Foveated Rendering Techniques in Modern Computer Graphics

Abstract

Foveated rendering coupled with eye-tracking has the potential to dramatically accelerate interactive 3D graphics with minimal loss of perceptual detail. I have developed a new foveated rendering technique: Kernel Foveated Rendering (KFR), which parameterizes foveated rendering by embedding polynomial kernel functions in log-polar mapping. This GPU-driven technique uses parameterized foveation that mimics the distribution of photoreceptors in the human retina. I present a two-pass kernel foveated rendering pipeline that maps well onto modern GPUs. In the first pass, I compute the kernel log-polar transformation and render to a reduced-resolution buffer. In the second pass, I have carried out the inverse-log-polar transformation with anti-aliasing to map the reduced-resolution rendering to the full-resolution screen. I carry out user studies to empirically identify the KFR parameters and observe a 2.8X-3.2X speedup in rendering on 4K displays. The eye-tracking-guided kernel foveated rendering can resolve the mutually conflicting goals of interactive rendering and perceptual realism.