Hi all,

We have an implementation of video.js that seems to regularly drop down from 1080p all the way to 480p multiple times per session for attendees. Those attendees are on relatively strong 100Mbps network, and it seems like video.js is really aggressive about this.

Does anyone have information on how it is determining to automatically adapt to a lower resolution ladder?

We don't want to force a fixed resolution in case of bad conditions, but we are seeing customers get hammered down with very little reason to 480p which makes text and some graphics illegible.

Is this an algorithm issue for detecting bandwidth? Should we adjust HLS chunk size?

We're sending HLS chunks from AWS CloudFront that we've encoded with ffmpeg on our own EC2 instance.
We currently have hls segment size set to 10 seconds.

Any tips appreciated. We're seeing this issue even after isolating out things like firewalls, vpns, etc.