What's the difference between an Object ID mask vs. separating objects as their own render layers?
In some ways, both serve a similar purpose. Namely the ability to make adjustments on a per-object-basis via the compositor. But while the Object ID pass merely generates factor info (black and white pixels) to essentially mask out each object in "image space", Render Layers are much more powerful outside of "image space". Render layers can retain their own passes, their own materials, lighting, etc.
Ah gotcha, make sense. Thanks Kent! In that case, it seems like Object ID's and alpha channel combining are more related. Is there a benefit to using one over the other as a factor input?