[FFmpeg-devel] [PATCH 1/2] avcodec/libx264: add a flush callback
Andreas Rheinhardt
andreas.rheinhardt at outlook.com
Mon Mar 27 23:33:37 EEST 2023
James Almer:
> Signed-off-by: James Almer <jamrial at gmail.com>
> ---
> libavcodec/libx264.c | 24 ++++++++++++++++++++++--
> 1 file changed, 22 insertions(+), 2 deletions(-)
>
> diff --git a/libavcodec/libx264.c b/libavcodec/libx264.c
> index 92828fabc3..72c3b6cedc 100644
> --- a/libavcodec/libx264.c
> +++ b/libavcodec/libx264.c
> @@ -178,8 +178,8 @@ static int encode_nals(AVCodecContext *ctx, AVPacket *pkt,
> memcpy(p, x4->sei, x4->sei_size);
> p += x4->sei_size;
> size -= x4->sei_size;
> - x4->sei_size = 0;
> - av_freep(&x4->sei);
> + /* Keep the value around in case of flush */
> + x4->sei_size = -x4->sei_size;
> }
>
> /* x264 guarantees the payloads of the NALs
> @@ -648,6 +648,24 @@ FF_ENABLE_DEPRECATION_WARNINGS
> return 0;
> }
>
> +static void X264_flush(AVCodecContext *avctx)
> +{
> + X264Context *x4 = avctx->priv_data;
> + x264_nal_t *nal;
> + int nnal, ret;
> + x264_picture_t pic_out = {0};
> +
> + do {
> + ret = x264_encoder_encode(x4->enc, &nal, &nnal, NULL, &pic_out);
> + } while (ret > 0 && x264_encoder_delayed_frames(x4->enc));
> +
> + for (int i = 0; i < x4->nb_reordered_opaque; i++)
> + opaque_uninit(&x4->reordered_opaque[i]);
> +
> + if (x4->sei_size < 0)
> + x4->sei_size = -x4->sei_size;
> +}
> +
> static av_cold int X264_close(AVCodecContext *avctx)
> {
> X264Context *x4 = avctx->priv_data;
> @@ -1335,12 +1353,14 @@ FFCodec ff_libx264_encoder = {
> .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_DELAY |
> AV_CODEC_CAP_OTHER_THREADS |
> AV_CODEC_CAP_ENCODER_REORDERED_OPAQUE |
> + AV_CODEC_CAP_ENCODER_FLUSH |
> AV_CODEC_CAP_ENCODER_RECON_FRAME,
> .p.priv_class = &x264_class,
> .p.wrapper_name = "libx264",
> .priv_data_size = sizeof(X264Context),
> .init = X264_init,
> FF_CODEC_ENCODE_CB(X264_frame),
> + .flush = X264_flush,
> .close = X264_close,
> .defaults = x264_defaults,
> #if X264_BUILD < 153
Wasn't AV_CODEC_CAP_ENCODER_FLUSH supposed to be only implemented for
hardware encoders where closing and reopening an encoder is costly?
- Andreas
More information about the ffmpeg-devel
mailing list