That way, YUV to RGBA conversion becomes 3 array looks up and bias+scale
instead of the 9 we currently do. Also, since CPUs are 64-bit now, the math ops
should be the same latency as the existing 16-bit ones. Win all around!
Original issue reported on code.google.com by utopiafa...@gmail.com on 23 Apr 2013 at 2:42
Original issue reported on code.google.com by
utopiafa...@gmail.com
on 23 Apr 2013 at 2:42