On 05/03/2013 11:14 AM, Anton Khirnov wrote: > From: Michael Niedermayer <[email protected]> > > The implementation is heavily based on Matthias Buercher's and Nicolas > Bertrand's vf_xyz2rgb > > Signed-off-by: Anton Khirnov <[email protected]> > --- > libswscale/swscale_internal.h | 9 +++++ > libswscale/swscale_unscaled.c | 82 > ++++++++++++++++++++++++++++++++++++++--- > libswscale/utils.c | 42 +++++++++++++++++++++ > 3 files changed, 128 insertions(+), 5 deletions(-) >
Summary of the patch: - it adds some way to signal that xyz is actually rgb48 (since that's the format it can be converted to) - it chains up the xyztorgb48 and then everything is managed as it is rgb48. The whole thing seems moderately horrible. What could be done: - set a format list to map the direct conversion available instead of having just a "can convert" boolean as we have now - make the converter setup function chain the conversion properly based on the above extension. - populate as needed. I'm not sure if I'll have time to do something about it soon. lu _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
