Hello
I have an issue regarding the way the denoiser are applied. In previous versions the denoiser calculated at the end of the render, but now it does it during the render progress, which increases render time.
Dont know, if this is a new feature, or a bug.
How to revert to previous behaviour?
Recorded the issue
https://mega.nz/file/vg9CSbKL#PLsUdmSH8du1VBi_EMMHlASQzTfr0RpyyMcdazfQTcs
Can i make a suggestion?
Keep both features and make it switchable with a checkbox or something like that. Or offer 2 versions, one with the new feature and the other one has the old denoiser behaviour.
In any case, wouldnt that be easier for you guys than optimizing? And especially more fair for users, who didnt want that new feature and now have to wait for the optimization...
Please make one of the above mentioned quick solutions possible.
@cera
Dont bother, they dont care.
They stubbornly ignore you, until you give up asking. Thats how Arnold development is...
I just hope alot users read this thread and realise how shabby these people are!
For me, im done with Arnold and will switch to a more superior renderer like Redshift!!!
I suggest you do that, too, because the dont give damn about this issue!
Opening a ticket, triaging and prioritizing it, scheduling it for development, and then actually working on it is not ignoring it.
We cannot publicly pre-announce when new features and changes will be available. That's the way it is with publicly traded companies.
You got couple times told to just give the option to use the old or new denoiser feature, which is done with a simple checkbox or different versions etc....
Simple solutions until you got that issue fixed, but you prefer to just ignore that!! And now people have to wait for that "optimization"...
No words, man
Even a simple toggle is a change in behavior that has to happen in the core and in the plugins. So, more than a simple "bug fix". Not necessarily complicated, but not a bug fix candidate.
Here is a temporary solution to solve this.
Thanks to Declan Russell for coding this and to Mads for showing me this a couple of months ago.
Go to https://github.com/DeclanRussell/IntelOIDenoiser/releases/tag/1.6
Download the Denoiser_v1.6.zip. Unzip it to your C:\ drive for example.
Then open the notepad or notepad+ and add these lines taken from Declan's script.
SET FILE_EXTENSION=exr SET PATH_TO_DENOISER=C:\Denoiser_v1.6\Denoiser_v1.6 SET OUTPUT_PREFIX=denoised_ for /r %%v in (*.%FILE_EXTENSION%) do %PATH_TO_DENOISER%\Denoiser.exe -i "%%~nv.%FILE_EXTENSION%" -o "%OUTPUT_PREFIX%%%~nv.%FILE_EXTENSION%" cmd /k
Change the exr to the extension you are using. Although you should be rendering out EXRs and not jpgs, pngs or tiffs.
Set path - is the folder where you unzip the Denoiser_v1.6zip file.
Save as denoiser.bat and save it inside the folder with the files you want to denoise.
Hi
Thank for the info, but I'm getting error
00:00:000 | Launching OIDN AI Denoiser command line app v1.6
00:00:000 | Created by Declan Russell (01/03/2019)
00:00:000 | Input image: arnold1.exr
00:00:004 | Loaded successfully
00:00:004 | Output image: denoised_arnold1.exr
00:00:178 ERROR | Failed to convert beauty to float3
any solutions for this?
Hi,
according to Declan as mentioned on his website:
Currently the only workflow is to render AOVs separately and then feed paths to each. I've also replied in DeclanRussell/NvidiaAIDenoiser#18 in regard to multilayer EXR support :)
So, are you trying to denoise all the AOVs in one go?
Hi
my bad, I didn't read the exception, and yes I was trying to denoise all AOVs in one go.
So I have to add a token <RenderPass> to get all the AVOs then use Nuke to shuffle them in one exr file.
work fine !
Thank you
With the new release, hoped this matter would be solved by now.
Couple month have passed, can you guys give an update?
Can't find what you're looking for? Ask the community or share your knowledge.