Relevant questions from the patch submission checklist: Explanation why it changes things like it does: Sometimes input data may have custom headers that ffmpeg is not expected to understand. It seems to me, this is why `-skip_initial_bytes` exists. From the documentation it sounds like this option should be supported on all muxers and demuxers but it doesn't seem to work with image2. Summary of user visible advantages: Users wanting to use the image2 format for files with custom headers can now do just that. Example using this change: `ffmpeg -y -skip_initial_bytes 32 -f image2 -c:v rawvideo -pix_fmt rgb24 -s:v 64x64 -ts_from_file 2 -pattern_type glob -i "*.RGB8" output.mp4` From within the example.zip I've attached. This zip file also includes `expected_output.mp4` for what the output should look like and `corrupt_output.mp4` for what I get without this patch. The patch contect itself: --- Add support for -skip_initial_bytes when using the image2 decoder. skip_initial_bytes is useful when the input data contains a custom header. Previously this was not supported when using the image2 format. Signed-off-by: Thomas Newton --- libavformat/img2dec.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/libavformat/img2dec.c b/libavformat/img2dec.c index 5f9d1f094f..19461aa200 100644 --- a/libavformat/img2dec.c +++ b/libavformat/img2dec.c @@ -474,6 +474,8 @@ int ff_img_read_packet(AVFormatContext *s1, AVPacket *pkt) ifmt = av_probe_input_format3(&pd, 1, &score); if (ifmt && ifmt->read_packet == ff_img_read_packet && ifmt->raw_codec_id) par->codec_id = ifmt->raw_codec_id; + } else { + avio_skip(f[i], s1->skip_initial_bytes); } if (par->codec_id == AV_CODEC_ID_RAWVIDEO && !par->width) -- 2.36.1 ---