Does the belinograph need syncing though? I imagine you would just have a waste ribbon around the photo and start it after the call with say a delay you both agreed too. Or maybe you could send light in some known pattern with a starter strip and the receiver could look at the lamp.

Attenuation most have been a problem though? I guess you could send some sort of calibration photo to adjust gain? I don't think practical FM signal modulation was a thing yet in the 30s?

You need to synchronize the spin rate of the drums on either end. If there's a difference of half a rotation over however many hundreds (thousands?) of rotations to transmit the photo, you'd end up with a completely distorted picture.

I had a pair of these transceivers when I was in my teens, from a somewhat later period, thanks to my grandfather who refurbished them from his time as a telegrapher at Western Union. Synchronization at startup took ten seconds or so while each unit adjusted its timing and apparently sent a correction signal to the other, drums whirling all the time and gradually matching initial positions. It looked like the most primitive process possible, but worked every time.

Yes, I was looking for the bit where they addressed skew. You can even see it in the demo with the string, horizontal lines come out wavy after transfer to the other spool due to small differences in synchronization.