E2E encryption is meaningless unless there is a way to prove that it is E2E, e.g. by showing us the source code of the client side and allowing us to compile it ourselves, which Signal does.
It would be super interesting if there was a way to abstract out encryption on the camera itself, where the video call software gets an encrypted video stream and its only job is to convey that stream to the other side, which decrypts it.
The hard part is sending an encrypted stream that can be programatically degraded based on available bandwidth, and still be cryptographically secure.