Okay, maybe this is a dumb question, but here it goes. Why is it impossible incorporate MSAA into a deferred pipeline? Personally, I could think of one or two ways that this could actually be achieved, but every single piece of middleware I see claims it is impossible. Could someone enlighten me? Much obliged.
Hi, generally, the problem is that MSAA renders multiple samples per pixel and interpolate the results to get the pixel value. Deferred rendering pipelines render one sample of all geometry information per pixel to the G-Buffer. Now imagine that you are rendering an edge: The geometry (depth, normal etc) values of multiple samples would be interpolated into a single pixel.
The interpolation of the surface normals can result in ugly artifacts during the lighting stage.
However, according to wikipedia, the deferred lighting approach can actually incorporate MSAA. I don’t know much about the Unity internals, so I can’t say which approach Unity is actually using.