Chapter 20: Framebuffers
Framebuffers are used when we want to draw somewhere other than the screen. If you bind a framebuffer, then any subsequent draw call will draw onto the framebuffer.
Typically, we attach a color texture and a depth texture to a framebuffer. Thus the color of every fragment rendered will end up in the attached texture, and the depth values will end up in the attached depth texture.
If we don’t care to access the values of a framebuffer attachment, we can attach a renderbuffer instead of a texture.
The most common use of this is for depth - we very commonly want a depth buffer, but typically don’t need to actually read these values back.
You can’t read and write to the same buffer in a single draw call. So if we want to do any sort of multi-pass rendering (e.g. a blur), we need to create two framebuffers and “ping pong” in-between them.
This artucle describes an optimized technique for computing Gaussian blur but the first half describes the simple version used in our Lab 8 basecode.
This tool can be used to calculate gaussian kernel coefficients!