I’m doing tests for a high (-ish) end graphics game and I usually end up using quite a few post fx (especially with Unity 5).
So in general, I have tonemapping, color adjustments, some bloom/glow and AA (and occasionally DOF). And in general they end up taking a significant amount of ms.
And I started thinking there is a lot of wasted processing and VRAM. Each post effect grabs a framebuffer (and/or depthbuffer) does stuff to it and then writes back to it.
Wouldn’t it make sense to combine them into one big post processing effect? So that the grabbing and writing happens only once? Isn’t there a lot to be gained by doing that, or am I overthinking it?
I searched the forums and answers quite a bit, but I couldn’t find a lot of information on the matter.
I’m going to try it anyway, and I’ll make sure to post the results here, but the problem I am foreseeing, is that every time I want to add an effect (say I get something from the asset store), I have to edit it in the big frankenstein monster of post effect I will have created. Which would make it harder to keep the effects updated as well. It sort of becomes unmaintainable…
(as a sidenote, couldn’t Unity handle that on its own? Make sure the appropriate buffers are grabbed, then all the post fx use the same data, and the result is written once at the end?)