[FFmpeg-devel] [PATCH] lavc/libx264: support AV_CODEC_CAP_ENCODER_RECON_FRAME

[Anton Khirnov]

Quoting James Almer (2022-07-18 19:41:27)
> Why not ff_encode_alloc_frame() for this? Can 
> csp_to_pixfmt(pic_out.img.i_csp) be different than ctx->pix_fmt?

Yes, the reconstructed frame is in x264's internal format, which will
typically be some flavor of nv12, even if the input format is fully
planar.

-- 
Anton Khirnov