On Fri, Oct 21, 2016 at 1:14 AM, Kagami Hiiragi <[email protected]> wrote: > On 21/10/16 06:14, James Zern wrote: >> On Thu, Oct 20, 2016 at 8:31 AM, Kagami Hiiragi <[email protected]> wrote: >>> vp9_cx_iface actually allows values in range [0..2]. >>> This fixes ticket #5894. >>> >>> Signed-off-by: Kagami Hiiragi <[email protected]> >>> --- >>> libavcodec/libvpxenc.c | 7 ++++++- >>> 1 file changed, 6 insertions(+), 1 deletion(-) >>> >> >> This is all right in the sense that the library exposes the option. In >> practice the feature was never fully developed and will cause failures >> in certain hardware decoders. > > vpxenc exposes it, good enough reason to allow that in ffmpeg too. > > >> The library will report an error (treated as a warning here) which is >> probably enough. > > Passing wrong option to encoder should be treated as error in my > opinion. Without that check library will just report warning and > encoding process won't be halted. >
That's the result for every other control we don't constrain. For now the local check is fine, but moving forward we should probably fail the control now that we don't try to set them without the user explicitly setting an option. _______________________________________________ ffmpeg-devel mailing list [email protected] http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
