1

I am rendering the depthbuffer of my scene to a texture, so that I can use it in a later pass for position and normal reconstruction. My question is whether the fact that the buffer is not power of 2 has any performance implications. And if that is so, is there a way to avoid this (excluding the case where I have to keep my canvas at a power of 2 resolution).

lightxbulb
  • 1,251
  • 12
  • 29
  • 1
    The only performance drawback is RAM as the frameBuffer will still need the closest power of 2 (Po2) above its size to be stored and used by the GPU. I believe that moving the frame buffer outside the GPU ie readPixels will be a little slower for non Po2 as the read is not a continuous memory transfer. You can avoid by making the frame buf closest smaller Po2 but you lose quality mapping it to the canvas res, or up size to closest Po2 but will need to render more pixels and mapping down to canvas res will create some artifacts. – Blindman67 Jun 11 '17 at 09:35

0 Answers0