multimattes, glass and "affect all channels"

[super gnu]

so ive got a building (surprise!) i want a solid multimatte of the glass, which requires "affect all channels" to be off.

however i also want multimattes of the materials behind the glass, which requires "affect all channels" to be on, and even then, the mattes are tinted.

any neat way to get both a solid glass matte and clean mattes for the geometry behind it?

id obviously like to avoid doing multiple renders.

[Svetlozar Draganov]

Use VrayObjectSelect Render Element in Raw mode for that purpose.
It can extract a solid alpha of the objects behind the glass. You can use the same RE for the glass also although it can be extracted via MultiMatte, ObjectID and etc Elements.

Here is a quick setup: http://ftp.chaosgroup.com/support/sc...nd_glass02.max

[super gnu]

great. just what i was looking for. many thanks.

[super gnu]

out of interest, im assuming if i have 8 or 9 of these in a scene with "consider for aa" turned on, it might slow my render down to the point that it would be much quicker to render the masks seperately?

[NicoC]

Use VrayObjectSelect Render Element in Raw mode for that purpose.
It can extract a solid alpha of the objects behind the glass. You can use the same RE for the glass also although it can be extracted via MultiMatte, ObjectID and etc Elements.

Here is a quick setup: http://ftp.chaosgroup.com/support/sc...nd_glass02.max

[ATTACH=CONFIG]22501[/ATTACH]
[ATTACH=CONFIG]22502[/ATTACH]
[ATTACH=CONFIG]22503[/ATTACH]
[ATTACH=CONFIG]22504[/ATTACH]
[ATTACH=CONFIG]22505[/ATTACH]

Nice ! I wasn't aware of that solution !! thanks a lot, Svetlozar

[super gnu]

an issue with this method, is that if you have glass on a building, and you can see through the building to the background, the alpha will be transparent in those areas.. i guess turning off "affect all channels" for the glass would fix this? would i still then be able to get the mattes of the objects behind the glass?

[Svetlozar Draganov]

out of interest, im assuming if i have 8 or 9 of these in a scene with "consider for aa" turned on, it might slow my render down to the point that it would be much quicker to render the masks seperately?

Not sure actually. Every additional Render Elements will slow down the rendering for sure but it is difficult to predict which approach (single or multiple passes) will be more effective.

an issue with this method, is that if you have glass on a building, and you can see through the building to the background, the alpha will be transparent in those areas.. i guess turning off "affect all channels" for the glass would fix this? would i still then be able to get the mattes of the objects behind the glass?

Would it be possible to send us an example scene of this issue please.
Let us also know the result you are looking for.

[super gnu]

well, i just knocked up a test scene, but it doesnt exhibit a transparent alpha on the glass objectselect element.. so ill have to try to figure out why my big scene is doing that. however im near a deadline and so far i dont need that renderelement in my psd..

[Svetlozar Draganov]

well, i just knocked up a test scene, but it doesnt exhibit a transparent alpha on the glass objectselect element.. so ill have to try to figure out why my big scene is doing that. however im near a deadline and so far i dont need that renderelement in my psd..

Let us know if you found what's going on.
You may also simplify the scene and sent it for investigation.

[theedge]

I was just looking at corona features at the moment and saw that it integrates a simple option to handle this. Look here : https://corona-renderer.com/features/workflow-tweaks/ and scroll down to "Not visible in masks" section. It's dead simple : at the material level you can set if it will be visible in masks or not.
Maybe it could be implemented the other way around : in the multimatte we could have an exclude list of materials that will be invisible in the current matte.

[LudvikKoutny]

I was just looking at corona features at the moment and saw that it integrates a simple option to handle this. Look here : https://corona-renderer.com/features/workflow-tweaks/ and scroll down to "Not visible in masks" section. It's dead simple : at the material level you can set if it will be visible in masks or not.
Maybe it could be implemented the other way around : in the multimatte we could have an exclude list of materials that will be invisible in the current matte.

That's actually a pretty bad solution and hopefully Corona will soon have the same thing Vray has. Currently, Vray handles transparency and refraction in render elements a lot more correctly than Corona, so there is no need to go one step back. Sure if you completely disable visibility to mask for say... glass windows, then it will work, but if you do the same for glass sphere, where some refraction actually happens, then using that mask for anything will result in a very wrong output.