Per-vertex versus Per-pixel

Granularity for shading computations



• Per-pixel computes shading at surface’s every pixel
  • More accurate, but more expensive
• Per-vertex computes shading at surface vertices
  • Then interpolates per-vertex results across pixels
  • Usually, more pixels than vertices so more efficient
  • Poor results when shading effect does not interpolate linearly
• Hybrid approaches such as “texgen”

Previous slide

Next slide

Back to first slide

View graphic version