Hello,

Struggling a bit with producing high quality streams even on highspeed network.

In my case, we push high quality 1920x1080 frames at 25fps
(frames are AI generted, we can't increase FPS).

We encode frames to H264/H265 on Nvidia GPU, which has a lot more resources available.

The resulting video stream has very poor quality even on fast networks... (It's blurry, with big pixel squares appearing...)

For H265, upscaling frames to 3840x2160 before sending helps a little... But it's still blurry...

Note: it's very neat when using AV1 codec, but it has too little support on IOS for us to use at the moment. 🥲

What can we do about this ?

Ideally we would encode with constant/min bitrate, but we would settle to have a target bitrate that the encoder use by default, and allowing decreases if the bandwidth is low.

I see that we can also encode the video ourselves with the ingress API. But it would require us to have two participants, one to ingest audio/video and one to publish generated content to the room. And we would like to avoid this if possible.

Thanks for your help!