Today I mostly watched this great tutorial on render layers then played around with it myself on a pre-existing scene.
In render layers, there is automatically a master layer which is setup to output all AOVs (arbitrary output variables… which is basically any data from a shader than can be output to a render for more flexibility in compositing). One can create custom layers (and collections of objects within that layer) to quite literally render multiple layers of a scene.
First one must go to render settings and move desired passes to active AOVs (this will allow them to show up in the render layers tab). Apparently there are some standard layers always included for ‘beauty’ passes in 3D, which are: diffuse (direct+indirect), emission, specular (direct+indirect), SSS (direct+indirect), transmission (direct+indirect), volume, the texture of the model, and background.
Then in render layers you add the objects + lights you want into a collection within a layer and when you render images you can now select different passes to view! I’m not sure how to save each individual one yet. I’m going to look into that tomorrow.
Apparently it’s typical to have a shadow pass and an AO pass. Which is done by right clicking on the collection and setting a material override (for shadows set aiShadowMatte and for AO set aiAmbientOcclusion). I found the shadow pass super useful for figuring out how many light samples to have in my scene. I actually got curious with this and made a little chart comparing light samples and render time. This is with my basic scene of two area lights and an HDRi (which it turns out can have light samples as well! How neat).
So at least for my scene setup it looks like light samples of 4-5 are really necessary.
I am still confused on:
How to export all the passes as separate images as a batch export (for the above pics I exported each one individually). Granted I still need to get to the exporting section of the tutorial but man am I excited to figure that out.
Additionally, how does one render out only the alpha channel? For the noise comparison (which was set to alpha) I just took a screenshot of each snapshot, but surely there’s a way to render it out?
All these questions shall hopefully be answered in tomorrows session. I do unfortunately have real school classes tomorrow so the time spent might be limited but I’ll try my darnedest. I’ll just focus on finishing the tutorial and doing one more render pass test (and hopefully not get distracted by testing light sampling hah).