Render a seperate masks pass

[Art48]

Hi,

I was wondering about the follwowing:

I have rendered an image with the "normal" passes like reflection, specular, lighting etc. (File A)
Now I want to render a seperate file which renders only the multimatte masks for all the materials/objects. (File B)

Now the AA settings of File A are the "foundation", and the masks obviously have to fit perfectly.
Since I want to render File B as fast as possible I would disable basically everything (lights, shadows, maps etc) to get it to render quickly.
Now this messes with the AA sampler of course since it basically renders a black image with only the masks in their elements visible.

What I do now is set the DMC sampler to garbage values and set the min/max of the AA to the max value of my beauty render.
This makes a lot of unneccassary samples on unneeded areas.
Is there some way of keeping the speed of a pure mask rendering but rendering out the exact same AA as the beauty?

Please do not debate reasons for or against doing a seperate masks pass. I have reasons and I don't do it this way usually

[Svetlozar Draganov]

Currently it is not possible to set up V-Ray to fire the same AA rays on two different scenes.
In order to get the maximum similarity you have to use a Fixed Image Sampler with Subdivs. equal to the Max Subdivs. value of the Adaptive IS of the Beauty Pass render.
This like you said will generate unnecessary samples but it will also ensure all the needed ones are there.

Have you tried to render Beauty and MM render elements on a single pass, then to render MM pass alone with the same IS settings and compare both results from the MM render element?
According to my test the MM render element gives the same results when using the same IS settings even though the scene is simplified a lot which makes me think that the samples fired for the MM render element have no connection to the ones fired for the Beauty Pass.

Can you run the same test in your environment and let us know the result from it?

[vlado]

Maybe it can be cheated if you create an ExtraTex element with the rendered image as a Bitmap texture in screen space attached to it, with "Consider for AA" enabled.

Best regards,
Vlado

[Dave_Wortley]

Could the SampleRate Render Element be used as an input in theory?

[samuel_bubat]

We sometimes render seperate mask, too. Usually when we need to render very large images and have to save every bit of Memory.
I than use an override material. Gi turned off ,lights off, default lights on. Like this your scene is bright enough for a good AA to happen.
If default lights are not enough I just use some more or less random placed omnis without shadows.

[Art48]

Currently it is not possible to set up V-Ray to fire the same AA rays on two different scenes.
In order to get the maximum similarity you have to use a Fixed Image Sampler with Subdivs. equal to the Max Subdivs. value of the Adaptive IS of the Beauty Pass render.
This like you said will generate unnecessary samples but it will also ensure all the needed ones are there.

Have you tried to render Beauty and MM render elements on a single pass, then to render MM pass alone with the same IS settings and compare both results from the MM render element?
According to my test the MM render element gives the same results when using the same IS settings even though the scene is simplified a lot which makes me think that the samples fired for the MM render element have no connection to the ones fired for the Beauty Pass.

Can you run the same test in your environment and let us know the result from it?

Hi,

I've already tried to just delete ever other elements from the beauty render, then switching off reflections/refractions, lights, shadows etc. Same AA settings. Gave me horrible sampling since the image was black. Basically 1 AA sample everywhere.
That what you mean?

The reason for an extra MM pass was basically RAM on a huge image ^^

[vlado]

Could the SampleRate Render Element be used as an input in theory?

Yes, in theory

Best regards,
Vlado