Well taking my past example I could aswer that everything depends on your shot. Making layers and stuff just for one angle of camera can take more time then just dropping an ogl alpha object to hide what you don’t wanna see and forget about it.
It can be used in compositing, yes. For say you have a sphere and 2 cylinders comme out from behind of that sphere on a background plate. You want to treat the sphere and the cylinders seperately from temselves and the background in after effects.
Well one method witch is most used is to render the shot strait and then make a black and white mask of the render (spheres white and cylinders black.) So you can use the luma of this map to isolate the spheres and you have to process another one to isolate de cylinders. With the ogl alpha method you also do 3 renders:
1- Spheres shaded and all the rest in ogl alpha.
2- Cylinders shaded and all the rest in ogl alpha.
3- Background shaded and all the rest in ogl alpha.
So both ways you have 3 renders total. But the ogl alpha way you get the objects alone (witch makes it easyer for the compositers) and if an object goes infront of another it hides it but doesn’t apear in the alpha channel witch is super praticle cause you can put your layers in more combinations of layering then if you’d see a black object pass in front (as with the b/w matte method) that you have to hide with the corresponding render that you put on top.
Hope you followed me.