On Sun, Mar 29, 2015 at 03:09:53PM +0200, Anton Khirnov wrote: > Quoting Diego Biurrun (2015-03-29 14:48:58) > > On Sun, Mar 29, 2015 at 02:18:54PM +0200, Anton Khirnov wrote: > > > --- a/libavcodec/hevc.c > > > +++ b/libavcodec/hevc.c > > > @@ -383,24 +383,63 @@ static int decode_lt_rps(HEVCContext *s, > > > LongTermRPS *rps, GetBitContext *gb) > > > + avctx->pix_fmt = sps->pix_fmt; > > > + avctx->coded_width = sps->width; > > > + avctx->coded_height = sps->height; > > > + avctx->width = sps->output_width; > > > + avctx->height = sps->output_height; > > > + avctx->has_b_frames = sps->temporal_layer[sps->max_sub_layers > > > - 1].num_reorder_pics; > > > + > > > - s->avctx->coded_width = sps->width; > > > - s->avctx->coded_height = sps->height; > > > - s->avctx->width = sps->output_width; > > > - s->avctx->height = sps->output_height; > > > - s->avctx->has_b_frames = > > > sps->temporal_layer[sps->max_sub_layers - 1].num_reorder_pics; > > > > Is there a reason to suddenly set pix_fmt as well? It sure looks odd in > > a refactoring-only patch. > > Yes, there is. > Currently, that function sets pixel format taking into account hwaccels, > but that should not be done during init.
I still don't understand. It's not done during init AFAICT. You're doing it in the same place now. Why wasn't it needed before and why doesn't the log message explain it? Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
