being slightly crippled in gameplay, no? -- Benjamin Clifford Davis www.moondog-animation.com office: +33 9 50 04 76 15 mobile: +33 6 88 48 54 50 6 bis avenue des Iles 74000 Annecy FRANCE On Wed, Mar 27, 2013 at 9:26 AM, Tim Leydecker bauero...@gmx.de wrote: Personally, I´m hesistant to using two or more cards with SLI because of micro stuttering: http://en.wikipedia.org/wiki/** Micro_stuttering http://en.wikipedia.org/wiki/Micro_stuttering If there would be a solution to that, I´d go with two GTX670 w/4GB VRAM, as they are the same GK104´s with a 915MHz chipspeed instead of a 1006Mhz chipspeed as in the reference design GTX680. That could save another 15-35% percent of investment compared to two single chip GTX 680 cards or one GTX Titan. Overclocked versions may use slightly different chip/shader speeds. In any case, as much VRAM as available, as that always helps in many progamms like Mudbox, Redshift and isn´t much of an added cost (comparing 2GB vs 4GB). At a company I worked Mari 1.5.x behaved bitchy unless it was given a Quadro or forced to ignore the actual card´s game heritage. But that may have been solved with 2.0... Cheers, tim On 27.03.2013 08:59, Mirko Jankovic wrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards
Re: Announcing Redshift - Biased GPU Renderer
There was a subject on Redshift Forum about having two grapphic cards. It seems to be possible to keep a quadro for dispaly (as it is significantly better at displaying), and have a Titan dedicated to rendering only (in Redshift you select which card is rendering) as they have a huge amount of cores and faster memory. I think I've red somewhere that Titan has 2600 cores against 256 for the Quadro 4000. After chating with Nicolas the Titan could be around 4 time faster than the Quadro4000 ...Which is huge :) Le 27/03/2013 09:26, Tim Leydecker a écrit : Personally, I´m hesistant to using two or more cards with SLI because of micro stuttering: http://en.wikipedia.org/wiki/Micro_stuttering If there would be a solution to that, I´d go with two GTX670 w/4GB VRAM, as they are the same GK104´s with a 915MHz chipspeed instead of a 1006Mhz chipspeed as in the reference design GTX680. That could save another 15-35% percent of investment compared to two single chip GTX 680 cards or one GTX Titan. Overclocked versions may use slightly different chip/shader speeds. In any case, as much VRAM as available, as that always helps in many progamms like Mudbox, Redshift and isn´t much of an added cost (comparing 2GB vs 4GB). At a company I worked Mari 1.5.x behaved bitchy unless it was given a Quadro or forced to ignore the actual card´s game heritage. But that may have been solved with 2.0... Cheers, tim On 27.03.2013 08:59, Mirko Jankovic wrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.com**wrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was
Re: Announcing Redshift - Biased GPU Renderer
SLI and crossfire dio not affect viewport performance in any of 3d application. On Wed, Mar 27, 2013 at 11:12 AM, olivier jeannel olivier.jean...@noos.frwrote: There was a subject on Redshift Forum about having two grapphic cards. It seems to be possible to keep a quadro for dispaly (as it is significantly better at displaying), and have a Titan dedicated to rendering only (in Redshift you select which card is rendering) as they have a huge amount of cores and faster memory. I think I've red somewhere that Titan has 2600 cores against 256 for the Quadro 4000. After chating with Nicolas the Titan could be around 4 time faster than the Quadro4000 ...Which is huge :) Le 27/03/2013 09:26, Tim Leydecker a écrit : Personally, I´m hesistant to using two or more cards with SLI because of micro stuttering: http://en.wikipedia.org/wiki/** Micro_stuttering http://en.wikipedia.org/wiki/Micro_stuttering If there would be a solution to that, I´d go with two GTX670 w/4GB VRAM, as they are the same GK104´s with a 915MHz chipspeed instead of a 1006Mhz chipspeed as in the reference design GTX680. That could save another 15-35% percent of investment compared to two single chip GTX 680 cards or one GTX Titan. Overclocked versions may use slightly different chip/shader speeds. In any case, as much VRAM as available, as that always helps in many progamms like Mudbox, Redshift and isn´t much of an added cost (comparing 2GB vs 4GB). At a company I worked Mari 1.5.x behaved bitchy unless it was given a Quadro or forced to ignore the actual card´s game heritage. But that may have been solved with 2.0... Cheers, tim On 27.03.2013 08:59, Mirko Jankovic wrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is
Re: Announcing Redshift - Biased GPU Renderer
I installed my Titan yesterday, and it bloody screams. Images soon as I get through this project deadline. On Wed, Mar 27, 2013 at 6:47 AM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: SLI and crossfire dio not affect viewport performance in any of 3d application. On Wed, Mar 27, 2013 at 11:12 AM, olivier jeannel olivier.jean...@noos.fr wrote: There was a subject on Redshift Forum about having two grapphic cards. It seems to be possible to keep a quadro for dispaly (as it is significantly better at displaying), and have a Titan dedicated to rendering only (in Redshift you select which card is rendering) as they have a huge amount of cores and faster memory. I think I've red somewhere that Titan has 2600 cores against 256 for the Quadro 4000. After chating with Nicolas the Titan could be around 4 time faster than the Quadro4000 ...Which is huge :) Le 27/03/2013 09:26, Tim Leydecker a écrit : Personally, I´m hesistant to using two or more cards with SLI because of micro stuttering: http://en.wikipedia.org/wiki/** Micro_stuttering http://en.wikipedia.org/wiki/Micro_stuttering If there would be a solution to that, I´d go with two GTX670 w/4GB VRAM, as they are the same GK104´s with a 915MHz chipspeed instead of a 1006Mhz chipspeed as in the reference design GTX680. That could save another 15-35% percent of investment compared to two single chip GTX 680 cards or one GTX Titan. Overclocked versions may use slightly different chip/shader speeds. In any case, as much VRAM as available, as that always helps in many progamms like Mudbox, Redshift and isn´t much of an added cost (comparing 2GB vs 4GB). At a company I worked Mari 1.5.x behaved bitchy unless it was given a Quadro or forced to ignore the actual card´s game heritage. But that may have been solved with 2.0... Cheers, tim On 27.03.2013 08:59, Mirko Jankovic wrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of
Re: Announcing Redshift - Biased GPU Renderer
Congrats! Wish I could get one of those puppies. On Wed, Mar 27, 2013 at 2:20 PM, Ed Manning etmth...@gmail.com wrote: I installed my Titan yesterday, and it bloody screams. Images soon as I get through this project deadline. On Wed, Mar 27, 2013 at 6:47 AM, Mirko Jankovic mirkoj.anima...@gmail.com wrote: SLI and crossfire dio not affect viewport performance in any of 3d application. On Wed, Mar 27, 2013 at 11:12 AM, olivier jeannel olivier.jean...@noos.fr wrote: There was a subject on Redshift Forum about having two grapphic cards. It seems to be possible to keep a quadro for dispaly (as it is significantly better at displaying), and have a Titan dedicated to rendering only (in Redshift you select which card is rendering) as they have a huge amount of cores and faster memory. I think I've red somewhere that Titan has 2600 cores against 256 for the Quadro 4000. After chating with Nicolas the Titan could be around 4 time faster than the Quadro4000 ...Which is huge :) Le 27/03/2013 09:26, Tim Leydecker a écrit : Personally, I´m hesistant to using two or more cards with SLI because of micro stuttering: http://en.wikipedia.org/wiki/** Micro_stuttering http://en.wikipedia.org/wiki/Micro_stuttering If there would be a solution to that, I´d go with two GTX670 w/4GB VRAM, as they are the same GK104´s with a 915MHz chipspeed instead of a 1006Mhz chipspeed as in the reference design GTX680. That could save another 15-35% percent of investment compared to two single chip GTX 680 cards or one GTX Titan. Overclocked versions may use slightly different chip/shader speeds. In any case, as much VRAM as available, as that always helps in many progamms like Mudbox, Redshift and isn´t much of an added cost (comparing 2GB vs 4GB). At a company I worked Mari 1.5.x behaved bitchy unless it was given a Quadro or forced to ignore the actual card´s game heritage. But that may have been solved with 2.0... Cheers, tim On 27.03.2013 08:59, Mirko Jankovic wrote: On the other hand Titan is more expensive than 2 gtx680 if I'm not mistaken... and i bet that with two 680 in SLI, when multi GPU is supported you will have better performance than with 1 titan right? On Wed, Mar 27, 2013 at 8:55 AM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now
Re: Announcing Redshift - Biased GPU Renderer
i bet, but damn... its an expensive card. On Wed, Mar 27, 2013 at 11:20 AM, Ed Manning etmth...@gmail.com wrote: I installed my Titan yesterday, and it bloody screams. Images soon as I get through this project deadline.
Re: Announcing Redshift - Biased GPU Renderer
yes, but reasonably less than Tesla and Maximus setups that it outperforms and uses less power than. And much less than a new computer. Makes my quad-xeon 2008 Mac Pro a viable workstation/renderbox for non-CPU tasks. On Wed, Mar 27, 2013 at 2:26 PM, Steven Caron car...@gmail.com wrote: i bet, but damn... its an expensive card. On Wed, Mar 27, 2013 at 11:20 AM, Ed Manning etmth...@gmail.com wrote: I installed my Titan yesterday, and it bloody screams. Images soon as I get through this project deadline.
Re: Announcing Redshift - Biased GPU Renderer
In what spare time I have I'm setting up a shootout between Octane standalone and redshift in SI.
Re: Announcing Redshift - Biased GPU Renderer
Ed, did Octane ever release their SI plugin? On Wed, Mar 27, 2013 at 3:23 PM, Ed Manning etmth...@gmail.com wrote: In what spare time I have I'm setting up a shootout between Octane standalone and redshift in SI.
Re: Announcing Redshift - Biased GPU Renderer
And Vimeo lets you allow people to download the video too, if you enable that option. On Tue, Mar 26, 2013 at 11:08 PM, Ahmidou.xsi ahmidou@gmail.com wrote: Hi Nicolas, you could maube consider vimeo as a better option. Cheers Le 27 mars 2013 à 13:43, Nicolas Burtnyk nico...@redshift3d.com a écrit : Hey guys, Just wanted to share a couple of very short videos we made that show the stability of the GI in Redshift. Unfortunately Youtube's compression kind of murdered the smoothness, but I assure you that any artifacts you see in these videos are from compression and not GI. http://youtu.be/3c0tHYdd-fg This video shows the dark side of a deforming gargoyle lit by physical sun sky. 25 seconds per frame for 1280x720 on a Core i7 3.07Ghz, 12GB RAM w/ NVIDIA Geforce GTX 470. http://youtu.be/ySCockShxTQ This video shows the same gargoyle being lit *strictly *by light bouncing off the floor. The setup is a white spot light shining onto the floor (off camera). The red glow you see on the floor around the gargoyle is light that has bounced off the floor, then off the gargoyle. 1 minute per frame for 1280x720 on a Core i7 3.07Ghz, 6GB RAM w/ NVIDIA Geforce GTX 670. We're still head-down fixing bugs and bringing new features online, but I plan to spend some time making more (and better) videos soon. -Nicolas On Tue, Mar 26, 2013 at 7:27 PM, Nicolas Burtnyk nico...@redshift3d.comwrote:
Re: Announcing Redshift - Biased GPU Renderer
Yeah we'll switch to Vimeo once we do our next batch of videos. Looks like we'll need a Pro account, which isn't free but the cost is pretty reasonable. -Nicolas On Wed, Mar 27, 2013 at 12:34 PM, Alan Fregtman alan.fregt...@gmail.comwrote: And Vimeo lets you allow people to download the video too, if you enable that option. On Tue, Mar 26, 2013 at 11:08 PM, Ahmidou.xsi ahmidou@gmail.comwrote: Hi Nicolas, you could maube consider vimeo as a better option. Cheers Le 27 mars 2013 à 13:43, Nicolas Burtnyk nico...@redshift3d.com a écrit : Hey guys, Just wanted to share a couple of very short videos we made that show the stability of the GI in Redshift. Unfortunately Youtube's compression kind of murdered the smoothness, but I assure you that any artifacts you see in these videos are from compression and not GI. http://youtu.be/3c0tHYdd-fg This video shows the dark side of a deforming gargoyle lit by physical sun sky. 25 seconds per frame for 1280x720 on a Core i7 3.07Ghz, 12GB RAM w/ NVIDIA Geforce GTX 470. http://youtu.be/ySCockShxTQ This video shows the same gargoyle being lit *strictly *by light bouncing off the floor. The setup is a white spot light shining onto the floor (off camera). The red glow you see on the floor around the gargoyle is light that has bounced off the floor, then off the gargoyle. 1 minute per frame for 1280x720 on a Core i7 3.07Ghz, 6GB RAM w/ NVIDIA Geforce GTX 670. We're still head-down fixing bugs and bringing new features online, but I plan to spend some time making more (and better) videos soon. -Nicolas On Tue, Mar 26, 2013 at 7:27 PM, Nicolas Burtnyk nico...@redshift3d.comwrote:
Re: Announcing Redshift - Biased GPU Renderer
We use a pro account for all of our stuff and it's a lot nicer than youtube - faster to upload, easy to upload multiple videos. There's also a fair few 3D-focused groups and channels on there, so I find the exposure to potential customers is much better. On 27 March 2013 15:37, Nicolas Burtnyk nico...@redshift3d.com wrote: Yeah we'll switch to Vimeo once we do our next batch of videos. Looks like we'll need a Pro account, which isn't free but the cost is pretty reasonable. -Nicolas On Wed, Mar 27, 2013 at 12:34 PM, Alan Fregtman alan.fregt...@gmail.comwrote: And Vimeo lets you allow people to download the video too, if you enable that option. On Tue, Mar 26, 2013 at 11:08 PM, Ahmidou.xsi ahmidou@gmail.comwrote: Hi Nicolas, you could maube consider vimeo as a better option. Cheers Le 27 mars 2013 à 13:43, Nicolas Burtnyk nico...@redshift3d.com a écrit : Hey guys, Just wanted to share a couple of very short videos we made that show the stability of the GI in Redshift. Unfortunately Youtube's compression kind of murdered the smoothness, but I assure you that any artifacts you see in these videos are from compression and not GI. http://youtu.be/3c0tHYdd-fg This video shows the dark side of a deforming gargoyle lit by physical sun sky. 25 seconds per frame for 1280x720 on a Core i7 3.07Ghz, 12GB RAM w/ NVIDIA Geforce GTX 470. http://youtu.be/ySCockShxTQ This video shows the same gargoyle being lit *strictly *by light bouncing off the floor. The setup is a white spot light shining onto the floor (off camera). The red glow you see on the floor around the gargoyle is light that has bounced off the floor, then off the gargoyle. 1 minute per frame for 1280x720 on a Core i7 3.07Ghz, 6GB RAM w/ NVIDIA Geforce GTX 670. We're still head-down fixing bugs and bringing new features online, but I plan to spend some time making more (and better) videos soon. -Nicolas On Tue, Mar 26, 2013 at 7:27 PM, Nicolas Burtnyk nico...@redshift3d.com wrote:
Re: Announcing Redshift - Biased GPU Renderer
Thanks Paul. You guys are the kings of cool videos, so your advice is well received! On Wed, Mar 27, 2013 at 12:49 PM, Paul Doyle technove...@gmail.com wrote: We use a pro account for all of our stuff and it's a lot nicer than youtube - faster to upload, easy to upload multiple videos. There's also a fair few 3D-focused groups and channels on there, so I find the exposure to potential customers is much better. On 27 March 2013 15:37, Nicolas Burtnyk nico...@redshift3d.com wrote: Yeah we'll switch to Vimeo once we do our next batch of videos. Looks like we'll need a Pro account, which isn't free but the cost is pretty reasonable. -Nicolas On Wed, Mar 27, 2013 at 12:34 PM, Alan Fregtman alan.fregt...@gmail.comwrote: And Vimeo lets you allow people to download the video too, if you enable that option. On Tue, Mar 26, 2013 at 11:08 PM, Ahmidou.xsi ahmidou@gmail.comwrote: Hi Nicolas, you could maube consider vimeo as a better option. Cheers Le 27 mars 2013 à 13:43, Nicolas Burtnyk nico...@redshift3d.com a écrit : Hey guys, Just wanted to share a couple of very short videos we made that show the stability of the GI in Redshift. Unfortunately Youtube's compression kind of murdered the smoothness, but I assure you that any artifacts you see in these videos are from compression and not GI. http://youtu.be/3c0tHYdd-fg This video shows the dark side of a deforming gargoyle lit by physical sun sky. 25 seconds per frame for 1280x720 on a Core i7 3.07Ghz, 12GB RAM w/ NVIDIA Geforce GTX 470. http://youtu.be/ySCockShxTQ This video shows the same gargoyle being lit *strictly *by light bouncing off the floor. The setup is a white spot light shining onto the floor (off camera). The red glow you see on the floor around the gargoyle is light that has bounced off the floor, then off the gargoyle. 1 minute per frame for 1280x720 on a Core i7 3.07Ghz, 6GB RAM w/ NVIDIA Geforce GTX 670. We're still head-down fixing bugs and bringing new features online, but I plan to spend some time making more (and better) videos soon. -Nicolas On Tue, Mar 26, 2013 at 7:27 PM, Nicolas Burtnyk nico...@redshift3d.com wrote:
Re: Announcing Redshift - Biased GPU Renderer
Don't call it a gimmick then (although it is with all the fashion and hype elements around it), call it a singularity, but if you're looking at benching and sorting videocards for performance and bang for buck you should exclude it. Unless you also want to include that massive liquid cooled asus radeon that is sold in a military grade carrying case and other things like that :) I've tried it btw as a friend's shop had a review return they kindly lent me for a week (they work closely with GB since one of the partners is an ex employee and another moonlights reviewing hardware). It was hardly a noticeable improvement over the GB OC 680 4GB I had (and still have) in there. The practical performance gains are far, far inferior to 35%. Only the added ram is nice, but nothing justifies a price tag that is more than doubled compared to the 680. It's a gimmick because you need a serious hardware fetish to justify forking out 1250-1400$ out for it compared to a benched OC 680 with 4GB that you can have for 550$ and have chances to trivially overclock and narrow the gap again. I run a dell 2711 and an additional 1980x1200 monitor with it btw. On Wed, Mar 27, 2013 at 6:55 PM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.com**wrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA
Re: Announcing Redshift - Biased GPU Renderer
The TITAN is not a gimmick with respect to Redshift. It's almost twice as fast as a GTX 670 on all the tests we've run. We don't have a GTX 680 so I don't have the numbers to compare against. Pricing wise, there TITAN costs $1K and the 680 4GB is $550 so the 680 wins for price/performance ratio (but probably not by a whole lot). For performance/watt, the TITAN wins by a lot. On Wed, Mar 27, 2013 at 3:09 PM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: Don't call it a gimmick then (although it is with all the fashion and hype elements around it), call it a singularity, but if you're looking at benching and sorting videocards for performance and bang for buck you should exclude it. Unless you also want to include that massive liquid cooled asus radeon that is sold in a military grade carrying case and other things like that :) I've tried it btw as a friend's shop had a review return they kindly lent me for a week (they work closely with GB since one of the partners is an ex employee and another moonlights reviewing hardware). It was hardly a noticeable improvement over the GB OC 680 4GB I had (and still have) in there. The practical performance gains are far, far inferior to 35%. Only the added ram is nice, but nothing justifies a price tag that is more than doubled compared to the 680. It's a gimmick because you need a serious hardware fetish to justify forking out 1250-1400$ out for it compared to a benched OC 680 with 4GB that you can have for 550$ and have chances to trivially overclock and narrow the gap again. I run a dell 2711 and an additional 1980x1200 monitor with it btw. On Wed, Mar 27, 2013 at 6:55 PM, Tim Leydecker bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.com**wrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was that
RE: Announcing Redshift - Biased GPU Renderer
I have been following this thread and have been wondering if the fact that the K6X0 and Quadro K5000 are more tuned for single precision is making the difference between them and Titan which from what I understand is more tuned for double precision? Or does that even matter for this or other renderers? I admit my knowledge in this area is pretty scarce. We just got some K5000's and were trying to get a handle on all of this before we purchased them. We never sorted it out so we went ahead with the K5000 which seem to be fine so far but I admit they have not yet been pushed for compute or rendering. jeff From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Nicolas Burtnyk Sent: Wednesday, March 27, 2013 6:20 PM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer The TITAN is not a gimmick with respect to Redshift. It's almost twice as fast as a GTX 670 on all the tests we've run. We don't have a GTX 680 so I don't have the numbers to compare against. Pricing wise, there TITAN costs $1K and the 680 4GB is $550 so the 680 wins for price/performance ratio (but probably not by a whole lot). For performance/watt, the TITAN wins by a lot. On Wed, Mar 27, 2013 at 3:09 PM, Raffaele Fragapane raffsxsil...@googlemail.commailto:raffsxsil...@googlemail.com wrote: Don't call it a gimmick then (although it is with all the fashion and hype elements around it), call it a singularity, but if you're looking at benching and sorting videocards for performance and bang for buck you should exclude it. Unless you also want to include that massive liquid cooled asus radeon that is sold in a military grade carrying case and other things like that :) I've tried it btw as a friend's shop had a review return they kindly lent me for a week (they work closely with GB since one of the partners is an ex employee and another moonlights reviewing hardware). It was hardly a noticeable improvement over the GB OC 680 4GB I had (and still have) in there. The practical performance gains are far, far inferior to 35%. Only the added ram is nice, but nothing justifies a price tag that is more than doubled compared to the 680. It's a gimmick because you need a serious hardware fetish to justify forking out 1250-1400$ out for it compared to a benched OC 680 with 4GB that you can have for 550$ and have chances to trivially overclock and narrow the gap again. I run a dell 2711 and an additional 1980x1200 monitor with it btw. On Wed, Mar 27, 2013 at 6:55 PM, Tim Leydecker bauero...@gmx.demailto:bauero...@gmx.de wrote: The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05tel:27.03.2013%2005:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.commailto:mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp
Re: Announcing Redshift - Biased GPU Renderer
While I'm not a huge fan of Anand, they do occasionally have a good article out. http://www.anandtech.com/show/6774/nvidias-geforce-gtx-titan-part-2-titans-performance-unveiled/3 This sheds some light on what you might be asking about, and on why some times you hear that the 580s are doing better than the 680s and why the latter is considered a crippled card for professional use. It does omit the fact that the (factory) OCed premium 680s, especially with the memory clocked higher, actually go up a fair chunk, and that if you have a 680 that hits 1400 then some of those tests, especially short span ones where Titan's turbo doesn't have the time to kick in, will actually see the 680 taking the lead over the titan in both numbers and power usage. Only benching I've done was CUDA and number crunching related because I've taken an interest in it a while ago and still toy with it on and off, and that includes the generic GEMM and FFT tests. I don't bother with game benchmarks or 3DMark or cinebench, but single precision the 680 stock cooled but OCed was constantly bang-on on par with the titan for a lower power draw. Double precision even OCed it (680) will fall back a fair chunk, and water cooled OCed 580s actually take the lead in bang for buck by a mile, but have horrible (high) power draw. You can consider the k5000 somewhat closer to the titan than to the 680. On Thu, Mar 28, 2013 at 10:16 AM, Jeff McFall jeff.mcf...@sas.com wrote: I have been following this thread and have been wondering if the fact that the K6X0 and Quadro K5000 are more tuned for single precision is making the difference between them and Titan which from what I understand is more tuned for double precision? Or does that even matter for this or other renderers? I admit my knowledge in this area is pretty scarce. ** ** We just got some K5000’s and were trying to get a handle on all of this before we purchased them. We never sorted it out so we went ahead with the K5000 which seem to be fine so far but I admit they have not yet been pushed for compute or rendering. ** ** ** ** jeff ** ** ** ** ** ** ** ** *From:* softimage-boun...@listproc.autodesk.com [mailto: softimage-boun...@listproc.autodesk.com] *On Behalf Of *Nicolas Burtnyk *Sent:* Wednesday, March 27, 2013 6:20 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer ** ** The TITAN is not a gimmick with respect to Redshift. It's almost twice as fast as a GTX 670 on all the tests we've run. We don't have a GTX 680 so I don't have the numbers to compare against. Pricing wise, there TITAN costs $1K and the 680 4GB is $550 so the 680 wins for price/performance ratio (but probably not by a whole lot). For performance/watt, the TITAN wins by a lot. ** ** On Wed, Mar 27, 2013 at 3:09 PM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: Don't call it a gimmick then (although it is with all the fashion and hype elements around it), call it a singularity, but if you're looking at benching and sorting videocards for performance and bang for buck you should exclude it. Unless you also want to include that massive liquid cooled asus radeon that is sold in a military grade carrying case and other things like that :) I've tried it btw as a friend's shop had a review return they kindly lent me for a week (they work closely with GB since one of the partners is an ex employee and another moonlights reviewing hardware). It was hardly a noticeable improvement over the GB OC 680 4GB I had (and still have) in there. The practical performance gains are far, far inferior to 35%. Only the added ram is nice, but nothing justifies a price tag that is more than doubled compared to the 680. It's a gimmick because you need a serious hardware fetish to justify forking out 1250-1400$ out for it compared to a benched OC 680 with 4GB that you can have for 550$ and have chances to trivially overclock and narrow the gap again. I run a dell 2711 and an additional 1980x1200 monitor with it btw. ** ** On Wed, Mar 27, 2013 at 6:55 PM, Tim Leydecker bauero...@gmx.de wrote:** ** The GTX Titan is not a gimmick but uses the successor to the chip series used in the GTX 680, e.g. the GT(X) 6xx series uses the GK104, while the GTX Titan uses the GK110. You can find the GK110 in the Tesla K20, too. You could describe the GTX690 as a gimmick, as it uses two GK104 on one card to maximize performance at the cost of higher powerconsumption, noise and heat. The performance gain between a GTX680 and a GTX Titan is roughly 35% and can be felt nicely when using it with higher screenresolutions like 1920x1200 or 2560x1440 and higher antialiasing in games. That´s where the 6GB VRAM of the GTX Titan come in handy, too. Cheers, tim On 27.03.2013 05:24, Raffaele Fragapane wrote: Benchmarking is more driver tuning than it's videocard
Re: Announcing Redshift - Biased GPU Renderer
The answer is... it depends :) If your scene is very large and doesn't fit in X GB of VRAM, then more VRAM will be a big performance win because you'll be going out of core less. That being said, even for simpler scenes that easily fit in VRAM, more VRAM can improve performance. Redshift can use excess VRAM to increase the size of its workloads which results in better utilzation of the GPU. So more memory is good and more/faster cores is good too, but it's impossible to give you a X factor for the performance difference between a 1GB and 2GB VRAM card with equivalent cores. It's just too scene dependent. -Nicolas On Mon, Mar 18, 2013 at 4:49 AM, Stefan Kubicek s...@tidbit-images.comwrote: Hi Nicolas, I'm curious in how far GPU memory impacts render time. To put it differently: Assuming the amount of cores is what makes the biggest difference in render time, what's the expected speed differences comparing a graphics card with 1gb to one equipped with 2gb or more? Cheers, Stefan Hi Mirko, Redshift does use the CPU for a couple things here and there so the CPU is not irrelevant to the performance, but it's not a big contributor. For example, the RT hierarchy construction (construction of the acceleration structure for raytracing) is done on the CPU as is texture conversion to our optimized tile format. Also the screen-space adaptive tessellation and some rebalancing of tree data structures for the point-based techniques run on the CPU. All in all though the GPU spec is what is really going to make the difference for Redshift, assuming we're not talking about a 486 or something :) So yeah I encourage you to dust off that old PC, pop in the GTX580 and take it for a spin. I think you'll be pleasantly surprised. -Nicolas On Fri, Mar 15, 2013 at 1:21 PM, Mirko Jankovic mirkoj.anima...@gmail.com**wrote: just wondering.. if everything is on GPU in theory it should work fine even with some slower older CPU? Got some older comp laying around and both PCI slots in comp are filled, maybe could use that one for GPU rendering station for testing :) any thoughts? On Fri, Mar 15, 2013 at 9:00 PM, Len Krenzler l...@creativecontrol.ca wrote: No kidding! I can't imagine why Nvidia/Arc/MR whatever they're called now couldn't have done this years ago. Between them and AD they can't even get they're basic features working. Money should be directed to these 3rd party guys as much as possible. AD must have called me about 10 times to renew my sub. This time I said no, I think I'll spend that on the guys getting results. If you haven't tried this yet, do! On 3/15/2013 1:54 PM, Mirko Jankovic wrote: Just thinking.. you really have to wonder why AD as huge company with resources that are probably hard to imagine by any of us never got to make ANYTHING nearly great as anything like what we see from 3rd party guys around. If you think about it like 99% of progress is NOT made by big companies. Really need to think in which direction money flow could start to change.. away from AD-like and more towards guys like Redshift team :) All the best! On Fri, Mar 15, 2013 at 8:47 PM, Christian Gotzinger cgo...@googlemail.com wrote: Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas -- __**___ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- --**- Stefan Kubicek --**- keyvis digital imagery Alfred Feierfeilstraße 3 A-2380 Perchtoldsdorf bei Wien Phone:+43/699/12614231 www.keyvis.at ste...@keyvis.at -- This email and its attachments are -- --confidential and for the recipient only--
Re: Announcing Redshift - Biased GPU Renderer
Hey guys, Just wanted to share a couple of very short videos we made that show the stability of the GI in Redshift. Unfortunately Youtube's compression kind of murdered the smoothness, but I assure you that any artifacts you see in these videos are from compression and not GI. http://youtu.be/3c0tHYdd-fg This video shows the dark side of a deforming gargoyle lit by physical sun sky. 25 seconds per frame for 1280x720 on a Core i7 3.07Ghz, 12GB RAM w/ NVIDIA Geforce GTX 470. http://youtu.be/ySCockShxTQ This video shows the same gargoyle being lit *strictly *by light bouncing off the floor. The setup is a white spot light shining onto the floor (off camera). The red glow you see on the floor around the gargoyle is light that has bounced off the floor, then off the gargoyle. 1 minute per frame for 1280x720 on a Core i7 3.07Ghz, 6GB RAM w/ NVIDIA Geforce GTX 670. We're still head-down fixing bugs and bringing new features online, but I plan to spend some time making more (and better) videos soon. -Nicolas On Tue, Mar 26, 2013 at 7:27 PM, Nicolas Burtnyk nico...@redshift3d.comwrote:
Re: Announcing Redshift - Biased GPU Renderer
Hi Nicolas, you could maube consider vimeo as a better option. Cheers Le 27 mars 2013 à 13:43, Nicolas Burtnyk nico...@redshift3d.com a écrit : Hey guys, Just wanted to share a couple of very short videos we made that show the stability of the GI in Redshift. Unfortunately Youtube's compression kind of murdered the smoothness, but I assure you that any artifacts you see in these videos are from compression and not GI. http://youtu.be/3c0tHYdd-fg This video shows the dark side of a deforming gargoyle lit by physical sun sky. 25 seconds per frame for 1280x720 on a Core i7 3.07Ghz, 12GB RAM w/ NVIDIA Geforce GTX 470. http://youtu.be/ySCockShxTQ This video shows the same gargoyle being lit strictly by light bouncing off the floor. The setup is a white spot light shining onto the floor (off camera). The red glow you see on the floor around the gargoyle is light that has bounced off the floor, then off the gargoyle. 1 minute per frame for 1280x720 on a Core i7 3.07Ghz, 6GB RAM w/ NVIDIA Geforce GTX 670. We're still head-down fixing bugs and bringing new features online, but I plan to spend some time making more (and better) videos soon. -Nicolas On Tue, Mar 26, 2013 at 7:27 PM, Nicolas Burtnyk nico...@redshift3d.com wrote:
Re: Announcing Redshift - Biased GPU Renderer
Benchmarking is more driver tuning than it's videocard performance, and if you want to look at number crunching you should look at the most recent gens. The 680 has brought nVIDIA back up top for number crunching (forgetting the silver editions or gimmicks like the titan), and close enough to bang for buck best, but AMD's response to that still has to come. Ironically, though, the 6xx gen is reported as a crippled, bad performer in DCC apps, although I can't say I noticed it myself. It sure as hell works admirably well in mudbox, mari, cuda work, and I've had no issues in maya or soft. I don't really benchmrak or obsess over numbers much though. When this will obsolesce, I will considering AMD again, probably in a couple years. For GPU rendering though, well, that's something you CAN bench reliably with the engine, and AMD might still win the FLOP per dollar run there, so it's not to be discounted. Would be good to know what the redshift guys have to say about it themselves though if they can spare the thought and can actually disclose. On Thu, Mar 21, 2013 at 9:04 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.comwrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA for a while in number crunching, the main problem is the drivers are still a coin toss chance, and that OCL isn't anywhere as popular as CUDA. With win7 or 8 and recent versions of Soft/Maya they can do well. nVIDIA didn't help with the crippling of the 6xx for professional use, and pissing off Linus. They are still ahead by a slight margin, for now, but I wouldn't discount AMD wholesale anymore. If the next generation is as disappointing as Kepler is, and AMD gets both Linux support AND decent (and properly OSS) drivers out, I'm moving time come for the next upgrade. For now I recently bought a 680 because it was kind of mandatory to not go insane with Mari and Mudbox, and because I like CUDA and I toy with it at home. On Wed, Mar 20, 2013 at 9:58 PM, Dan Yargici danyarg...@gmail.comwrote: Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... Really? I don't remember anyone ever suggesting ATI was anything other than shit! DAN -- Our users will know fear and cower before our software! Ship it! Ship it and let them flee like the dogs they are!
Re: Announcing Redshift - Biased GPU Renderer
Yep, doesn't flicker one bit! On 26/03/2013 10:43 PM, Nicolas Burtnyk wrote: Hey guys, Just wanted to share a couple of very short videos we made that show the stability of the GI in Redshift. Unfortunately Youtube's compression kind of murdered the smoothness, but I assure you that any artifacts you see in these videos are from compression and not GI. http://youtu.be/3c0tHYdd-fg This video shows the dark side of a deforming gargoyle lit by physical sun sky. 25 seconds per frame for 1280x720 on a Core i7 3.07Ghz, 12GB RAM w/ NVIDIA Geforce GTX 470. http://youtu.be/ySCockShxTQ This video shows the same gargoyle being lit _strictly _by light bouncing off the floor. The setup is a white spot light shining onto the floor (off camera). The red glow you see on the floor around the gargoyle is light that has bounced off the floor, then off the gargoyle. 1 minute per frame for 1280x720 on a Core i7 3.07Ghz, 6GB RAM w/ NVIDIA Geforce GTX 670. We're still head-down fixing bugs and bringing new features online, but I plan to spend some time making more (and better) videos soon. -Nicolas On Tue, Mar 26, 2013 at 7:27 PM, Nicolas Burtnyk nico...@redshift3d.com mailto:nico...@redshift3d.com wrote:
Re: Announcing Redshift - Biased GPU Renderer
My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA for a while in number crunching, the main problem is the drivers are still a coin toss chance, and that OCL isn't anywhere as popular as CUDA. With win7 or 8 and recent versions of Soft/Maya they can do well. nVIDIA didn't help with the crippling of the 6xx for professional use, and pissing off Linus. They are still ahead by a slight margin, for now, but I wouldn't discount AMD wholesale anymore. If the next generation is as disappointing as Kepler is, and AMD gets both Linux support AND decent (and properly OSS) drivers out, I'm moving time come for the next upgrade. For now I recently bought a 680 because it was kind of mandatory to not go insane with Mari and Mudbox, and because I like CUDA and I toy with it at home. On Wed, Mar 20, 2013 at 9:58 PM, Dan Yargici danyarg...@gmail.com wrote: Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... Really? I don't remember anyone ever suggesting ATI was anything other than shit! DAN
Re: Announcing Redshift - Biased GPU Renderer
well no idea about pro cards.. really never got financial justification to get one, quadro 4000 in old company didn;t really felt anything much better than gaming cards so... but in gaming segment.. opengl scores in sinebench for example: gtx 580: ~55 7970: ~90 to start with not to mention annoying issue with high segment rotating cube in viewport in SI. 7970 smooth at ~170 fps with gtx580 bfore that.. to point out that the rest of comp is identical only switched card... for the first 30-50sec frame rate was stuck at something like 17 fps... and after that it kinda jump to ~70-80fps... in any case with gaming cards ati vs nvidia there is no doubt. and if you are not using CUDA much then no need to even thing which way to go. Now redshift is game changer heheh but I'm still hoping that OpenCL will be supported and I'm looking forward to test it out with two of 7970 in crossfire :) btw I'm not much into programming waters but is it really OpenCL programming that as I understood should work on ALL cards, is that much more complex than for CUDA which is limited to nvidia only? Wouldn't it be more logical to go with solution that is covering a lot more market than something limited to one manufacturer? On Thu, Mar 21, 2013 at 10:55 AM, Arvid Björn arvidbj...@gmail.com wrote: My beef with ATI last time I tried FirePro was that it had a hard time locking into 25fps playback in some apps, as if the refresh rate was locked to 30/60. Realtime playback in Softimage would stutter annoyingly IIRC. Plus it seemed to draw text slightly differently in some apps. Nvidia just feels.. comfy. On Thu, Mar 21, 2013 at 5:21 AM, Raffaele Fragapane raffsxsil...@googlemail.com wrote: These days if you hit the right combination of drivers and planet alignment they are OK. Performance wise they have been ahead of nVIDIA for a while in number crunching, the main problem is the drivers are still a coin toss chance, and that OCL isn't anywhere as popular as CUDA. With win7 or 8 and recent versions of Soft/Maya they can do well. nVIDIA didn't help with the crippling of the 6xx for professional use, and pissing off Linus. They are still ahead by a slight margin, for now, but I wouldn't discount AMD wholesale anymore. If the next generation is as disappointing as Kepler is, and AMD gets both Linux support AND decent (and properly OSS) drivers out, I'm moving time come for the next upgrade. For now I recently bought a 680 because it was kind of mandatory to not go insane with Mari and Mudbox, and because I like CUDA and I toy with it at home. On Wed, Mar 20, 2013 at 9:58 PM, Dan Yargici danyarg...@gmail.comwrote: Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... Really? I don't remember anyone ever suggesting ATI was anything other than shit! DAN
Re: Announcing Redshift - Biased GPU Renderer
testing it a bit and looks great! amazing work guys, grats. any ETA for production ready version? also reall shame again that it is nvidia only for now. Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... having this support openCL would be great! But everything in it's time. Grats! On Mon, Mar 18, 2013 at 9:15 PM, Stefan Andersson sander...@gmail.comwrote: That was pretty neat! :) I can't wait to see some more test! regards stefan On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.comwrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- *Stefan Andersson | Digital Janitor* blog http://sanders3d.wordpress.com | showreelhttp://vimeo.com/sanders3d| twitter http://twitter.com/sanders3d | LinkedInhttp://www.linkedin.com/in/sanders3d| cell: +46-73-6268850 | skype:sanders3d
Re: Announcing Redshift - Biased GPU Renderer
Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... Really? I don't remember anyone ever suggesting ATI was anything other than shit! DAN On Wed, Mar 20, 2013 at 12:32 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: testing it a bit and looks great! amazing work guys, grats. any ETA for production ready version? also reall shame again that it is nvidia only for now. Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... having this support openCL would be great! But everything in it's time. Grats! On Mon, Mar 18, 2013 at 9:15 PM, Stefan Andersson sander...@gmail.comwrote: That was pretty neat! :) I can't wait to see some more test! regards stefan On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.comwrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- *Stefan Andersson | Digital Janitor* blog http://sanders3d.wordpress.com | showreelhttp://vimeo.com/sanders3d| twitter http://twitter.com/sanders3d | LinkedInhttp://www.linkedin.com/in/sanders3d| cell: +46-73-6268850 | skype:sanders3d
Re: Announcing Redshift - Biased GPU Renderer
http://www.si-community.com/community/viewtopic.php?f=36t=3526 latest one. ati rigth now leaves nvidia in the dust On Wed, Mar 20, 2013 at 11:58 AM, Dan Yargici danyarg...@gmail.com wrote: Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... Really? I don't remember anyone ever suggesting ATI was anything other than shit! DAN On Wed, Mar 20, 2013 at 12:32 PM, Mirko Jankovic mirkoj.anima...@gmail.com wrote: testing it a bit and looks great! amazing work guys, grats. any ETA for production ready version? also reall shame again that it is nvidia only for now. Ati was tested over and over and showing a lot better viewport results in Softimage than nvidia... having this support openCL would be great! But everything in it's time. Grats! On Mon, Mar 18, 2013 at 9:15 PM, Stefan Andersson sander...@gmail.comwrote: That was pretty neat! :) I can't wait to see some more test! regards stefan On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- *Stefan Andersson | Digital Janitor* blog http://sanders3d.wordpress.com | showreelhttp://vimeo.com/sanders3d| twitter http://twitter.com/sanders3d | LinkedInhttp://www.linkedin.com/in/sanders3d| cell: +46-73-6268850 | skype:sanders3d
Re: Announcing Redshift - Biased GPU Renderer
Hi Nicolas, I'm curious in how far GPU memory impacts render time. To put it differently: Assuming the amount of cores is what makes the biggest difference in render time, what's the expected speed differences comparing a graphics card with 1gb to one equipped with 2gb or more? Cheers, Stefan Hi Mirko, Redshift does use the CPU for a couple things here and there so the CPU is not irrelevant to the performance, but it's not a big contributor. For example, the RT hierarchy construction (construction of the acceleration structure for raytracing) is done on the CPU as is texture conversion to our optimized tile format. Also the screen-space adaptive tessellation and some rebalancing of tree data structures for the point-based techniques run on the CPU. All in all though the GPU spec is what is really going to make the difference for Redshift, assuming we're not talking about a 486 or something :) So yeah I encourage you to dust off that old PC, pop in the GTX580 and take it for a spin. I think you'll be pleasantly surprised. -Nicolas On Fri, Mar 15, 2013 at 1:21 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: just wondering.. if everything is on GPU in theory it should work fine even with some slower older CPU? Got some older comp laying around and both PCI slots in comp are filled, maybe could use that one for GPU rendering station for testing :) any thoughts? On Fri, Mar 15, 2013 at 9:00 PM, Len Krenzler l...@creativecontrol.cawrote: No kidding! I can't imagine why Nvidia/Arc/MR whatever they're called now couldn't have done this years ago. Between them and AD they can't even get they're basic features working. Money should be directed to these 3rd party guys as much as possible. AD must have called me about 10 times to renew my sub. This time I said no, I think I'll spend that on the guys getting results. If you haven't tried this yet, do! On 3/15/2013 1:54 PM, Mirko Jankovic wrote: Just thinking.. you really have to wonder why AD as huge company with resources that are probably hard to imagine by any of us never got to make ANYTHING nearly great as anything like what we see from 3rd party guys around. If you think about it like 99% of progress is NOT made by big companies. Really need to think in which direction money flow could start to change.. away from AD-like and more towards guys like Redshift team :) All the best! On Fri, Mar 15, 2013 at 8:47 PM, Christian Gotzinger cgo...@googlemail.com wrote: Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca -- --- Stefan Kubicek --- keyvis digital imagery Alfred Feierfeilstraße 3 A-2380 Perchtoldsdorf bei Wien Phone:+43/699/12614231 www.keyvis.at ste...@keyvis.at -- This email and its attachments are -- --confidential and for the recipient only--
Re: Announcing Redshift - Biased GPU Renderer
To add to the subject, is there a Redshift benchmark with different graphic cards ? Will be fun to have renderfarm only filled with graphic cards :) -Where's your renderfarm ? -It's the little box on floor... Le 18/03/2013 12:49, Stefan Kubicek a écrit : Hi Nicolas, I'm curious in how far GPU memory impacts render time. To put it differently: Assuming the amount of cores is what makes the biggest difference in render time, what's the expected speed differences comparing a graphics card with 1gb to one equipped with 2gb or more? Cheers, Stefan Hi Mirko, Redshift does use the CPU for a couple things here and there so the CPU is not irrelevant to the performance, but it's not a big contributor. For example, the RT hierarchy construction (construction of the acceleration structure for raytracing) is done on the CPU as is texture conversion to our optimized tile format. Also the screen-space adaptive tessellation and some rebalancing of tree data structures for the point-based techniques run on the CPU. All in all though the GPU spec is what is really going to make the difference for Redshift, assuming we're not talking about a 486 or something :) So yeah I encourage you to dust off that old PC, pop in the GTX580 and take it for a spin. I think you'll be pleasantly surprised. -Nicolas On Fri, Mar 15, 2013 at 1:21 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: just wondering.. if everything is on GPU in theory it should work fine even with some slower older CPU? Got some older comp laying around and both PCI slots in comp are filled, maybe could use that one for GPU rendering station for testing :) any thoughts? On Fri, Mar 15, 2013 at 9:00 PM, Len Krenzler l...@creativecontrol.cawrote: No kidding! I can't imagine why Nvidia/Arc/MR whatever they're called now couldn't have done this years ago. Between them and AD they can't even get they're basic features working. Money should be directed to these 3rd party guys as much as possible. AD must have called me about 10 times to renew my sub. This time I said no, I think I'll spend that on the guys getting results. If you haven't tried this yet, do! On 3/15/2013 1:54 PM, Mirko Jankovic wrote: Just thinking.. you really have to wonder why AD as huge company with resources that are probably hard to imagine by any of us never got to make ANYTHING nearly great as anything like what we see from 3rd party guys around. If you think about it like 99% of progress is NOT made by big companies. Really need to think in which direction money flow could start to change.. away from AD-like and more towards guys like Redshift team :) All the best! On Fri, Mar 15, 2013 at 8:47 PM, Christian Gotzinger cgo...@googlemail.com wrote: Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca
Re: Announcing Redshift - Biased GPU Renderer
That was pretty neat! :) I can't wait to see some more test! regards stefan On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.comwrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- *Stefan Andersson | Digital Janitor* blog http://sanders3d.wordpress.com | showreelhttp://vimeo.com/sanders3d| twitter http://twitter.com/sanders3d | LinkedInhttp://www.linkedin.com/in/sanders3d| cell: +46-73-6268850 | skype:sanders3d
Re: Announcing Redshift - Biased GPU Renderer
Hi Peter, First of all, let me apologize for taking forever to respond. We've had a pretty crazy last couple of days with the alpha launch. You're absolutely right that speed is worth very little or even nothing if you can't actually get the image you or the client wants out of the damned thing, whether it's due to missing features, stability issues, limitations on content complexity, lack of flexibility or ease of use. We're very sensitive to that and while I can't claim that we're there yet, we do plan to have all the bells and whistles, stability, flexibility and ergonomics to make Redshift a legitimate choice for production rendering. That being said, speed can be important for a number of reasons. A big one is iteration times. Everything else being equal faster rendering results in better images because you have more opportunity to iterate, experiment, tweak and generally be creative. Another one is cost. This will vary a lot for different types of users, but if you suddenly don't need a render farm because your workstation renders just as fast, you've saved money. Or if you need a farm with only 100 nodes instead of 1000, you've saved some more money. I should point out that Redshift doesn't just do basic raytracing and GI but actually already supports many of the features you mentioned. We do point-based SSS, motion blur (not deformation blur yet, but we're working on that right now), instancing and refractions. For a 3rd party renderer, I would say that our support for the native Softimage shaders is probably about on par or possibly better than the others. And we're not done yet! Proper ICE support is a big one, as is proper support for AOV/render channels. Hair is another. These are all in the plan and have already had some significant thought (and in some cases initial work) put into them. -Nicolas On Thu, Mar 14, 2013 at 1:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
Re: Announcing Redshift - Biased GPU Renderer
Hi Mirko, Redshift does use the CPU for a couple things here and there so the CPU is not irrelevant to the performance, but it's not a big contributor. For example, the RT hierarchy construction (construction of the acceleration structure for raytracing) is done on the CPU as is texture conversion to our optimized tile format. Also the screen-space adaptive tessellation and some rebalancing of tree data structures for the point-based techniques run on the CPU. All in all though the GPU spec is what is really going to make the difference for Redshift, assuming we're not talking about a 486 or something :) So yeah I encourage you to dust off that old PC, pop in the GTX580 and take it for a spin. I think you'll be pleasantly surprised. -Nicolas On Fri, Mar 15, 2013 at 1:21 PM, Mirko Jankovic mirkoj.anima...@gmail.comwrote: just wondering.. if everything is on GPU in theory it should work fine even with some slower older CPU? Got some older comp laying around and both PCI slots in comp are filled, maybe could use that one for GPU rendering station for testing :) any thoughts? On Fri, Mar 15, 2013 at 9:00 PM, Len Krenzler l...@creativecontrol.cawrote: No kidding! I can't imagine why Nvidia/Arc/MR whatever they're called now couldn't have done this years ago. Between them and AD they can't even get they're basic features working. Money should be directed to these 3rd party guys as much as possible. AD must have called me about 10 times to renew my sub. This time I said no, I think I'll spend that on the guys getting results. If you haven't tried this yet, do! On 3/15/2013 1:54 PM, Mirko Jankovic wrote: Just thinking.. you really have to wonder why AD as huge company with resources that are probably hard to imagine by any of us never got to make ANYTHING nearly great as anything like what we see from 3rd party guys around. If you think about it like 99% of progress is NOT made by big companies. Really need to think in which direction money flow could start to change.. away from AD-like and more towards guys like Redshift team :) All the best! On Fri, Mar 15, 2013 at 8:47 PM, Christian Gotzinger cgo...@googlemail.com wrote: Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca
Re: Announcing Redshift - Biased GPU Renderer
no, that was last summer. if anything needs to be known about a software, an official website with actual information would be a good starting point. anyways – lets not spoil Redshift’s thread with talk about other software. From: Steven Caron Sent: Saturday, March 16, 2013 8:36 PM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer still, sitoa isn't beta anymore. peter, that was years ago when you started evaluation, right? it needs to be known that the reason it's not distributed widely isn't because arnold or sitoa is beta software. it's because they can't support everyone that wants access at this time. so no reason to make an announcement about lifting the beta tag. the sitoa list isn't called the beta list by its users or moderators anymore. sorry for derailing nic's thread. *written with my thumbs On Mar 16, 2013, at 7:08 AM, pete...@skynet.be wrote: fair enough – I was indeed referring to SItoA and MtoA , not Arnold standalone. From: Stephen Blair Sent: Saturday, March 16, 2013 12:50 PM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer SItoA beta and Mtoa beta, not Arnold beta On 16/03/2013 5:15 AM, pete...@skynet.be wrote: well, I had to join a betalist to get info and test it, downloadables were called beta, and so was the quote (afaik – it wasn’t adressed to me) – I was under the impression that it is still beta, because that’s what it said on the surface. Smoke and mirrors? Semantics? Perhaps – but not to the producer who had to make the decision on purchasing. From: Vladimir Jankijevic Sent: Saturday, March 16, 2013 12:03 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I have to back up Steven. Arnold is NOT in beta. I had more to say about this subject but it's not the place for that. I'm really curious what the Redshift guys are able to deliver for a production environment. I'll keep an eye on this for sure! On Fri, Mar 15, 2013 at 10:10 PM, Steven Caron car...@gmail.com wrote: arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.de wrote: and today it's (officially) still in beta. -- --- Vladimir Jankijevic Technical Direction Elefant Studios AG Lessingstrasse 15 CH-8002 Zürich +41 44 500 48 20 www.elefantstudios.ch ---
Re: Announcing Redshift - Biased GPU Renderer
thanks Nicolas – sounds very good. the images as well as the video look very promising – my interest is certainly aroused . From: Nicolas Burtnyk Sent: Sunday, March 17, 2013 7:33 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Hi Peter, First of all, let me apologize for taking forever to respond. We've had a pretty crazy last couple of days with the alpha launch. You're absolutely right that speed is worth very little or even nothing if you can't actually get the image you or the client wants out of the damned thing, whether it's due to missing features, stability issues, limitations on content complexity, lack of flexibility or ease of use. We're very sensitive to that and while I can't claim that we're there yet, we do plan to have all the bells and whistles, stability, flexibility and ergonomics to make Redshift a legitimate choice for production rendering. That being said, speed can be important for a number of reasons. A big one is iteration times. Everything else being equal faster rendering results in better images because you have more opportunity to iterate, experiment, tweak and generally be creative. Another one is cost. This will vary a lot for different types of users, but if you suddenly don't need a render farm because your workstation renders just as fast, you've saved money. Or if you need a farm with only 100 nodes instead of 1000, you've saved some more money. I should point out that Redshift doesn't just do basic raytracing and GI but actually already supports many of the features you mentioned. We do point-based SSS, motion blur (not deformation blur yet, but we're working on that right now), instancing and refractions. For a 3rd party renderer, I would say that our support for the native Softimage shaders is probably about on par or possibly better than the others. And we're not done yet! Proper ICE support is a big one, as is proper support for AOV/render channels. Hair is another. These are all in the plan and have already had some significant thought (and in some cases initial work) put into them. -Nicolas On Thu, Mar 14, 2013 at 1:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. wlEmoticon-winkingsmile[1].png
Re: Announcing Redshift - Biased GPU Renderer
well, I had to join a betalist to get info and test it, downloadables were called beta, and so was the quote (afaik – it wasn’t adressed to me) – I was under the impression that it is still beta, because that’s what it said on the surface. Smoke and mirrors? Semantics? Perhaps – but not to the producer who had to make the decision on purchasing. From: Vladimir Jankijevic Sent: Saturday, March 16, 2013 12:03 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I have to back up Steven. Arnold is NOT in beta. I had more to say about this subject but it's not the place for that. I'm really curious what the Redshift guys are able to deliver for a production environment. I'll keep an eye on this for sure! On Fri, Mar 15, 2013 at 10:10 PM, Steven Caron car...@gmail.com wrote: arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.de wrote: and today it's (officially) still in beta. -- --- Vladimir Jankijevic Technical Direction Elefant Studios AG Lessingstrasse 15 CH-8002 Zürich +41 44 500 48 20 www.elefantstudios.ch ---
Re: Announcing Redshift - Biased GPU Renderer
SItoA beta and Mtoa beta, not Arnold beta On 16/03/2013 5:15 AM, pete...@skynet.be wrote: well, I had to join a betalist to get info and test it, downloadables were called beta, and so was the quote (afaik – it wasn’t adressed to me) – I was under the impression that it is still beta, because that’s what it said on the surface. Smoke and mirrors? Semantics? Perhaps – but not to the producer who had to make the decision on purchasing. *From:* Vladimir Jankijevic mailto:vladi...@elefantstudios.ch *Sent:* Saturday, March 16, 2013 12:03 AM *To:* softimage@listproc.autodesk.com mailto:softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer I have to back up Steven. Arnold is NOT in beta. I had more to say about this subject but it's not the place for that. I'm really curious what the Redshift guys are able to deliver for a production environment. I'll keep an eye on this for sure! On Fri, Mar 15, 2013 at 10:10 PM, Steven Caron car...@gmail.com mailto:car...@gmail.com wrote: arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.de mailto:sixsi_l...@imagefront.de wrote: and today it's (officially) still in beta. -- --- Vladimir Jankijevic Technical Direction Elefant Studios AG Lessingstrasse 15 CH-8002 Zürich +41 44 500 48 20 www.elefantstudios.ch http://www.elefantstudios.ch ---
RE: Announcing Redshift - Biased GPU Renderer
ok, I stand corrected. Thanks for clearing that up. From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Vladimir Jankijevic Sent: Saturday, March 16, 2013 0:04 To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I have to back up Steven. Arnold is NOT in beta. I had more to say about this subject but it's not the place for that. I'm really curious what the Redshift guys are able to deliver for a production environment. I'll keep an eye on this for sure! On Fri, Mar 15, 2013 at 10:10 PM, Steven Caron car...@gmail.com wrote: arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.de wrote: and today it's (officially) still in beta. -- --- Vladimir Jankijevic Technical Direction Elefant Studios AG Lessingstrasse 15 CH-8002 Zürich +41 44 500 48 20 www.elefantstudios.ch ---
Re: Announcing Redshift - Biased GPU Renderer
fair enough – I was indeed referring to SItoA and MtoA , not Arnold standalone. From: Stephen Blair Sent: Saturday, March 16, 2013 12:50 PM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer SItoA beta and Mtoa beta, not Arnold beta On 16/03/2013 5:15 AM, pete...@skynet.be wrote: well, I had to join a betalist to get info and test it, downloadables were called beta, and so was the quote (afaik – it wasn’t adressed to me) – I was under the impression that it is still beta, because that’s what it said on the surface. Smoke and mirrors? Semantics? Perhaps – but not to the producer who had to make the decision on purchasing. From: Vladimir Jankijevic Sent: Saturday, March 16, 2013 12:03 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I have to back up Steven. Arnold is NOT in beta. I had more to say about this subject but it's not the place for that. I'm really curious what the Redshift guys are able to deliver for a production environment. I'll keep an eye on this for sure! On Fri, Mar 15, 2013 at 10:10 PM, Steven Caron car...@gmail.com wrote: arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.de wrote: and today it's (officially) still in beta. -- --- Vladimir Jankijevic Technical Direction Elefant Studios AG Lessingstrasse 15 CH-8002 Zürich +41 44 500 48 20 www.elefantstudios.ch ---
Re: Announcing Redshift - Biased GPU Renderer
still, sitoa isn't beta anymore. peter, that was years ago when you started evaluation, right? it needs to be known that the reason it's not distributed widely isn't because arnold or sitoa is beta software. it's because they can't support everyone that wants access at this time. so no reason to make an announcement about lifting the beta tag. the sitoa list isn't called the beta list by its users or moderators anymore. sorry for derailing nic's thread. *written with my thumbs On Mar 16, 2013, at 7:08 AM, pete...@skynet.be wrote: fair enough – I was indeed referring to SItoA and MtoA , not Arnold standalone. From: Stephen Blair Sent: Saturday, March 16, 2013 12:50 PM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer SItoA beta and Mtoa beta, not Arnold beta On 16/03/2013 5:15 AM, pete...@skynet.be wrote: well, I had to join a betalist to get info and test it, downloadables were called beta, and so was the quote (afaik – it w asn’t adressed to me) – I was under the impression that it is still beta, because that’s what it said on the surface. Smoke and mirrors? Semantics? Perhaps – but not to the producer who had to make the decision on purchasing. From: Vladimir Jankijevic Sent: Saturday, March 16, 2013 12:03 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I have to back up Steven. Arnold is NOT in beta. I had more to say about this subject but it's not the place for that. I'm really curious what the Redshift guys are able to deliver for a production environment. I'll keep an eye on this for sure! On Fri, Mar 15, 2013 at 10:10 PM, Steven Caron car...@gmail.com wrote: arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.de wrote: and today it's (officially) still in beta. -- --- Vladimir Jankijevic Technical Direction Elefant Studios AG Lessingstrasse 15 CH-8002 Zürich +41 44 500 48 20 www.elefantstudios.ch ---
Re: Announcing Redshift - Biased GPU Renderer
Guys, this is fantastic! Exactly the simple workflow and high speed anyone should come to expect these days! It had to be a bunch of true independent nerds to pave the path, again... (meant as a compliment!) I haven't got resources left for testing, but I'm very much looking forward to 1.0. Am 15.03.2013 03:35, schrieb Nicolas Burtnyk: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be mailto:pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer mailto:andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com mailto:softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be mailto:pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
Re: Announcing Redshift - Biased GPU Renderer
Very interesting indeed! Definitely shooting you guys an email! :) On Fri, Mar 15, 2013 at 10:50 AM, Eugen Sares softim...@keyvis.at wrote: Guys, this is fantastic! Exactly the simple workflow and high speed anyone should come to expect these days! It had to be a bunch of true independent nerds to pave the path, again... (meant as a compliment!) I haven't got resources left for testing, but I'm very much looking forward to 1.0. Am 15.03.2013 03:35, schrieb Nicolas Burtnyk: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
Re: Announcing Redshift - Biased GPU Renderer
+1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.com mailto:s...@shedmtl.com killer congrats to you and team Nicolas!! sly *Sylvain Lebeau // SHED** *V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025WWW.SHEDMTL.COM http://www.shedmtl.com/http://www.shedmtl.com/http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be mailto:pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer mailto:andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com mailto:softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be mailto:pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca
Re: Announcing Redshift - Biased GPU Renderer
hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.cawrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.com killer congrats to you and team Nicolas!! sly *Sylvain Lebeau // SHED** *V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COM http://www.shedmtl.com/ http://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca
Re: Announcing Redshift - Biased GPU Renderer
Everything is supported Mirko! It is like having the old and crumpy MR reborn with power, speed and awsome result. Integration with Softimage is seamless. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.com hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.cawrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.com killer congrats to you and team Nicolas!! sly *Sylvain Lebeau // SHED** *V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COM http://www.shedmtl.com/http://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca --
Re: Announcing Redshift - Biased GPU Renderer
u soo nice! now just to wait for OpenCL version whenever it comes.. I moved away from nvidia completely :) On Fri, Mar 15, 2013 at 2:33 PM, Emilio Hernandez emi...@e-roja.com wrote: Everything is supported Mirko! It is like having the old and crumpy MR reborn with power, speed and awsome result. Integration with Softimage is seamless. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.com hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.cawrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.com killer congrats to you and team Nicolas!! sly *Sylvain Lebeau // SHED** *V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COM http://www.shedmtl.com/http://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer. -- -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca --
Re: Announcing Redshift - Biased GPU Renderer
Well Mirko as Len said. You might just reconsider going back to Nvidia. CUDA is coming strong on a lot of apps. And getting first than ATI. You can buy a GTX 470 for 200 bucks. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.com well honestly... I'm on gaming cards because pro cards really are not justified with price in my case, and with gaming line ati right now is twice the speed of nvidia really... so just for rendering t o sacrifice all viewport performance.. I'm not sure that is something I would be willing to do :) not sure how mixing cards on same board would work with different drivers and everything to have one nvidis just for rendering.. anyway that is all different story and not really relevant in this case. in any case this is so refreshing On Fri, Mar 15, 2013 at 2:39 PM, Len Krenzler l...@creativecontrol.cawrote: You might want to move back just for this...just sayin'... On 3/15/2013 7:35 AM, Mirko Jankovic wrote: u soo nice! now just to wait for OpenCL version whenever it comes.. I moved away from nvidia completely :) On Fri, Mar 15, 2013 at 2:33 PM, Emilio Hernandez emi...@e-roja.comwrote: Everything is supported Mirko! It is like having the old and crumpy MR reborn with power, speed and awsome result. Integration with Softimage is seamless. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.com hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.cawrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.com killer congrats to you and team Nicolas!! sly *Sylvain Lebeau // SHED** *V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COMhttp://www.shedmtl.com/http://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features
Re: Announcing Redshift - Biased GPU Renderer
actualy I already have an 580 in another comp so that itself is not problem :) On Fri, Mar 15, 2013 at 3:29 PM, Emilio Hernandez emi...@e-roja.com wrote: Well Mirko as Len said. You might just reconsider going back to Nvidia. CUDA is coming strong on a lot of apps. And getting first than ATI. You can buy a GTX 470 for 200 bucks. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.com well honestly... I'm on gaming cards because pro cards really are not justified with price in my case, and with gaming line ati right now is twice the speed of nvidia really... so just for rendering t o sacrifice all viewport performance.. I'm not sure that is something I would be willing to do :) not sure how mixing cards on same board would work with different drivers and everything to have one nvidis just for rendering.. anyway that is all different story and not really relevant in this case. in any case this is so refreshing On Fri, Mar 15, 2013 at 2:39 PM, Len Krenzler l...@creativecontrol.cawrote: You might want to move back just for this...just sayin'... On 3/15/2013 7:35 AM, Mirko Jankovic wrote: u soo nice! now just to wait for OpenCL version whenever it comes.. I moved away from nvidia completely :) On Fri, Mar 15, 2013 at 2:33 PM, Emilio Hernandez emi...@e-roja.comwrote: Everything is supported Mirko! It is like having the old and crumpy MR reborn with power, speed and awsome result. Integration with Softimage is seamless. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.com hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.ca wrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.com killer congrats to you and team Nicolas!! sly *Sylvain Lebeau // SHED** *V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COMhttp://www.shedmtl.com/http://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite
RE: Announcing Redshift - Biased GPU Renderer
I wonder how this looks with a render region with alpha blending turned on. The renderer would need to output RGBA and support the render region. Does it? GL From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Len Krenzler Sent: Friday, March 15, 2013 11:13 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I only have a GTX470 and it flies even with that! I'm testing a scene right now with 4.5 mil polys and a 12k HDR lighting texture as well as other large textures and no problem. On 3/15/2013 8:58 AM, Tim Crowson wrote: I've been really impressed with the performance and integration so far. I still need to throw some heavy scenes at it thow. But considering what it can do on a single card, I can't wait to see how it will run once multiple cards are supported. Either way, this is already a win for the Softimage community. Big thanks to Nicolas and his team! -Tim On 3/15/2013 9:32 AM, Mirko Jankovic wrote: actualy I already have an 580 in another comp so that itself is not problem :) On Fri, Mar 15, 2013 at 3:29 PM, Emilio Hernandez emi...@e-roja.commailto:emi...@e-roja.com wrote: Well Mirko as Len said. You might just reconsider going back to Nvidia. CUDA is coming strong on a lot of apps. And getting first than ATI. You can buy a GTX 470 for 200 bucks. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.commailto:mirkoj.anima...@gmail.com well honestly... I'm on gaming cards because pro cards really are not justified with price in my case, and with gaming line ati right now is twice the speed of nvidia really... so just for rendering t o sacrifice all viewport performance.. I'm not sure that is something I would be willing to do :) not sure how mixing cards on same board would work with different drivers and everything to have one nvidis just for rendering.. anyway that is all different story and not really relevant in this case. in any case this is so refreshing On Fri, Mar 15, 2013 at 2:39 PM, Len Krenzler l...@creativecontrol.camailto:l...@creativecontrol.ca wrote: You might want to move back just for this...just sayin'... On 3/15/2013 7:35 AM, Mirko Jankovic wrote: u soo nice! now just to wait for OpenCL version whenever it comes.. I moved away from nvidia completely :) On Fri, Mar 15, 2013 at 2:33 PM, Emilio Hernandez emi...@e-roja.commailto:emi...@e-roja.com wrote: Everything is supported Mirko! It is like having the old and crumpy MR reborn with power, speed and awsome result. Integration with Softimage is seamless. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.commailto:mirkoj.anima...@gmail.com hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.camailto:l...@creativecontrol.ca wrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.commailto:s...@shedmtl.com killer congrats to you and team Nicolas!! sly Sylvain Lebeau // SHED V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COMhttp://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.bemailto:pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. From: Andy Moorermailto:andymoo...@gmail.com Sent: Thursday, March 14, 2013 9:02 PM To: softimage@listproc.autodesk.commailto:softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.bemailto:pete
Re: Announcing Redshift - Biased GPU Renderer
This is a great achievement Nicolas . It's really surprise how many amazing tech are available in those days ( Fabric Engine - Arnold - Alembic ) and who do it is not the biggest (resources, money) company , this renderer look promising and will be on of them , keep up the great work guys . On Fri, Mar 15, 2013 at 8:31 AM, Guillaume Laferriere guillaume.laferri...@autodesk.com wrote: I wonder how this looks with a render region with alpha blending turned on. The renderer would need to output RGBA and support the render region. Does it? GL From: softimage-boun...@listproc.autodesk.com [mailto: softimage-boun...@listproc.autodesk.com] On Behalf Of Len Krenzler Sent: Friday, March 15, 2013 11:13 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I only have a GTX470 and it flies even with that! I'm testing a scene right now with 4.5 mil polys and a 12k HDR lighting texture as well as other large textures and no problem. On 3/15/2013 8:58 AM, Tim Crowson wrote: I've been really impressed with the performance and integration so far. I still need to throw some heavy scenes at it thow. But considering what it can do on a single card, I can't wait to see how it will run once multiple cards are supported. Either way, this is already a win for the Softimage community. Big thanks to Nicolas and his team! -Tim On 3/15/2013 9:32 AM, Mirko Jankovic wrote: actualy I already have an 580 in another comp so that itself is not problem :) On Fri, Mar 15, 2013 at 3:29 PM, Emilio Hernandez emi...@e-roja.com mailto:emi...@e-roja.com wrote: Well Mirko as Len said. You might just reconsider going back to Nvidia. CUDA is coming strong on a lot of apps. And getting first than ATI. You can buy a GTX 470 for 200 bucks. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.commailto: mirkoj.anima...@gmail.com well honestly... I'm on gaming cards because pro cards really are not justified with price in my case, and with gaming line ati right now is twice the speed of nvidia really... so just for rendering t o sacrifice all viewport performance.. I'm not sure that is something I would be willing to do :) not sure how mixing cards on same board would work with different drivers and everything to have one nvidis just for rendering.. anyway that is all different story and not really relevant in this case. in any case this is so refreshing On Fri, Mar 15, 2013 at 2:39 PM, Len Krenzler l...@creativecontrol.ca mailto:l...@creativecontrol.ca wrote: You might want to move back just for this...just sayin'... On 3/15/2013 7:35 AM, Mirko Jankovic wrote: u soo nice! now just to wait for OpenCL version whenever it comes.. I moved away from nvidia completely :) On Fri, Mar 15, 2013 at 2:33 PM, Emilio Hernandez emi...@e-roja.com mailto:emi...@e-roja.com wrote: Everything is supported Mirko! It is like having the old and crumpy MR reborn with power, speed and awsome result. Integration with Softimage is seamless. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.commailto: mirkoj.anima...@gmail.com hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.ca mailto:l...@creativecontrol.ca wrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.commailto:s...@shedmtl.com killer congrats to you and team Nicolas!! sly Sylvain Lebeau // SHED V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COMhttp://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.bemailto: pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. From: Andy Moorermailto:andymoo...@gmail.com Sent: Thursday, March 14, 2013 9:02 PM To: softimage@listproc.autodesk.commailto:softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative
Re: Announcing Redshift - Biased GPU Renderer
Yes it does. On 3/15/2013 9:31 AM, Guillaume Laferriere wrote: I wonder how this looks with a render region with alpha blending turned on. The renderer would need to output RGBA and support the render region. Does it? GL From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Len Krenzler Sent: Friday, March 15, 2013 11:13 AM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer I only have a GTX470 and it flies even with that! I'm testing a scene right now with 4.5 mil polys and a 12k HDR lighting texture as well as other large textures and no problem. On 3/15/2013 8:58 AM, Tim Crowson wrote: I've been really impressed with the performance and integration so far. I still need to throw some heavy scenes at it thow. But considering what it can do on a single card, I can't wait to see how it will run once multiple cards are supported. Either way, this is already a win for the Softimage community. Big thanks to Nicolas and his team! -Tim On 3/15/2013 9:32 AM, Mirko Jankovic wrote: actualy I already have an 580 in another comp so that itself is not problem :) On Fri, Mar 15, 2013 at 3:29 PM, Emilio Hernandez emi...@e-roja.commailto:emi...@e-roja.com wrote: Well Mirko as Len said. You might just reconsider going back to Nvidia. CUDA is coming strong on a lot of apps. And getting first than ATI. You can buy a GTX 470 for 200 bucks. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.commailto:mirkoj.anima...@gmail.com well honestly... I'm on gaming cards because pro cards really are not justified with price in my case, and with gaming line ati right now is twice the speed of nvidia really... so just for rendering t o sacrifice all viewport performance.. I'm not sure that is something I would be willing to do :) not sure how mixing cards on same board would work with different drivers and everything to have one nvidis just for rendering.. anyway that is all different story and not really relevant in this case. in any case this is so refreshing On Fri, Mar 15, 2013 at 2:39 PM, Len Krenzler l...@creativecontrol.camailto:l...@creativecontrol.ca wrote: You might want to move back just for this...just sayin'... On 3/15/2013 7:35 AM, Mirko Jankovic wrote: u soo nice! now just to wait for OpenCL version whenever it comes.. I moved away from nvidia completely :) On Fri, Mar 15, 2013 at 2:33 PM, Emilio Hernandez emi...@e-roja.commailto:emi...@e-roja.com wrote: Everything is supported Mirko! It is like having the old and crumpy MR reborn with power, speed and awsome result. Integration with Softimage is seamless. 2013/3/15 Mirko Jankovic mirkoj.anima...@gmail.commailto:mirkoj.anima...@gmail.com hey I haven't really seen if region rendering is supported as well or only preview window? just wondering On Fri, Mar 15, 2013 at 2:17 PM, Len Krenzler l...@creativecontrol.camailto:l...@creativecontrol.ca wrote: +1! Absolutely out of this world! How you guys got all this done so fast is mind blowing. Integrated into SI too, not just an export plugin. This is truly ground breaking! On 3/14/2013 10:06 PM, Emilio Hernandez wrote: Let me tell you that I just put my hands on this baby and wow!!! This is going to rock the rendering world. And for Softimage Awsome guys congratulations on this one. My quadro 3000 finally is awake!!! 2013/3/14 Sylvain Lebeau s...@shedmtl.commailto:s...@shedmtl.com killer congrats to you and team Nicolas!! sly Sylvain Lebeau // SHED V-P/Visual effects supervisor 1410, RUE STANLEY, 11E ÉTAGE MONTRÉAL (QUÉBEC) H3A 1P8 T 514 849-1555 F 514 849-5025 WWW.SHEDMTL.COMhttp://www.shedmtl.com/ http://WWW.SHEDMTL.COM On 3/14/2013 10:35 PM, Nicolas Burtnyk wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.bemailto:pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. From: Andy Moorermailto:andymoo...@gmail.com Sent: Thursday, March 14, 2013 9:02 PM To: softimage@listproc.autodesk.commailto:softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad
Re: Announcing Redshift - Biased GPU Renderer
Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.comwrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas
Re: Announcing Redshift - Biased GPU Renderer
Just thinking.. you really have to wonder why AD as huge company with resources that are probably hard to imagine by any of us never got to make ANYTHING nearly great as anything like what we see from 3rd party guys around. If you think about it like 99% of progress is NOT made by big companies. Really need to think in which direction money flow could start to change.. away from AD-like and more towards guys like Redshift team :) All the best! On Fri, Mar 15, 2013 at 8:47 PM, Christian Gotzinger cgo...@googlemail.comwrote: Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.comwrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas
Re: Announcing Redshift - Biased GPU Renderer
No kidding! I can't imagine why Nvidia/Arc/MR whatever they're called now couldn't have done this years ago. Between them and AD they can't even get they're basic features working. Money should be directed to these 3rd party guys as much as possible. AD must have called me about 10 times to renew my sub. This time I said no, I think I'll spend that on the guys getting results. If you haven't tried this yet, do! On 3/15/2013 1:54 PM, Mirko Jankovic wrote: Just thinking.. you really have to wonder why AD as huge company with resources that are probably hard to imagine by any of us never got to make ANYTHING nearly great as anything like what we see from 3rd party guys around. If you think about it like 99% of progress is NOT made by big companies. Really need to think in which direction money flow could start to change.. away from AD-like and more towards guys like Redshift team :) All the best! On Fri, Mar 15, 2013 at 8:47 PM, Christian Gotzinger cgo...@googlemail.com mailto:cgo...@googlemail.com wrote: Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.com mailto:nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas -- _ Len Krenzler - Creative Control Media Productions Phone: 780.463.3126 www.creativecontrol.ca - l...@creativecontrol.ca
RE: Announcing Redshift - Biased GPU Renderer
It is not that simple. I think to make a renderer that looks very promising is one thing. Establish it in the market is the hard part. For example lets take Arnold. it took them over ten years to make it something we consider a product and today it's (officially) still in beta. There were other renderers (I don't remember right now. Brazil was one of them). Great renderer, faster than some others. Now its abandoned. This was actually a renderer used some years, but there were many others that didn't survive their first couple of years while in developing. Redshift looks indeed very nice and promising. I hope it will make its way into the market. sven From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Mirko Jankovic Sent: Friday, March 15, 2013 20:54 To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Just thinking.. you really have to wonder why AD as huge company with resources that are probably hard to imagine by any of us never got to make ANYTHING nearly great as anything like what we see from 3rd party guys around. If you think about it like 99% of progress is NOT made by big companies. Really need to think in which direction money flow could start to change.. away from AD-like and more towards guys like Redshift team :) All the best! On Fri, Mar 15, 2013 at 8:47 PM, Christian Gotzinger cgo...@googlemail.com wrote: Excuse the language, but: Holy shit! Mighty impressive stuff! On Fri, Mar 15, 2013 at 3:35 AM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas
Re: Announcing Redshift - Biased GPU Renderer
arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.dewrote: and today it's (officially) still in beta.
Re: Announcing Redshift - Biased GPU Renderer
I have to back up Steven. Arnold is NOT in beta. I had more to say about this subject but it's not the place for that. I'm really curious what the Redshift guys are able to deliver for a production environment. I'll keep an eye on this for sure! On Fri, Mar 15, 2013 at 10:10 PM, Steven Caron car...@gmail.com wrote: arnold is NOT in beta... but your point about market success is made. On Fri, Mar 15, 2013 at 1:46 PM, Sven Constable sixsi_l...@imagefront.dewrote: and today it's (officially) still in beta. -- --- Vladimir Jankijevic Technical Direction Elefant Studios AG Lessingstrasse 15 CH-8002 Zürich +41 44 500 48 20 www.elefantstudios.ch ---
Re: Announcing Redshift - Biased GPU Renderer
Hi Manuel, Redshift doesn't currently support multiple framebuffers (or render elements), but it's in the plan. -Nicolas On Wed, Mar 13, 2013 at 7:53 PM, Manuel Huertas Marchena lito...@hotmail.com wrote: Hi Nicolas, This looks very interesting, props to you guys, I am looking forward for the video as well, if possible! I have a question, what about framebuffer support in softimage with redshif, is it similar worflow as mr/vray? ...Haven't found that on the documentation, maybe I missed that. Thanks Cheers -Manuel Date: Thu, 14 Mar 2013 13:01:28 +1100 Subject: Re: Announcing Redshift - Biased GPU Renderer From: ahmidou@gmail.com To: softimage@listproc.autodesk.com 77 secs for the Living room, that's impressive!! --- Ahmidou Lyazidi Director | TD | CG artist http://vimeo.com/ahmidou/videos 2013/3/14 Nicolas Burtnyk nico...@redshift3d.com: Hey guys, Thanks for the great responses. I think I've responded to everyone who sent an alpha request, but if you think I missed you, please shoot me an email to remind me :) Also, I wanted to share some render times as we ran some more tests this afternoon comparing the GTX 470, GTX 670 and GTX Titan (which we actually received after the announcement went out) for the scenes we posted. Some of this info is also on our announcement thread on CGTalk (http://forums.cgsociety.org/showthread.php?f=59t=1098062). Gargoyle 1280x720 (jp_studio_icp_1280.png) GTX 470: 35 seconds GTX 670: 27 seconds GTX Titan: 17 seconds Car 1024x683 (mazda_1024.png) GTX 470: 75 seconds GTX 670: 65 seconds GTX Titan: 39 seconds Evermotion Living Room 1200x1000 (AI_V8_S10_1200.png) GTX 470: 155 seconds GTX 670: 123 seconds GTX Titan: 77 seconds Classroom 1024x512 (classroom.png) GTX 470: 129 seconds - ok I exaggerated a bit when I said 2 minutes :) GTX 670: 96 seconds GTX Titan: 49 seconds Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) On Wed, Mar 13, 2013 at 12:17 PM, Len Krenzler l...@creativecontrol.ca wrote: Exactly! This is VERY interesting. Hope I can test (sent request already) :) On 3/13/2013 12:33 PM, Nicolas Burtnyk wrote: Yes - I'll try to make a video of that if I can get set up correctly for it. Note that this is 2 mins on a GTX 470 which is nothing special in terms of GPUs. You can expect significantly better times with a GTX 580 for example. I don't have official times for that card, but I'd guess under 1.5 minutes. These kinds of times really underscore the power of biased rendering. When you need to reduce noise, you have a lot more options than let's just throw a ton more samples at the whole thing. On Wed, Mar 13, 2013 at 11:28 AM, olivier jeannel olivier.jean...@noos.fr wrote: The classroom is really 2min render ? Congrats to you, sending a request :) Le 13/03/2013 19:18, Steven Caron a écrit : congrats to you and your team! i was wondering when we would see/hear about your work. it would be great to see a video demonstration of redshift in softimage. On Wed, Mar 13, 2013 at 11:12 AM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hello folks, In March of last year, 2 colleagues and I left our jobs as software developers in the games industry to form our own company - Redshift. Our goal was to apply our experience with graphics hardware to the problem of offline rendering. Artists friends had been asking us for years why Mental Ray and other renderers were not taking advantage of the GPU. As the ideas bounced around in our heads, we figured we'd take a crack at it. As it turns out, it's really freakin' hard, but not impossible! Today, we're very excited to announce the official launch of Redshift v0.1 alpha, to our knowledge, the world's first fully GPU-accelerated biased renderer. Redshift supports multiple GI solutions: Brute-Force GI, Irradiance Caching (aka Final Gather), Irradiance Point Cloud (aka Light Cache) and Photon Mapping (GI and Caustics). All are fully GPU-accelerated and perform many times faster than similar CPU-based offerings. A problem that plagues many GPU renderers on the market is that they are limited by the available VRAM on the graphics card (and most systems have significantly less VRAM than main memory). Redshift addresses this by using an out-of-core architecture for geometry and textures allowing you to render scenes with tens of millions of polygons and gigabytes of textures with off-the-shelf, inexpensive hardware. Redshift currently integrates directly with Softimage 2011 through 2013 and Maya 2011 through 2013 on Windows
Re: Announcing Redshift - Biased GPU Renderer
Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
Re: Announcing Redshift - Biased GPU Renderer
you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. From: Andy Moorer Sent: Thursday, March 14, 2013 9:02 PM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
Re: Announcing Redshift - Biased GPU Renderer
Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
Re: Announcing Redshift - Biased GPU Renderer
thanks a lot for the video, gives me a good idea of the integration. honestly i have no time for testing... but when you guys announce a price i will see if its right for me to jump on :) On Thu, Mar 14, 2013 at 7:35 PM, Nicolas Burtnyk nico...@redshift3d.comwrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. *From:* Andy Moorer andymoo...@gmail.com *Sent:* Thursday, March 14, 2013 9:02 PM *To:* softimage@listproc.autodesk.com *Subject:* Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
Re: Announcing Redshift - Biased GPU Renderer
Wow ! Sent from my iPhone On 2013-03-14, at 10:35 PM, Nicolas Burtnyk nico...@redshift3d.com wrote: Hey guys, I'm going to respond to the last few messages regarding the importance of speed later, but in the meantime here is a video of some live rendering in Softimage. http://youtu.be/fjCguRdSlV0 -Nicolas On Thu, Mar 14, 2013 at 1:17 PM, pete...@skynet.be wrote: you are right of course, as always. what is really needed is a fine balance between quality and speed, at a pricepoint that is affordable yet high enough to sustain development, and available before my retirement. From: Andy Moorer Sent: Thursday, March 14, 2013 9:02 PM To: softimage@listproc.autodesk.com Subject: Re: Announcing Redshift - Biased GPU Renderer Well said, but speed is still important, deadlines are tight and particularly in the iterative direction phase often re-rendering takes much more time than making a directed change. Dailies reflect this... A series of several directed tweaks to a shot can stretch over several days in part to allow time to make changes and get them rendered... A major limitation to working with rendered VFX elements versus composite effects which can often be altered in near realtime. Sent from my iPad On Mar 14, 2013, at 4:21 AM, pete...@skynet.be wrote: Please also bear in mind that we're still just in alpha and constantly improving performance. We're kind of obsessed with speed :) speed is great of course – but IMO it’s not the most important factor. over the years we have all been doing productions with rather long rendertimes, running into hours per frame and more. The bottom line was rarely “it has to be rendered in X amount of time” – clients couldn’t care less. It has to be good enough first and rendered in time for delivery. it’s been a long time I’m looking forward for a viewport/GPU mental ray replacement in softimage. Hopefully staying below 5 minutes for complex HD images and within 1 minute for more simple stuff – but more importantly, it should have the bells and whistles of a modern raytracer, and deliver production quality rendering – that can be very precisely tweaked by the user. It’s very frustrating to get a promising image very fast, but not being able to make the image really final - some remaining artifacts, sampling problem or no ability to finetune this or that effect or simply lack of a feature you really require – so in turn you have to bite the bullet and go back to good old offline rendering – and the corresponding rendertimes will be twice as frustrating. Very extensive support for lighting features – not just GI / AO / softshadows / softreflections – but also SSS, raytraced refractions, motion blur, volumetrics, ICE support, instancing, hair – and a good set of shaders and support for the rendertree and as many of the factory shaders as possible. Mental ray never became the standard it was because of speed – but because of what one can achieve with it. (and then you have to turn off a few things left and right for final renders in order to make rendertimes acceptable) Obviously in this day and age it’s features are getting long in the tooth as well, which opens the door wide open for others – but it remains a reference for what a renderer should at least aspire to. just some thoughts and hints of what matters to me when considering a new renderer.
