Dear Linphone developers,
When using Linphone iOS app, with the iOS native H.264 encoder/decoder which uses VideoToolbox.framework, when call to some video phone model, it is one way video. The video phone cannot see the video, but Linphone iOS can see the video.
After hours of investigation, I found out that, when using VideoToolbox.framework to encode the video, it will call
rfc3984_set_mode(&ctx->packer_ctx, 1);
to tell the RTP packer to do fragmentation, i.e. packetization-mode=1, which results in FU-A packets. I compared the WireShark capture, I think those video phones which cannot see the video may not support that mode. So, I tried to do
rfc3984_set_mode(&ctx->packer_ctx, 0);
ctx->packer_ctx.maxsz = INT_MAX;
and hoping that IP layer can take care of the fragmentation if necessary. Then the video quality becomes very bad...
Is it possible to make the encoder to support packetization-mode=0 while still having the good performance when using VideoToolbox.framework?
Looking forward to your reply.
Best Regards,
Li