RE: Particle collision behaves differently after particle count 4000
Hi Stefan, Thanks for the repro steps and I am able to repro the crash. As for Why does the output type change at all? The instance is a piece of geometry just like the non-instanced sphere, or at least it should be interpreted like that, I will file this as an improvement. Many thanks. Chris -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Stefan Kubicek Sent: Sunday, August 12, 2012 3:39 PM To: softimage@listproc.autodesk.com Subject: Re: Particle collision behaves differently after particle count 4000 Hi Chris, here are the steps to reproduce: 1 Create a point cloud with an ice tree (simulated or not does not matter). 2 Create a sphere and make a model from it, make an instance of that model. 3 Create another sphere without a model, create a group with the second sphere as it's _only_ member. 4 In the ice tree of the point cloud, create an Emit from Geometry node, connect it to the ICE Tree's Port 1. 5 From an explorer, drag the group containing the sphere into the point cloud's ICE tree (Get Group node is created), plug it into the Emitter 1 port of the Emit from Geometry node. 6 In the explorer, drag the model instance we created before into the group - Soft crashes. Thanks for looking into it. From what I can tell the Get Data node tries to adapt to the changing input data by changing the output data type. Something downstream doesn't seem to like that. The question is: Why does the output type change at all? The instance is a piece of geometry just like the non-instanced sphere, or at least it should be interpreted like that. That seems to crash both 2012 and 2013. It seems nobody ever tried emitting particles from instanced geometry? Stefan Hi Stefan, Would you mind if you could share the 'group with model having referenced iceteee' crash scene? And did you work on 2013SP1? Chris On 8 Aug, 2012, at 5:14 AM, Stefan Kubicek s...@tidbit-images.com wrote: Guys, I just had a different hickup with the exact same setup where suddenly collision with one of my collision objects would suddenly not work anymore. After saving and reloading the scene it went back to normal behavior. Something _is_ afoul with Bullet RBD in ICE, see my other post from yesterday regarding different issues. And I just hit yet another one: When assigning an instanced model to a group thats already referenced in an ICE tree (e.g. as a GetData node from a drag and drop action from explorer into the ice tree) and that node is already connected to another node Softimage will bomb out to the CER form and then exit. Correction: This one has nothing to do with Bullet RBD, it even happens with the basic Emit from geometry preset ICE Tree. Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen -- --- Stefan Kubicek Co-founder --- keyvis digital imagery Wehrgasse 9 - Grüner Hof 1050 Vienna Austria Phone:+43/699/12614231 --- www.keyvis.at ste...@keyvis.at --- -- This email and its attachments are --confidential and for the recipient only-- -- --- Stefan Kubicek Co-founder --- keyvis digital imagery Wehrgasse 9 - Grüner Hof 1050 Vienna Austria Phone:+43/699/12614231 --- www.keyvis.at ste...@keyvis.at --- -- This email and its attachments are --confidential and for the recipient only-- attachment: winmail.dat
Re: Particle collision behaves differently after particle count 4000
Hi Chris, here are the steps to reproduce: 1 Create a point cloud with an ice tree (simulated or not does not matter). 2 Create a sphere and make a model from it, make an instance of that model. 3 Create another sphere without a model, create a group with the second sphere as it's _only_ member. 4 In the ice tree of the point cloud, create an Emit from Geometry node, connect it to the ICE Tree's Port 1. 5 From an explorer, drag the group containing the sphere into the point cloud's ICE tree (Get Group node is created), plug it into the Emitter 1 port of the Emit from Geometry node. 6 In the explorer, drag the model instance we created before into the group - Soft crashes. Thanks for looking into it. From what I can tell the Get Data node tries to adapt to the changing input data by changing the output data type. Something downstream doesn't seem to like that. The question is: Why does the output type change at all? The instance is a piece of geometry just like the non-instanced sphere, or at least it should be interpreted like that. That seems to crash both 2012 and 2013. It seems nobody ever tried emitting particles from instanced geometry? Stefan Hi Stefan, Would you mind if you could share the 'group with model having referenced iceteee' crash scene? And did you work on 2013SP1? Chris On 8 Aug, 2012, at 5:14 AM, Stefan Kubicek s...@tidbit-images.com wrote: Guys, I just had a different hickup with the exact same setup where suddenly collision with one of my collision objects would suddenly not work anymore. After saving and reloading the scene it went back to normal behavior. Something _is_ afoul with Bullet RBD in ICE, see my other post from yesterday regarding different issues. And I just hit yet another one: When assigning an instanced model to a group thats already referenced in an ICE tree (e.g. as a GetData node from a drag and drop action from explorer into the ice tree) and that node is already connected to another node Softimage will bomb out to the CER form and then exit. Correction: This one has nothing to do with Bullet RBD, it even happens with the basic Emit from geometry preset ICE Tree. Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen -- --- Stefan Kubicek Co-founder --- keyvis digital imagery Wehrgasse 9 - Grüner Hof 1050 Vienna Austria Phone:+43/699/12614231 --- www.keyvis.at ste...@keyvis.at --- -- This email and its attachments are --confidential and for the recipient only-- -- --- Stefan Kubicek Co-founder --- keyvis digital imagery Wehrgasse 9 - Grüner Hof 1050 Vienna Austria Phone:+43/699/12614231 --- www.keyvis.at ste...@keyvis.at --- -- This email and its attachments are --confidential and for the recipient only--
Re: Particle collision behaves differently after particle count 4000
Great! Thank you, Chris. Am 10.08.2012 12:48, schrieb Chris Chia: Hi Eugen, The bullet rigid bodies defect has been logged. We will quickly look into it. Regards, Chris -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Eugen Sares Sent: Thursday, August 09, 2012 5:56 PM To: softimage@listproc.autodesk.com Subject: Re: Particle collision behaves differently after particle count 4000 Am 09.08.2012 10:46, schrieb Chris Chia: Hi Eugen, Are you on SI2013SP1? Yes. I will check with the other version if needed... Chris On 9 Aug, 2012, at 2:27 PM, Eugen Sares softim...@keyvis.at wrote: Hello Chris, tried a precision value of 1000. Still the spheres overlap. Also, thought that a slower trajectory might give the collision check a better chance, and reduced gravity. Does not help. That burst effect was collision check suddenly working at a certain particle count, but I could not reproduce it any more. - a bug most certainly, don't you think? Thanks! Best regards, Eugen Try tweaking the precision value if you strictly don't want any particles to touch one another. A quick prove: something like 1000 should fix the penetration problem; but of course resulting in an extreme slowdown in performance. And no out burst at frames when running that simulation. Cheers. Chris On 8 Aug, 2012, at 12:38 AM, Eugen Sares softim...@keyvis.at wrote: Am 07.08.2012 17:27, schrieb Jeff McFall: I ran the sim all the way through and it looks correct here. None of the burst effect from the video. Thanks for answering, Jeff! Do the particles penetrate each other? They do here. in your video example it almost looks like the particle count suddenly jumps Nope, the count is continuous. Strangely, after closing and opening the scene, the burst effect at frame 200 does not show up any more, meaning the particles overlap quite often. Weird... Precision is set to 60 in the Simulate node. Am I missing some other parameter relevant for collision? Jeff -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Eugen Sares Sent: Tuesday, August 07, 2012 9:44 AM To: softimage@listproc.autodesk.com Subject: Particle collision behaves differently after particle count 4000 Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen
Re: Particle collision behaves differently after particle count 4000
Hi Eugen, Are you on SI2013SP1? I will check with the other version if needed... Chris On 9 Aug, 2012, at 2:27 PM, Eugen Sares softim...@keyvis.at wrote: Hello Chris, tried a precision value of 1000. Still the spheres overlap. Also, thought that a slower trajectory might give the collision check a better chance, and reduced gravity. Does not help. That burst effect was collision check suddenly working at a certain particle count, but I could not reproduce it any more. - a bug most certainly, don't you think? Thanks! Best regards, Eugen Try tweaking the precision value if you strictly don't want any particles to touch one another. A quick prove: something like 1000 should fix the penetration problem; but of course resulting in an extreme slowdown in performance. And no out burst at frames when running that simulation. Cheers. Chris On 8 Aug, 2012, at 12:38 AM, Eugen Sares softim...@keyvis.at wrote: Am 07.08.2012 17:27, schrieb Jeff McFall: I ran the sim all the way through and it looks correct here. None of the burst effect from the video. Thanks for answering, Jeff! Do the particles penetrate each other? They do here. in your video example it almost looks like the particle count suddenly jumps Nope, the count is continuous. Strangely, after closing and opening the scene, the burst effect at frame 200 does not show up any more, meaning the particles overlap quite often. Weird... Precision is set to 60 in the Simulate node. Am I missing some other parameter relevant for collision? Jeff -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Eugen Sares Sent: Tuesday, August 07, 2012 9:44 AM To: softimage@listproc.autodesk.com Subject: Particle collision behaves differently after particle count 4000 Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen attachment: winmail.dat
Re: Particle collision behaves differently after particle count 4000
Am 09.08.2012 10:46, schrieb Chris Chia: Hi Eugen, Are you on SI2013SP1? Yes. I will check with the other version if needed... Chris On 9 Aug, 2012, at 2:27 PM, Eugen Sares softim...@keyvis.at wrote: Hello Chris, tried a precision value of 1000. Still the spheres overlap. Also, thought that a slower trajectory might give the collision check a better chance, and reduced gravity. Does not help. That burst effect was collision check suddenly working at a certain particle count, but I could not reproduce it any more. - a bug most certainly, don't you think? Thanks! Best regards, Eugen Try tweaking the precision value if you strictly don't want any particles to touch one another. A quick prove: something like 1000 should fix the penetration problem; but of course resulting in an extreme slowdown in performance. And no out burst at frames when running that simulation. Cheers. Chris On 8 Aug, 2012, at 12:38 AM, Eugen Sares softim...@keyvis.at wrote: Am 07.08.2012 17:27, schrieb Jeff McFall: I ran the sim all the way through and it looks correct here. None of the burst effect from the video. Thanks for answering, Jeff! Do the particles penetrate each other? They do here. in your video example it almost looks like the particle count suddenly jumps Nope, the count is continuous. Strangely, after closing and opening the scene, the burst effect at frame 200 does not show up any more, meaning the particles overlap quite often. Weird... Precision is set to 60 in the Simulate node. Am I missing some other parameter relevant for collision? Jeff -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Eugen Sares Sent: Tuesday, August 07, 2012 9:44 AM To: softimage@listproc.autodesk.com Subject: Particle collision behaves differently after particle count 4000 Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen
Re: Particle collision behaves differently after particle count 4000
Hi Stefan, Would you mind if you could share the 'group with model having referenced iceteee' crash scene? And did you work on 2013SP1? Chris On 8 Aug, 2012, at 5:14 AM, Stefan Kubicek s...@tidbit-images.com wrote: Guys, I just had a different hickup with the exact same setup where suddenly collision with one of my collision objects would suddenly not work anymore. After saving and reloading the scene it went back to normal behavior. Something _is_ afoul with Bullet RBD in ICE, see my other post from yesterday regarding different issues. And I just hit yet another one: When assigning an instanced model to a group thats already referenced in an ICE tree (e.g. as a GetData node from a drag and drop action from explorer into the ice tree) and that node is already connected to another node Softimage will bomb out to the CER form and then exit. Correction: This one has nothing to do with Bullet RBD, it even happens with the basic Emit from geometry preset ICE Tree. Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen -- --- Stefan Kubicek Co-founder --- keyvis digital imagery Wehrgasse 9 - Grüner Hof 1050 Vienna Austria Phone:+43/699/12614231 --- www.keyvis.at ste...@keyvis.at --- -- This email and its attachments are --confidential and for the recipient only-- attachment: winmail.dat
RE: Particle collision behaves differently after particle count 4000
I ran the sim all the way through and it looks correct here. None of the burst effect from the video. in your video example it almost looks like the particle count suddenly jumps Jeff -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Eugen Sares Sent: Tuesday, August 07, 2012 9:44 AM To: softimage@listproc.autodesk.com Subject: Particle collision behaves differently after particle count 4000 Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen
RE: Particle collision behaves differently after particle count 4000
they do not appear to intersect in my sim here. I did notice that they do seem to intersect prior to the burst in your video, but I could not tell for sure. Also I tried to compare the last frame resting positions of the particles in my sim and they were close but not exact to that you posted, for whatever that is worth. Strange effect as it seems to be a pretty basic setup. Precision here is 60 as well. That is weird.I would think it should work as is. Reboot and a fresh cup of coffee? -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Eugen Sares Sent: Tuesday, August 07, 2012 12:38 PM To: softimage@listproc.autodesk.com Subject: Re: Particle collision behaves differently after particle count 4000 Am 07.08.2012 17:27, schrieb Jeff McFall: I ran the sim all the way through and it looks correct here. None of the burst effect from the video. Thanks for answering, Jeff! Do the particles penetrate each other? They do here. in your video example it almost looks like the particle count suddenly jumps Nope, the count is continuous. Strangely, after closing and opening the scene, the burst effect at frame 200 does not show up any more, meaning the particles overlap quite often. Weird... Precision is set to 60 in the Simulate node. Am I missing some other parameter relevant for collision? Jeff -Original Message- From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Eugen Sares Sent: Tuesday, August 07, 2012 9:44 AM To: softimage@listproc.autodesk.com Subject: Particle collision behaves differently after particle count 4000 Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen
Re: Particle collision behaves differently after particle count 4000
Guys, I just had a different hickup with the exact same setup where suddenly collision with one of my collision objects would suddenly not work anymore. After saving and reloading the scene it went back to normal behavior. Something _is_ afoul with Bullet RBD in ICE, see my other post from yesterday regarding different issues. And I just hit yet another one: When assigning an instanced model to a group thats already referenced in an ICE tree (e.g. as a GetData node from a drag and drop action from explorer into the ice tree) and that node is already connected to another node Softimage will bomb out to the CER form and then exit. Correction: This one has nothing to do with Bullet RBD, it even happens with the basic Emit from geometry preset ICE Tree. Hi, I've got a simple setup here with Simulate Bullet Rigid Bodies, where I'm trying to fill a volume with small spherical particles (radius 0,1). Testing 2013SP1. As long as the particle count is below 4000 (for whatever reason), particles tend to penetrate/clump. Suddenly at the frame where the count reaches 4000, all particles push themselves out of their neighbours at once, then simulation continues normally, with no penetrations. A bug? I didn't touch any simulation range settings. Video (possibly not converted yet): https://vimeo.com/47089050 Scene: http://www.sendspace.com/file/c0xke5 Thanks a lot! Eugen -- --- Stefan Kubicek Co-founder --- keyvis digital imagery Wehrgasse 9 - Grüner Hof 1050 Vienna Austria Phone:+43/699/12614231 --- www.keyvis.at ste...@keyvis.at --- -- This email and its attachments are --confidential and for the recipient only--
