Does anyone know what the performance implications are using a texture as depth/stencil buffer for an fbo? As i see it, i have 2 choices:
1) Use glFramebufferTexture2D as depth buffer, render everything then use the texture for post processing.
2) Use glFramebufferRenderbuffer as depth buffer, render everything, copy buffer to texture then use texture for post processing.
My question is, does using a texture as depth/stencil buffer impact rendering performance? Do i lose performance features like hierarchical z?
Which version would probably be faster, also for future hardware generations?