Hello everyone,
I am using livav-ffmpeg (ubuntu 16-04 LTS liav-ffmpeg packages) to
forward RTP streams from a webRTC gateway to a streaming server. I have
to remux my incoming RTP streams (i.e my audio and my video stream)
using a RTSP muxer in order to contact my streaming server:
- Incoming RTP audio packets are encoded in OPUS, I transcode them into
AAC before sending them to the RTSP muxer, which works fine
- On the video side, I get h264 packets from the incoming RTP video
stream, that I just want to re-inject into the output RTSP muxer. I am
struggling to do this correctly.
On the video side, what I do basically is:
- I create an AVPacket from my incoming h264 payload I get from my
incoming RTP video stream
- I setup my packet.stream_index to make it point to my RTSP video stream
- then I call av_write_frame(rtsp_format_context, &packet)
Only the audio is sent (and received correcly by my streaming server),
on the video side, it enters rtp_write_packet function (I get "0: write
len=..." messages), and then nothing, packet is never sent. Once
rtp_write_packet is entered I don't understand why the packet is not
sent?? What is weird is that if I don't create an audio stream (so my
RTSP muxer contains only my video stream), my video is sent correctly.
So the problem happens when I have 2 streams in my RTSP muxer. Do you
have any explanation for this?
I know, anyway that what I am doing is not correct: As my RTSP video
stream is initialized with a H264 codec id, rtp_write_packet is
computing a new NAL for each new packet. But I already have a correct
NAL in my incoming RTP payload that I would not like to strip out.
Basically I would just like to forward my incoming RTP h264 payload with
no modification into my RTSP muxer, is there a simple way to do this?
Thanks a lot,
Emmanuel.
_______________________________________________
Libav-user mailing list
[email protected]
http://ffmpeg.org/mailman/listinfo/libav-user
