Denoiser

[vlado]

Do they only take the RGB image as input?

No; a bunch of render elements are used (lighting, reflections, shadows, normals, world position and others). I'm not sure how exactly they are used though.

Best regards,
Vlado

[3LP]

This is sick!
Can't wait to test this out!

Is it any near being implemented in the max nightlies?

Cheers
Stan

[super gnu]

This is sick!
Can't wait to test this out!

Is it any near being implemented in the max nightlies?

Cheers
Stan

i get the impression this is a very early discussion between the companies. its not owned or made by Chaos, doesnt officially support vray yet, is still in beta even for mental ray on Maya, and im guessing some fairly protracted negotiations would be required to see it included in vray.


TLDR: dont hold your breath.

[vlado]

Is it any near being implemented in the max nightlies?

Hehe the thing isn't even out yet and I can't commit to anything until I actually play with it myself. Also, I don't know if and under what conditions those guys would be interested in licensing it - f.e. they might prefer to keep it a separate tool, or it might be prohibitively expensive for us. But it's exciting stuff nevertheless...

Best regards,
Vlado

[RotemS]

Hehe the thing isn't even out yet and I can't commit to anything until I actually play with it myself. Also, I don't know if and under what conditions those guys would be interested in licensing it - f.e. they might prefer to keep it a separate tool, or it might be prohibitively expensive for us. But it's exciting stuff nevertheless...

Best regards,
Vlado

So in about 1-2 weeks, yes?

[3LP]

Hehe the thing isn't even out yet and I can't commit to anything until I actually play with it myself. Also, I don't know if and under what conditions those guys would be interested in licensing it - f.e. they might prefer to keep it a separate tool, or it might be prohibitively expensive for us. But it's exciting stuff nevertheless...

Best regards,
Vlado

Aie... buuuuut I know you'll make you cute eyes and it will work out.

Otherwise we will have to make ours :


Plus, we all know your second name is Santa 2.0!

In the mean time, still hoping

Stan

[AndresAhumada]

woooo I hope this implementation is also for 3d max.

[Alex_M]

Holy Smokes! What is this magic? Those example images look impeccable after denoising! The one with the hair is particularly impressive. I would normally expect the hair to become a blob of colors but this is some pretty serious stuff right there!

When can we have this? I think we can expect some significant render speed improvements if this is implemented? I think I already know what I want for Christmas.

[jstrob]

Impressive! But yes I want to see some animatons. Do you have a link?

[super gnu]

just the one:

https://vimeo.com/112342126

[snivlem]

Impressive tech.

[Laserschwert]

We've already made denoising part of our common workflow when rendering animations. I strongly recommend you guys taking a look at AviSynth, a freeware frameserving/scripting-tool with tons of free plugins, including several denoisers. I mainly use MC_Spuds for denoising, but by now there might be better options.

Of course, a solution integrated in VRay would be much more convenient!

[VISUALMAN_TOKYO]

sorry chaosgroup with this topic I created a monster

[innobright]

The altus is a screen space denoising application that uses rendered features to preserve detail so that there is noise removal but no loss of important details.

Our current feature input list is as follows.

There are two inputs needed per AOV formatted with b0 and b1 in the name. The b0 and b1 renders also need to be rendered with different sample seeds.

rgb = beauty render
visibility = shadow pass; In terms of what the filter is expecting it is what is visible to the lights.
normals = forward facing normals that preserve bump and displacement
position = WP
albedo = unshaded diffuse textures
caustics = This is an optional AOV used only if you have caustics in your scene.

Each input plays a role in how a feature is preserved and how the noise is filtered.

Our current software package is based on Robust Denoising using Feature and Color information by Rouselle. (RDFC)
The paper can be found here: http://cgg.unibe.ch/publications/201...or-information

The idea behind the tool is that it uses a non traditional approach for screen space noise removal. It uses the normals and world position aovs to understand the dimension of what is visible to camera. The other AOVs are used to preserve user created detail IE texture, bump, and displacement detail that could be interpretted as noise by a traditional screen space filter. This will prevent loss of such detail.

The current download that is available on our website is our beta cycle version, we will be releasing our full production version in the next few days that has adjustments and fixes from our beta cycle.

We currently have a maya plug-in that will be packaged with the VRay version of our software that will generate render files that have the predetermined AOVs setup for you.

[Donfarese]

I know this is for Maya thread, but this is important for all vray plugins. will there be a 3ds Max version?