Hi there, I'm loving the results from the internal denoiser applied by Houdini with the (Output) Pixel filters, when rendering with Mantra Locally..

But I can't use it when rendering on the cloud with Gridmarkets. So my brother helped me download and use https://github.com/DeclanRussell/NvidiaAIDenoiser/releases [github.com] (Nvidia OptiX 6.0.0 & Denoiser v2.2) ..

He's denoised externally with much success in Blender. By rendering out 3 images. The image, Albedo and normal. But when it comes to Houdini, these are the settings which we used to try and get the same results.. The denoiser does a poor job of removing the noise. It does like 20% of what Houdini does internally with it. (See images attached)

If you can let us know which would be the best way of rendering the image passes to feed into the denoiser externally would be awesome for our cloud pipeline. Thanks in advance for any advice.

Any hints on what to experiment with? The albedo pass looks correct. Will see if the normal pass needs some work.

This is something everybody doing semi-complex work should be trying to figure out..
Rendering with less cpu power time and utilizing nvidia's fast post-processing.
Especially when it comes to rendering on the cloud!