Mastodon Feed
Boosted by kornel ("Kornel"):
lina@vt.social ("Asahi Lina (朝日リナ) // nullptr::live") wrote:
I'm going to argue that basically every YUV conversion matrix used in gamma space is wrong!
BT.709 says Y = 0.2126R + 0.7152G + 0.0722B, but those coefficients are designed for linear light. They relatively overemphasize green and underemphasize blue in gamma space.
I think 0.299R + 0.518G + 0.183B or so would work better. For 2.2 gamma (simplifying the EOTF for the example), those give the 0.2126... *relative* intensities for pure R,G,B after linear conversion.
Saturated colors are always going to be too dark if you do grayscale conversion in gamma space, but at least with those coefficients all of R,G,B are *equally* too dark, instead of R and B getting way darker than G.