On Sun, Aug 22, 2021 at 11:55 PM Jan Ekström <jee...@gmail.com> wrote: > > Fixes passing through mismatching metadata from the input side > when RGB input (from f.ex. H.264 or HEVC) gets converted to YCbCr. > > Fixes #9132 > --- > libavfilter/vf_scale.c | 9 +++++++++ > 1 file changed, 9 insertions(+) > > diff --git a/libavfilter/vf_scale.c b/libavfilter/vf_scale.c > index ce39217515..7ca833bbb1 100644 > --- a/libavfilter/vf_scale.c > +++ b/libavfilter/vf_scale.c > @@ -738,6 +738,15 @@ scale: > out->width = outlink->w; > out->height = outlink->h; > > + // Sanity check: If we've got the RGB/XYZ (identity) matrix configured, > and > + // the output is no longer RGB, unset the matrix. > + // In theory this should be in swscale itself as the > AVFrame > + // based API gets in, so that not every swscale API user > has > + // to go through duplicating such sanity checks. > + if (out->colorspace == AVCOL_SPC_RGB && > + !(av_pix_fmt_desc_get(out->format)->flags & AV_PIX_FMT_FLAG_RGB)) > + out->colorspace = AVCOL_SPC_UNSPECIFIED; > + > if (scale->output_is_pal) > avpriv_set_systematic_pal2((uint32_t*)out->data[1], outlink->format > == AV_PIX_FMT_PAL8 ? AV_PIX_FMT_BGR8 : outlink->format); > > -- > 2.31.1 >
Unless there are objections, I will pull this in soon. I will also attempt to see if I can add a programmatic FATE test for all of the allowed configurations of vf_scale, to see if there are any other clear-cut cases where not all metadata from the input AVFrame can be passed on as-is and has to be reset. Jan _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
