Towards Better Chroma Subsampling - SMPTE Presentation Notes
Slide: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Solving the out of gamut problem

• If we start off with a perfectly legal image, it should be possible to reconstruct it without creating erroneous values. I call this in-range chroma reconstruction.
  
  
• Methods for doing this can be pictured as pouring water (chroma) into glasses of different heights (one for each pixel). The height of each glass represents how much chroma that pixel can hold before out-of-range values occur.

		First pass	Second pass
Typical	“Proportion” method - chroma is distributed proportional to the heights of the glasses.	“Spill” method - Any spilled chroma is distributed into non-full glasses.