Re: [Nuke-users] GPU FFT?
I have made one small video about it in my last training at fxphd. To remove a bad signal baked in the image. Not using it everyday, but might works like magic in some cases Sent from my iPhone > On Mar 20, 2017, at 2:17 PM, Ron Ganbarwrote: > > He's a question: > Can somebody properly explain FFT? > (I'm feeling so inadequate right now. Am I the only one left who doesn't know > what this magical acronym is?) > > > > > > Ron Ganbar > email: ron...@gmail.com > tel: +44 (0)7968 007 309 [UK] > +972 (0)54 255 9765 [Israel] > url: http://ronganbar.wordpress.com/ > >> On Mon, Mar 20, 2017 at 3:42 PM, Magno Borgo wrote: >> A non-nuke alternative would be to pre-render the FFT using a python package >> like scipy or similar, which already have fast FFT implementations, then >> load/read it inside NUKE. That might help on some pipelines, though the >> inverse edited FFT would still be needed to handled inside Nuke to be >> practical. >> >> >> >> >> >> On Mon, 20 Mar 2017 11:09:30 -0400, jon parker wrote: >> >>> Thanks for the responses! I'm not a full-time comper so am unaware of >>> some of the tricks out there, like using ZDefocus. And I should have >>> added that large kernels / images are involved so the standard >>> convolve node falls behind at the resolution we are working with. >>> >>> Magno, I wasn't aware of your blinkscript. If I have some time I'll >>> peek at the code. >>> >>> Someone should put together a commercial implementation for Nuke some >>> day, sounds like there is demand out there for it. >>> >>> -Jon >>> On Mon, Mar 20, 2017 at 7:18 AM, Magno Borgo wrote: I coded a naive Blinkscript DCT (and inverse) implementation a while back, you can check the code and use as a start for a better implementation. http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform Magno. On Mon, 20 Mar 2017 03:39:40 -0400, Mads Lund wrote: we did some CPU vs FFT tests aswell, but I can't remember the exact kernel size where FFT started to be more efficient, but it was rather low. On the flip side Convolve on the GPU seem to have some memory problems and (at least for us) cause some random out of memory issues, even on beefy cards. We found that using gaming techniques: up-scaling low frequency areas (then convolving) and only full convolving of high frequency areas to be to be quite efficient for the vast majority of our renders where the plate is gigantic or a large kernel size is needed. The result is perceptually indistinguishable. > man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid : > > The convolve node was fixed up quite a while a go(4-5 years ago, it is > what the zdefocus is based on). In our tests though we still find the FFT > nodes faster on the farm vs convolve in CPU mode. If you have a gpu farm > then convolve is faster. > > > On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht > wrote: >> >> The latest version of the convolve node has gpu support. >> >> Am 20.03.2017 um 01:20 schrieb jon parker: >> > Greetings Nuke users, >> > >> > I'm just wondering if there are any faster, more robust FFT tools >> > available for Nuke besides the (hidden) built-in nodes? >> > >> > The built-ins do the job, but they are pretty slow and definitely >> > prone to crashing fairly often. >> > >> > Or, alternatively, something that does fast image convolution, some >> > other way, under the hood could work too. >> > >> > Cheers, >> > Jon >> > ___ >> > Nuke-users mailing list >> > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >> > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> > >> ___ >> Nuke-users mailing list >> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users -- Best regards. Mads Hagbarth Lund -- Using Opera's mail client: http://www.opera.com/mail/ ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>> ___ >>>
Re: [Nuke-users] GPU FFT?
Thanks Peter! Ron Ganbar email: ron...@gmail.com tel: +44 (0)7968 007 309 [UK] +972 (0)54 255 9765 [Israel] url: http://ronganbar.wordpress.com/ On Mon, Mar 20, 2017 at 6:23 PM, Peter McAuleywrote: > https://betterexplained.com/articles/an-interactive-guide- > to-the-fourier-transform/ > > On 20 March 2017 at 14:17, Ron Ganbar wrote: > >> He's a question: >> Can somebody properly explain FFT? >> (I'm feeling so inadequate right now. Am I the only one left who doesn't >> know what this magical acronym is?) >> >> >> >> >> >> Ron Ganbar >> email: ron...@gmail.com >> tel: +44 (0)7968 007 309 <+44%207968%20007309> [UK] >> +972 (0)54 255 9765 <+972%2054-255-9765> [Israel] >> url: http://ronganbar.wordpress.com/ >> >> On Mon, Mar 20, 2017 at 3:42 PM, Magno Borgo wrote: >> >>> A non-nuke alternative would be to pre-render the FFT using a python >>> package like scipy or similar, which already have fast FFT implementations, >>> then load/read it inside NUKE. That might help on some pipelines, though >>> the inverse edited FFT would still be needed to handled inside Nuke to be >>> practical. >>> >>> >>> >>> >>> >>> On Mon, 20 Mar 2017 11:09:30 -0400, jon parker >>> wrote: >>> >>> Thanks for the responses! I'm not a full-time comper so am unaware of some of the tricks out there, like using ZDefocus. And I should have added that large kernels / images are involved so the standard convolve node falls behind at the resolution we are working with. Magno, I wasn't aware of your blinkscript. If I have some time I'll peek at the code. Someone should put together a commercial implementation for Nuke some day, sounds like there is demand out there for it. -Jon On Mon, Mar 20, 2017 at 7:18 AM, Magno Borgo wrote: > I coded a naive Blinkscript DCT (and inverse) implementation a while > back, > you can check the code and use as a start for a better implementation. > > http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform > > Magno. > > > > > > On Mon, 20 Mar 2017 03:39:40 -0400, Mads Lund > wrote: > > we did some CPU vs FFT tests aswell, but I can't remember the exact > kernel > size where FFT started to be more efficient, but it was rather low. On > the > flip side Convolve on the GPU seem to have some memory problems and (at > least for us) cause some random out of memory issues, even on beefy > cards. > > We found that using gaming techniques: up-scaling low frequency areas > (then > convolving) and only full convolving of high frequency areas to be to > be > quite efficient for the vast majority of our renders where the plate is > gigantic or a large kernel size is needed. > The result is perceptually indistinguishable. > > > > > man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid >: > >> >> The convolve node was fixed up quite a while a go(4-5 years ago, it is >> what the zdefocus is based on). In our tests though we still find >> the FFT >> nodes faster on the farm vs convolve in CPU mode. If you have a gpu >> farm >> then convolve is faster. >> >> >> On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht >> wrote: >> >>> >>> The latest version of the convolve node has gpu support. >>> >>> Am 20.03.2017 um 01:20 schrieb jon parker: >>> > Greetings Nuke users, >>> > >>> > I'm just wondering if there are any faster, more robust FFT tools >>> > available for Nuke besides the (hidden) built-in nodes? >>> > >>> > The built-ins do the job, but they are pretty slow and definitely >>> > prone to crashing fairly often. >>> > >>> > Or, alternatively, something that does fast image convolution, some >>> > other way, under the hood could work too. >>> > >>> > Cheers, >>> > Jon >>> > ___ >>> > Nuke-users mailing list >>> > Nuke-users@support.thefoundry.co.uk, >>> http://forums.thefoundry.co.uk/ >>> > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuk >>> e-users >>> > >>> ___ >>> Nuke-users mailing list >>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>> >> >> ___ >> Nuke-users mailing list >> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> > > -- > Best regards. Mads Hagbarth Lund > > > > >
Re: [Nuke-users] GPU FFT?
https://betterexplained.com/articles/an-interactive-guide-to-the-fourier-transform/ On 20 March 2017 at 14:17, Ron Ganbarwrote: > He's a question: > Can somebody properly explain FFT? > (I'm feeling so inadequate right now. Am I the only one left who doesn't > know what this magical acronym is?) > > > > > > Ron Ganbar > email: ron...@gmail.com > tel: +44 (0)7968 007 309 <+44%207968%20007309> [UK] > +972 (0)54 255 9765 <+972%2054-255-9765> [Israel] > url: http://ronganbar.wordpress.com/ > > On Mon, Mar 20, 2017 at 3:42 PM, Magno Borgo wrote: > >> A non-nuke alternative would be to pre-render the FFT using a python >> package like scipy or similar, which already have fast FFT implementations, >> then load/read it inside NUKE. That might help on some pipelines, though >> the inverse edited FFT would still be needed to handled inside Nuke to be >> practical. >> >> >> >> >> >> On Mon, 20 Mar 2017 11:09:30 -0400, jon parker >> wrote: >> >> Thanks for the responses! I'm not a full-time comper so am unaware of >>> some of the tricks out there, like using ZDefocus. And I should have >>> added that large kernels / images are involved so the standard >>> convolve node falls behind at the resolution we are working with. >>> >>> Magno, I wasn't aware of your blinkscript. If I have some time I'll >>> peek at the code. >>> >>> Someone should put together a commercial implementation for Nuke some >>> day, sounds like there is demand out there for it. >>> >>> -Jon >>> >>> On Mon, Mar 20, 2017 at 7:18 AM, Magno Borgo wrote: >>> I coded a naive Blinkscript DCT (and inverse) implementation a while back, you can check the code and use as a start for a better implementation. http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform Magno. On Mon, 20 Mar 2017 03:39:40 -0400, Mads Lund wrote: we did some CPU vs FFT tests aswell, but I can't remember the exact kernel size where FFT started to be more efficient, but it was rather low. On the flip side Convolve on the GPU seem to have some memory problems and (at least for us) cause some random out of memory issues, even on beefy cards. We found that using gaming techniques: up-scaling low frequency areas (then convolving) and only full convolving of high frequency areas to be to be quite efficient for the vast majority of our renders where the plate is gigantic or a large kernel size is needed. The result is perceptually indistinguishable. man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid : > > The convolve node was fixed up quite a while a go(4-5 years ago, it is > what the zdefocus is based on). In our tests though we still find the > FFT > nodes faster on the farm vs convolve in CPU mode. If you have a gpu > farm > then convolve is faster. > > > On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht > wrote: > >> >> The latest version of the convolve node has gpu support. >> >> Am 20.03.2017 um 01:20 schrieb jon parker: >> > Greetings Nuke users, >> > >> > I'm just wondering if there are any faster, more robust FFT tools >> > available for Nuke besides the (hidden) built-in nodes? >> > >> > The built-ins do the job, but they are pretty slow and definitely >> > prone to crashing fairly often. >> > >> > Or, alternatively, something that does fast image convolution, some >> > other way, under the hood could work too. >> > >> > Cheers, >> > Jon >> > ___ >> > Nuke-users mailing list >> > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk >> / >> > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> > >> ___ >> Nuke-users mailing list >> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> > > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > -- Best regards. Mads Hagbarth Lund -- Using Opera's mail client: http://www.opera.com/mail/ ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>> ___ >>> Nuke-users mailing list >>>
Re: [Nuke-users] GPU FFT?
He's a question: Can somebody properly explain FFT? (I'm feeling so inadequate right now. Am I the only one left who doesn't know what this magical acronym is?) Ron Ganbar email: ron...@gmail.com tel: +44 (0)7968 007 309 [UK] +972 (0)54 255 9765 [Israel] url: http://ronganbar.wordpress.com/ On Mon, Mar 20, 2017 at 3:42 PM, Magno Borgowrote: > A non-nuke alternative would be to pre-render the FFT using a python > package like scipy or similar, which already have fast FFT implementations, > then load/read it inside NUKE. That might help on some pipelines, though > the inverse edited FFT would still be needed to handled inside Nuke to be > practical. > > > > > > On Mon, 20 Mar 2017 11:09:30 -0400, jon parker > wrote: > > Thanks for the responses! I'm not a full-time comper so am unaware of >> some of the tricks out there, like using ZDefocus. And I should have >> added that large kernels / images are involved so the standard >> convolve node falls behind at the resolution we are working with. >> >> Magno, I wasn't aware of your blinkscript. If I have some time I'll >> peek at the code. >> >> Someone should put together a commercial implementation for Nuke some >> day, sounds like there is demand out there for it. >> >> -Jon >> >> On Mon, Mar 20, 2017 at 7:18 AM, Magno Borgo wrote: >> >>> I coded a naive Blinkscript DCT (and inverse) implementation a while >>> back, >>> you can check the code and use as a start for a better implementation. >>> >>> http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform >>> >>> Magno. >>> >>> >>> >>> >>> >>> On Mon, 20 Mar 2017 03:39:40 -0400, Mads Lund >>> wrote: >>> >>> we did some CPU vs FFT tests aswell, but I can't remember the exact >>> kernel >>> size where FFT started to be more efficient, but it was rather low. On >>> the >>> flip side Convolve on the GPU seem to have some memory problems and (at >>> least for us) cause some random out of memory issues, even on beefy >>> cards. >>> >>> We found that using gaming techniques: up-scaling low frequency areas >>> (then >>> convolving) and only full convolving of high frequency areas to be to be >>> quite efficient for the vast majority of our renders where the plate is >>> gigantic or a large kernel size is needed. >>> The result is perceptually indistinguishable. >>> >>> >>> >>> >>> man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid : >>> The convolve node was fixed up quite a while a go(4-5 years ago, it is what the zdefocus is based on). In our tests though we still find the FFT nodes faster on the farm vs convolve in CPU mode. If you have a gpu farm then convolve is faster. On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht wrote: > > The latest version of the convolve node has gpu support. > > Am 20.03.2017 um 01:20 schrieb jon parker: > > Greetings Nuke users, > > > > I'm just wondering if there are any faster, more robust FFT tools > > available for Nuke besides the (hidden) built-in nodes? > > > > The built-ins do the job, but they are pretty slow and definitely > > prone to crashing fairly often. > > > > Or, alternatively, something that does fast image convolution, some > > other way, under the hood could work too. > > > > Cheers, > > Jon > > ___ > > Nuke-users mailing list > > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>> >>> -- >>> Best regards. Mads Hagbarth Lund >>> >>> >>> >>> >>> -- >>> Using Opera's mail client: http://www.opera.com/mail/ >>> >>> ___ >>> Nuke-users mailing list >>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>> >> ___ >> Nuke-users mailing list >> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> > > > -- > Using Opera's mail client: http://www.opera.com/mail/ > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >
Re: [Nuke-users] GPU FFT?
A non-nuke alternative would be to pre-render the FFT using a python package like scipy or similar, which already have fast FFT implementations, then load/read it inside NUKE. That might help on some pipelines, though the inverse edited FFT would still be needed to handled inside Nuke to be practical. On Mon, 20 Mar 2017 11:09:30 -0400, jon parkerwrote: Thanks for the responses! I'm not a full-time comper so am unaware of some of the tricks out there, like using ZDefocus. And I should have added that large kernels / images are involved so the standard convolve node falls behind at the resolution we are working with. Magno, I wasn't aware of your blinkscript. If I have some time I'll peek at the code. Someone should put together a commercial implementation for Nuke some day, sounds like there is demand out there for it. -Jon On Mon, Mar 20, 2017 at 7:18 AM, Magno Borgo wrote: I coded a naive Blinkscript DCT (and inverse) implementation a while back, you can check the code and use as a start for a better implementation. http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform Magno. On Mon, 20 Mar 2017 03:39:40 -0400, Mads Lund wrote: we did some CPU vs FFT tests aswell, but I can't remember the exact kernel size where FFT started to be more efficient, but it was rather low. On the flip side Convolve on the GPU seem to have some memory problems and (at least for us) cause some random out of memory issues, even on beefy cards. We found that using gaming techniques: up-scaling low frequency areas (then convolving) and only full convolving of high frequency areas to be to be quite efficient for the vast majority of our renders where the plate is gigantic or a large kernel size is needed. The result is perceptually indistinguishable. man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid : The convolve node was fixed up quite a while a go(4-5 years ago, it is what the zdefocus is based on). In our tests though we still find the FFT nodes faster on the farm vs convolve in CPU mode. If you have a gpu farm then convolve is faster. On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht wrote: The latest version of the convolve node has gpu support. Am 20.03.2017 um 01:20 schrieb jon parker: > Greetings Nuke users, > > I'm just wondering if there are any faster, more robust FFT tools > available for Nuke besides the (hidden) built-in nodes? > > The built-ins do the job, but they are pretty slow and definitely > prone to crashing fairly often. > > Or, alternatively, something that does fast image convolution, some > other way, under the hood could work too. > > Cheers, > Jon > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users -- Best regards. Mads Hagbarth Lund -- Using Opera's mail client: http://www.opera.com/mail/ ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users -- Using Opera's mail client: http://www.opera.com/mail/ ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
Re: [Nuke-users] GPU FFT?
Thanks for the responses! I'm not a full-time comper so am unaware of some of the tricks out there, like using ZDefocus. And I should have added that large kernels / images are involved so the standard convolve node falls behind at the resolution we are working with. Magno, I wasn't aware of your blinkscript. If I have some time I'll peek at the code. Someone should put together a commercial implementation for Nuke some day, sounds like there is demand out there for it. -Jon On Mon, Mar 20, 2017 at 7:18 AM, Magno Borgowrote: > I coded a naive Blinkscript DCT (and inverse) implementation a while back, > you can check the code and use as a start for a better implementation. > > http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform > > Magno. > > > > > > On Mon, 20 Mar 2017 03:39:40 -0400, Mads Lund wrote: > > we did some CPU vs FFT tests aswell, but I can't remember the exact kernel > size where FFT started to be more efficient, but it was rather low. On the > flip side Convolve on the GPU seem to have some memory problems and (at > least for us) cause some random out of memory issues, even on beefy cards. > > We found that using gaming techniques: up-scaling low frequency areas (then > convolving) and only full convolving of high frequency areas to be to be > quite efficient for the vast majority of our renders where the plate is > gigantic or a large kernel size is needed. > The result is perceptually indistinguishable. > > > > > man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid : >> >> The convolve node was fixed up quite a while a go(4-5 years ago, it is >> what the zdefocus is based on). In our tests though we still find the FFT >> nodes faster on the farm vs convolve in CPU mode. If you have a gpu farm >> then convolve is faster. >> >> >> On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht >> wrote: >>> >>> The latest version of the convolve node has gpu support. >>> >>> Am 20.03.2017 um 01:20 schrieb jon parker: >>> > Greetings Nuke users, >>> > >>> > I'm just wondering if there are any faster, more robust FFT tools >>> > available for Nuke besides the (hidden) built-in nodes? >>> > >>> > The built-ins do the job, but they are pretty slow and definitely >>> > prone to crashing fairly often. >>> > >>> > Or, alternatively, something that does fast image convolution, some >>> > other way, under the hood could work too. >>> > >>> > Cheers, >>> > Jon >>> > ___ >>> > Nuke-users mailing list >>> > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>> > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >>> > >>> ___ >>> Nuke-users mailing list >>> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> >> ___ >> Nuke-users mailing list >> Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > -- > Best regards. Mads Hagbarth Lund > > > > > -- > Using Opera's mail client: http://www.opera.com/mail/ > > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
Re: [Nuke-users] GPU FFT?
I coded a naive Blinkscript DCT (and inverse) implementation a while back, you can check the code and use as a start for a better implementation. http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform Magno. On Mon, 20 Mar 2017 03:39:40 -0400, Mads Lundwrote: we did some CPU vs FFT tests aswell, but I can't remember the exact kernel size where FFT started to be more efficient, but it was rather low. On the flip side Convolve >on the GPU seem to have some memory problems and (at least for us) cause some random out of memory issues, even on beefy cards. We found that using gaming techniques: up-scaling low frequency areas (then convolving) and only full convolving of high frequency areas to be to be quite efficient for >the vast majority of our renders where the plate is gigantic or a large kernel size is needed. The result is perceptually indistinguishable. man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid : The convolve node was fixed up quite a while a go(4-5 years ago, it is what the zdefocus is based on). In our tests though we still find the FFT nodes faster on the >>farm vs convolve in CPU mode. If you have a gpu farm then convolve is faster. On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht wrote: The latest version of the convolve node has gpu support. Am 20.03.2017 um 01:20 schrieb jon parker: Greetings Nuke users, I'm just wondering if there are any faster, more robust FFT tools available for Nuke besides the (hidden) built-in nodes? The built-ins do the job, but they are pretty slow and definitely prone to crashing fairly often. Or, alternatively, something that does fast image convolution, some other way, under the hood could work too. Cheers, Jon ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users --Best regards. Mads Hagbarth Lund -- Using Opera's mail client: http://www.opera.com/mail/___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
Re: [Nuke-users] GPU FFT?
we did some CPU vs FFT tests aswell, but I can't remember the exact kernel size where FFT started to be more efficient, but it was rather low. On the flip side Convolve on the GPU seem to have some memory problems and (at least for us) cause some random out of memory issues, even on beefy cards. We found that using gaming techniques: up-scaling low frequency areas (then convolving) and only full convolving of high frequency areas to be to be quite efficient for the vast majority of our renders where the plate is gigantic or a large kernel size is needed. The result is perceptually indistinguishable. man. 20. mar. 2017 kl. 06.14 skrev Deke Kincaid: > The convolve node was fixed up quite a while a go(4-5 years ago, it is > what the zdefocus is based on). In our tests though we still find the FFT > nodes faster on the farm vs convolve in CPU mode. If you have a gpu farm > then convolve is faster. > > > On Sun, Mar 19, 2017 at 6:23 PM Michael Habenicht > wrote: > > The latest version of the convolve node has gpu support. > > Am 20.03.2017 um 01:20 schrieb jon parker: > > Greetings Nuke users, > > > > I'm just wondering if there are any faster, more robust FFT tools > > available for Nuke besides the (hidden) built-in nodes? > > > > The built-ins do the job, but they are pretty slow and definitely > > prone to crashing fairly often. > > > > Or, alternatively, something that does fast image convolution, some > > other way, under the hood could work too. > > > > Cheers, > > Jon > > ___ > > Nuke-users mailing list > > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users -- Best regards. Mads Hagbarth Lund ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
Re: [Nuke-users] GPU FFT?
The convolve node was fixed up quite a while a go(4-5 years ago, it is what the zdefocus is based on). In our tests though we still find the FFT nodes faster on the farm vs convolve in CPU mode. If you have a gpu farm then convolve is faster. On Sun, Mar 19, 2017 at 6:23 PM Michael Habenichtwrote: > The latest version of the convolve node has gpu support. > > Am 20.03.2017 um 01:20 schrieb jon parker: > > Greetings Nuke users, > > > > I'm just wondering if there are any faster, more robust FFT tools > > available for Nuke besides the (hidden) built-in nodes? > > > > The built-ins do the job, but they are pretty slow and definitely > > prone to crashing fairly often. > > > > Or, alternatively, something that does fast image convolution, some > > other way, under the hood could work too. > > > > Cheers, > > Jon > > ___ > > Nuke-users mailing list > > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > > ___ > Nuke-users mailing list > Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
Re: [Nuke-users] GPU FFT?
The latest version of the convolve node has gpu support. Am 20.03.2017 um 01:20 schrieb jon parker: Greetings Nuke users, I'm just wondering if there are any faster, more robust FFT tools available for Nuke besides the (hidden) built-in nodes? The built-ins do the job, but they are pretty slow and definitely prone to crashing fairly often. Or, alternatively, something that does fast image convolution, some other way, under the hood could work too. Cheers, Jon ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
Re: [Nuke-users] GPU FFT?
Not quite FFT but there is a similar DCT implemented as a blinkscript on Nukepedia, http://www.nukepedia.com/blink/other/dct-discrete-cosine-transform There is also just the Convolve node (buggy and a bit slow at least on CPU), or you can (mis)use the ZDefocus node in "math: direct" to do convolution On 20/03/17 10:50, jon parker wrote: Greetings Nuke users, I'm just wondering if there are any faster, more robust FFT tools available for Nuke besides the (hidden) built-in nodes? The built-ins do the job, but they are pretty slow and definitely prone to crashing fairly often. Or, alternatively, something that does fast image convolution, some other way, under the hood could work too. Cheers, Jon ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users -- ben dickson 2D TD | ben.dick...@rsp.com.au rising sun pictures | www.rsp.com.au ___ Nuke-users mailing list Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users