This is similar to the solution I found for showing hands and face in a piano lesson. I spent a long time trying to get two cameras working. First by composing the video streams with ffmpeg or gstreamer and then by screen sharing a application which showed a second video. But in all cases the synchronisation was really bad.

Then I realised a mirror composes two images really well :)

Later I used transparent perspex with lighting (peppers ghost) as the mirror so the laptop screen on the piano is visible.

It is described here: https://www.youtube.com/watch?v=c4Qp8iHre9o&t=1s