[jitsi-dev] Options to process video feed


#1

Hi!

We are building a custom jitsi-meet client and have a requirement that we
process the user video to show different level of privacy effect using
semi-opaque overlay and blur. We are currently doing on receiving end but
obviously it can be removed by fiddling with css. We want to add the
privacy feature so that it can be removed by disabling the css. So what are
the options?

Currently HTMLCanvasElement.captureStream() is experimental and not
available on all browsers so seem to run out of options on client-side.

Processing on video-bridge end will remove the advantage in terms of
processing requirements because after having a brief look, video bridge
does not even decode and simply forwards the stream. Correct me if I am
wrong?

Secondly we want to reduce the bandwidth utilization if the privacy is
turned on. It seems there are some options in SDP (apart from using other
tricks like reducing video size). So the question is that whether it is
possible to re-negotiate SDP during the call?

Best regards
Babar.


#2

Hello,

Hi!

We are building a custom jitsi-meet client and have a requirement that
we process the user video to show different level of privacy effect
using semi-opaque overlay and blur. We are currently doing on receiving
end but obviously it can be removed by fiddling with css. We want to add
the privacy feature so that it can be removed by disabling the css. So
what are the options?

Currently HTMLCanvasElement.captureStream() is experimental and not
available on all browsers so seem to run out of options on client-side.

Processing on video-bridge end will remove the advantage in terms of
processing requirements because after having a brief look, video bridge
does not even decode and simply forwards the stream. Correct me if I am
wrong?

Correct. It would have to re-encode the stream which is expensive, somewhat complex, adds a delay and loses quality.

Secondly we want to reduce the bandwidth utilization if the privacy is
turned on. It seems there are some options in SDP (apart from using
other tricks like reducing video size). So the question is that whether
it is possible to re-negotiate SDP during the call?

Unless I'm missing something you should be able to do a round of setLocalDescription/setRemoteDescription and update the SDPs on the browser without actually notifying the remote side.

Regards,
Boris

ยทยทยท

On 14/07/15 18:41, Babar Asghar wrote: