Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
|
|
Thread Tools | Search this Thread | Display Modes |
15th October 2019, 20:00 | #1 | Link |
Registered User
Join Date: Jul 2019
Posts: 73
|
VSGAN - VapourSynth GAN Implementation, based on ESRGAN's implementation
https://github.com/rlaphoenix/VSGAN
docs: https://vsgan.phoeniix.dev VSGAN as of 1.6.0 is a port of multiple kinds of machine learning GAN Networks and Architectures. As of 1.6.0, ESRGAN (and its derivatives), and EGVSR is supported. For much more information, getting started, more details on supported models, networks, architectures, check the docs. Last edited by PRAGMA; 24th January 2022 at 09:03. |
15th October 2019, 23:45 | #2 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,377
|
Thanks, it works well
Any lists for nice pre-trained models ? I found a bunch on Google , but does anyone have any "favorites" that work well in specific situations? (including Pragma's own AD model) https://upscale.wiki/wiki/Model_Database Last edited by poisondeathray; 15th October 2019 at 23:51. |
16th October 2019, 06:28 | #3 | Link | |
Registered User
Join Date: Jul 2019
Posts: 73
|
Quote:
https://github.com/WolframRhodium/Super-Resolution-Zoo |
|
16th October 2019, 06:34 | #4 | Link | |
Registered User
Join Date: Sep 2007
Posts: 5,377
|
Quote:
But these are converted models (params , json ; not pth), are they drop in compatible ? (And some of the links to the original projects and papers have pth models) Last edited by poisondeathray; 16th October 2019 at 06:42. |
|
16th October 2019, 09:30 | #5 | Link | |
Registered User
Join Date: Dec 2005
Location: Germany
Posts: 1,795
|
Quote:
The "spongebob model" looked also not bad for some older animes but had some blurry spots for some reason.
__________________
AVSRepoGUI // VSRepoGUI - Package Manager for AviSynth // VapourSynth VapourSynth Portable FATPACK || VapourSynth Database |
|
16th October 2019, 09:40 | #6 | Link |
Registered User
Join Date: Dec 2005
Location: Germany
Posts: 1,795
|
I get out of memory errors for 1080p content using a GTX 1070. Does it really need what much ram? used model 1x_DeSharpen
CUDA out of memory. Tried to allocate 1.48 GiB (GPU 0; 8.00 GiB total capacity; 2.56 GiB already allocated; 72.05 MiB free; 3.65 GiB cached) Restarting vsedit didn't help.
__________________
AVSRepoGUI // VSRepoGUI - Package Manager for AviSynth // VapourSynth VapourSynth Portable FATPACK || VapourSynth Database |
16th October 2019, 15:04 | #7 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,377
|
I think so...I can only use 1x_Desharpen on smaller dimensions . You can look at GPU caps viewer or similar utilities and it looks like it's using all
I wonder if there is a way to share system memory with CUDA memory ? eg. Although it's slower, some 3D/CG renderers can offload graphics card memory to system memory when doing calculations (like a shared pool) enabling you to complete if the scene is too large |
16th October 2019, 21:36 | #8 | Link | |
Registered User
Join Date: Sep 2016
Posts: 67
|
Quote:
PyTorch does have a "empty cache" function. Maybe you could call it every few frames with FrameEval. https://pytorch.org/docs/stable/cuda...ory-management |
|
17th October 2019, 15:08 | #9 | Link | ||
Registered User
Join Date: Sep 2007
Posts: 5,377
|
Quote:
Quote:
Last edited by poisondeathray; 17th October 2019 at 15:14. |
||
20th October 2019, 13:03 | #10 | Link | ||
Registered User
Join Date: Jul 2019
Posts: 73
|
Quote:
Quote:
But yes, generally speaking x1080 resolution content require quite a bit of memory, i'm trying to work out a way to split up the input frame into multiple split images, to then run them all separately and then combine them back to 1 frame and return, that way using the same amount of memory, but less memory simultaneously, would be slower, but would be the only viable option for most GPU's. I'm having a bit of trouble working it out right now but hopefully I or someone else, figures out a way to do so. |
||
20th October 2019, 13:52 | #11 | Link | |
Registered User
Join Date: Jan 2016
Posts: 162
|
Quote:
|
|
20th October 2019, 14:54 | #12 | Link | |
Registered User
Join Date: Jul 2019
Posts: 73
|
Quote:
I didn't end up using anything like EDSR though, instead I simply cropped the frame, and then ran them. |
|
20th October 2019, 16:33 | #13 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,377
|
Thanks for the update
Request: If you have time would it be possible to port EDVR ? Looks great and deals with video specific issues, temporal and motion blur https://xinntao.github.io/projects/EDVR https://github.com/xinntao/EDVR |
22nd October 2019, 22:35 | #15 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,377
|
Request#2 , another pytorch implementation, PPON
https://github.com/Zheng222/PPON Someone forked it and wrote some demo code , would it be possible to implement in vapoursynth ? https://github.com/vlad3996/PPON/blo...rence_utils.py https://github.com/vlad3996/PPON/blob/master/demo.ipynb Thanks |
23rd October 2019, 05:00 | #17 | Link | |
Registered User
Join Date: Apr 2009
Posts: 478
|
Quote:
As an aside, I wonder what speeds are you folks getting? With a GTX 1070, upscaling 480p content using the "PSNR_x4_DB.pth" model, I'm getting like... 1fps :/ |
|
24th October 2019, 00:34 | #18 | Link | |
Registered User
Join Date: Jul 2019
Posts: 73
|
Quote:
RTX 2080ti takes about 1.4s per frame for the same input. |
|
Tags |
esrgan, gan, upscale, vapoursynth |
|
|