Message ID | 20221127170351.11477-23-anton@khirnov.net |
---|---|
State | New |
Headers | show |
Series | [FFmpeg-devel,01/30] lavc/libx264: factor out setting up the input frame | expand |
Context | Check | Description |
---|---|---|
yinshiyou/make_loongarch64 | success | Make finished |
yinshiyou/make_fate_loongarch64 | success | Make fate finished |
andriy/make_x86 | success | Make finished |
andriy/make_fate_x86 | success | Make fate finished |
Anton Khirnov: > --- > libavcodec/ffv1enc.c | 16 +++++++++++++++- > 1 file changed, 15 insertions(+), 1 deletion(-) > > diff --git a/libavcodec/ffv1enc.c b/libavcodec/ffv1enc.c > index 0237ac48eb..553abb558f 100644 > --- a/libavcodec/ffv1enc.c > +++ b/libavcodec/ffv1enc.c > @@ -1233,7 +1233,20 @@ static int encode_frame(AVCodecContext *avctx, AVPacket *pkt, > pkt->size = buf_p - pkt->data; > pkt->pts = > pkt->dts = pict->pts; > + pkt->duration = pict->duration; > pkt->flags |= AV_PKT_FLAG_KEY * f->key_frame; > + > + if (avctx->flags & AV_CODEC_FLAG_COPY_OPAQUE) { > + pkt->opaque = pict->opaque; > + if (pict->opaque_ref) { > + pkt->opaque_ref = av_buffer_ref(pict->opaque_ref); > + if (!pkt->opaque_ref) > + return AVERROR(ENOMEM); > + } > + } > + > + avctx->reordered_opaque = pict->reordered_opaque; > + > *got_packet = 1; > > return 0; > @@ -1272,7 +1285,8 @@ const FFCodec ff_ffv1_encoder = { > .p.type = AVMEDIA_TYPE_VIDEO, > .p.id = AV_CODEC_ID_FFV1, > .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_DELAY | > - AV_CODEC_CAP_SLICE_THREADS, > + AV_CODEC_CAP_SLICE_THREADS | > + AV_CODEC_CAP_ENCODER_REORDERED_OPAQUE, > .priv_data_size = sizeof(FFV1Context), > .init = encode_init, > FF_CODEC_ENCODE_CB(encode_frame), FFV1's does not have frame reordering (no AV_CODEC_PROP_REORDER for the codec descriptor); the encoder just has the AV_CODEC_CAP_DELAY set to write stats in case this was the first pass of a two-pass encoding (this could actually be moved . Otherwise the encoder is a simple one-in-one-out encoder with no delay at all. Something similar is true for the FLAC encoder (it has the DELAY cap so that it can output a side-data only packet with updated extradata when flushing) and adxenc. Maybe we should have an internal capability to allow to set these properties generically for these codecs, too? - Andreas
diff --git a/libavcodec/ffv1enc.c b/libavcodec/ffv1enc.c index 0237ac48eb..553abb558f 100644 --- a/libavcodec/ffv1enc.c +++ b/libavcodec/ffv1enc.c @@ -1233,7 +1233,20 @@ static int encode_frame(AVCodecContext *avctx, AVPacket *pkt, pkt->size = buf_p - pkt->data; pkt->pts = pkt->dts = pict->pts; + pkt->duration = pict->duration; pkt->flags |= AV_PKT_FLAG_KEY * f->key_frame; + + if (avctx->flags & AV_CODEC_FLAG_COPY_OPAQUE) { + pkt->opaque = pict->opaque; + if (pict->opaque_ref) { + pkt->opaque_ref = av_buffer_ref(pict->opaque_ref); + if (!pkt->opaque_ref) + return AVERROR(ENOMEM); + } + } + + avctx->reordered_opaque = pict->reordered_opaque; + *got_packet = 1; return 0; @@ -1272,7 +1285,8 @@ const FFCodec ff_ffv1_encoder = { .p.type = AVMEDIA_TYPE_VIDEO, .p.id = AV_CODEC_ID_FFV1, .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_DELAY | - AV_CODEC_CAP_SLICE_THREADS, + AV_CODEC_CAP_SLICE_THREADS | + AV_CODEC_CAP_ENCODER_REORDERED_OPAQUE, .priv_data_size = sizeof(FFV1Context), .init = encode_init, FF_CODEC_ENCODE_CB(encode_frame),