Hi, On Tue, Jan 10, 2012 at 4:07 PM, Ronald S. Bultje <[email protected]> wrote: > On Tue, Jan 10, 2012 at 2:24 PM, Ronald S. Bultje <[email protected]> wrote: >> Compare: >> ./avplay -vf format=rgb555be ~/Movies/fate-suite/kega/kgv1.avi >> ./avplay -vf format=rgb24 ~/Movies/fate-suite/kega/kgv1.avi >> ./avplay -vf format=rgb555be,format=rgb24 ~/Movies/fate-suite/kega/kgv1.avi >> >> Looks to me like rgb555be-to-rgb24 conversion is broken. > > Reason is that this is handled by rgbToRgbWrapper(), which only > handles native-endianness conversions but also is triggered for > non-native-endianness conversions.
Also, regardless of this bug, I don't see how this could affect the fate test itself. When I look at LE and BE output from the same system (which would be the same as comparing LE output on a BE or LE system), it is identical apart from byte order. Mike, are you perhaps using a git revision before Janne committed this? commit b4dc68803bd7fa73c2e37f55ae73b5fdb293c68d Author: Janne Grunau <[email protected]> Date: Fri Dec 16 23:04:30 2011 +0100 swscale: add endian conversion for RGB555 and RGB444 pixel formats Add a macro to shorten the if condition. The only other thing I can think of is the use of the float offset... Can you try removing the -ss option in the fate test and see if they now generate the same results? Ronald _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
