Yeah I heard the same rumour which was why I thought lets just get something sketched out in form of .diff
But I dont think that’s happened yet, So once complete I will not be distributing builds as would break the license, Ill post the .diff blender.dev.org and people if want to try it will have to build their own versions. Best I can do until the license issue is solved
Also felt this would be relevant here. Here’s the optix denoiser running within Blender to clear up a photo I took. These are promising results. Essentially something like this (hopefully) could be used for separate passes (particularly troublesome noisy passes) previous renders, photos, etc. It actually denoised better than photoshop/lightroom as it preserved detail in his shirt. Granted this photo is a bit out of focus, (I realized later.)
ooooowwwww, things just got interesting. Were have to start looking into this, at the mo I also need optix_denoiser.51.dll, but maybe that’s now included in the optix.dll as before it was called optix.51.dll. Maybe the new nvoptix.dll is combined
cudnn.dll might be needed for training speed increase but not a must, just speeds up training CNN
Do you have contacts at nvidia to query about this?
I compared the entry points between the one in the optix package and the one with the driver, the nvoptix.dll seems almost identical to the optix.51.dll from the sdk.
however it doesn’t seem to have any of the entry points of optix_denoiser.51.dll so sadly it doesn’t seem like they merged the two. (also no denoiser dlls in the driver by the looks of it)
I have no contacts at nvidia, not even contacts i can’t share… sorry
Brecht surly has nvidia contacts, think we should get him to ask sme questions. It would be stupid on nvidia’s side to now add optix directly but not the denoiser. Maybe we can encourage them to do so
Fascinating stuff ! Exciting times ahead, less samples = faster render times = more complex renders can be done on a home PC. Just out of interest heres a reupload of the image but after AI supersampling? Basically i ran it through Topaz AI gigapixel and AI upscaled it 400% but then bicubic sharper downsampled it back to its original size… Also the Topaz was set to strong -denoise and blur. You won’t notice the difference unless you scrutinise both images.
Hi, apologies for the noob question here but I’ve been up and down the thread and can’t find any information on where the downloaded files need to be saved to. I’ve tried putting them on the root of my C:\ drive including the .py addon and I’ve also tried saving everything to a folder in my scripts folder, ether way blender spews a load of errors when I press the “Denoise” button, I’m certain it’s due to the fact I’m not putting Optix in the correct place and therefore Blender can’t find the .exe
This is the error I’m seeing:
Traceback (most recent call last):
File "C:\Users\James\AppData\Roaming\Blender Foundation\Blender\2.79\scripts\addons\blender_optix_denoiser.py", line 77, in execute
subprocess.call(args)
File "C:\Program Files\Blender Foundation\Blender\2.79\python\lib\subprocess.py", line 247, in call
with Popen(*popenargs, **kwargs) as p:
File "C:\Program Files\Blender Foundation\Blender\2.79\python\lib\subprocess.py", line 676, in __init__
restore_signals, start_new_session)
File "C:\Program Files\Blender Foundation\Blender\2.79\python\lib\subprocess.py", line 955, in _execute_child
startupinfo)
FileNotFoundError: [WinError 2] The system cannot find the file specified
location: <unknown location>:-1