LuxCoreRender Forums

A very impressive feature

Really great feature!

Thanks, I'm pretty proud of it
It will get even more useful when I implement support for the mask channels that are anti-aliased: MATERIAL_ID_MASK, BY_MATERIAL_ID, OBJECT_ID_MASK, BY_OBJECT_ID.

Anti-aliased mask channels? Sounds very interesting. How's that gonna work?

Are there maybe future plans to support cryptomatte as well?

SteffenD wrote: Mon Feb 26, 2018 5:54 pm Anti-aliased mask channels? Sounds very interesting. How's that gonna work?

You can see an example in the old wiki: http://www.luxrender.net/wiki/Render_pa ... AL_ID_MASK
MATERIAL_ID_MASK is an image with float values in the range 0..1.
You specify a material ID and it creates a mask for it.
BY_MATERIAL_ID is just the mask applied to the RGB_IMAGEPIPELINE channel (the "combined" pass in Blender).

OIC, thanks!

So I have one further stupid question:
Does this mean you can select a single material id to be output as a mask channel or can you add let's say 20 of these all at once to be output in 20 separate AOVs / channels?

SteffenD wrote: Mon Feb 26, 2018 7:10 pm Does this mean you can select a single material id to be output as a mask channel or can you add let's say 20 of these all at once to be output in 20 separate AOVs / channels?

You can add multiple of these. I think there's no limit on the CPU, but the OpenCL engines the following limits apply:
Only 1 of each allowed: MATERIAL_ID_MASK, BY_MATERIAL_ID, OBJECT_ID_MASK, BY_OBJECT_ID
Only 8 allowed: RADIANCE_GROUP (note that one of them is always used: the default light group)
(RADIANCE_GROUP is a lightgroup channel)

edit: I corrected my post.

Cool! Looking forward to it then!

SteffenD wrote: Mon Feb 26, 2018 7:16 pm Cool! Looking forward to it then!

I had a look at the LuxBlend code and the limits on OpenCL are different, I corrected my post.

While reading through an old thread, I found this ancient bug report: http://www.luxrender.net/forum/viewtopi ... 80#p109906
The bug was also in BlendLuxCore (wrong calculation of the distance between camera and depth of field object).
I fixed it now.