Hi, On December 1, 2014 11:34:44 PM GMT+01:00, Michael Niedermayer <michae...@gmx.at> wrote: >On Mon, Dec 01, 2014 at 11:41:41AM +0100, Benoit Fouet wrote: >> --- >> Tested against all the materials I have at hand. >> There is an artifact showing for >https://raw.githubusercontent.com/maxcom/lorsource/master/src/test/resources/images/i_want_to_be_a_hero__apng_animated__by_tamalesyatole-d5ht8eu.png >> which I don't really understand, as it seems the individual frames >are correct >> for our decoder, but the disposal that's done for other decoders >(tested >> firefox and chrome) is not the same for the end of the cape. >> --- >> libavcodec/pngdec.c | 93 >++++++++++++++++++++++++++++++++++++++++------------- >> 1 file changed, 71 insertions(+), 22 deletions(-) >> >> diff --git a/libavcodec/pngdec.c b/libavcodec/pngdec.c >> index 9e52d0b..2ca3dee 100644 >> --- a/libavcodec/pngdec.c >> +++ b/libavcodec/pngdec.c >> @@ -23,6 +23,7 @@ >> >> #include "libavutil/bprint.h" >> #include "libavutil/imgutils.h" >> +#include "libavutil/thread.h" >> #include "avcodec.h" >> #include "bytestream.h" >> #include "internal.h" >> @@ -38,9 +39,16 @@ typedef struct PNGDecContext { >> AVCodecContext *avctx; >> >> GetByteContext gb; >> + ThreadFrame previous_picture; >> ThreadFrame last_picture; >> ThreadFrame picture; >> >> +#if CONFIG_APNG_DECODER >> + AVMutex mutex; >> + int frame_id; >> + int *pframe_id; >> +#endif > >why do you need a mutex ? >
Actually, the only thing I need is the frame index. The best place for that would be in the demuxer, but I didn't find a place where this information is accessible. Did I miss something (I hope so)? Do you think I should be using side data for this? To answer the question though, the access is done is all the decoder threads, so I did not want the reset to happen between the reading and the writing of the ++. Thinking more about this, I think it's wrong anyway. I really need the demuxer to handle this, it would be simpler and more correct... -- Ben _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
