This not only reduces the amount of code, it also eliminates a redundant conversion: compute_frame_ssimulacra2 currently converts back to LRGB from the XYB that is passed to it.
This makes the code a bit faster, ~10% from simple testing.
Thanks, this must have been left in here from when the algorithm got updated--it used to require Xyb only, then it changed upstream to require Linear RGB. :+1:
This not only reduces the amount of code, it also eliminates a redundant conversion:
compute_frame_ssimulacra2
currently converts back to LRGB from the XYB that is passed to it.This makes the code a bit faster, ~10% from simple testing.