On Fri, May 30, 2014 at 02:38:05PM +0200, Dirk Ausserhaus wrote: > On Fri, May 30, 2014 at 7:38 AM, Anton Khirnov <[email protected]> wrote: > > > > > > On Thu, 29 May 2014 13:44:34 +0200, Dirk Ausserhaus <[email protected]> > > wrote: > > > diff --git a/libavcodec/ivi_common.c b/libavcodec/ivi_common.c > > > index cf4df18..ef4df10 100644 > > > --- a/libavcodec/ivi_common.c > > > +++ b/libavcodec/ivi_common.c > > > @@ -968,6 +968,21 @@ int ff_ivi_decode_frame(AVCodecContext *avctx, void > > > *data, int *got_frame, > > > if (ctx->gop_invalid) > > > return AVERROR_INVALIDDATA; > > > > > > + if (avctx->codec_id == AV_CODEC_ID_INDEO4 && > > > + ctx->frame_type == IVI4_FRAMETYPE_NULL_LAST) { > > > + if (ctx->got_p_frame) { > > > + if ((result = av_frame_ref(data, ctx->p_frame)) < 0) > > > + return result; > > > + av_frame_free(&ctx->p_frame); > > > + ctx->p_frame = av_frame_alloc(); > > > > The above 4 lines should be replaced with av_frame_move_ref(data, > > ctx->p_frame) > > > > The rest looks fine to me. > > > Here're the updated patches in case you've not committed them yet. > > P.S. I might look at I263 too. What are the samples with IP-frames there?
Thank you very much. I've misremembered - I263 has PB-frames which are essentially the same things. Files like https://samples.libav.org/A-codecs/IMC/mensaje.avi https://samples.libav.org/A-codecs/IMC/neal73_saber.avi or https://samples.libav.org/A-codecs/IMC/nyc.avi contain them. But decoder is relying on dreaded MpegEncContext and it might be somewhat tricky to find where to hack it. _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
