The math actually isn’t bad. You don’t have to run 640x480 at 24 bit - you can run 160x160 with black and white for example - that’s like a 3.5k frame buffer. What more do you need to show a few status boxes or maybe a remote oscilloscope display?
The compression schemes for VNC is very simple but effective… Once the initial frame is transmitted, only changes are sent after that - down to 4x4 pixel tiles if you want it to. Even if you don’t output to VNC, RFP is still and interesting protocol to implement… Instead of outputting every frame over SPI, output just the differences. Less CPU and less power too.
Ethernet will be ideal, but I think this can be done easily down to Bluetooth level, which can do over 1Mb/s. It won’t play full motion video, but it sure will work for simple graphical display updates.
What I’m after here is to use the power of the Presentation classes, which are there anyway, but with a screen a little distance away (maybe a phone). I have this image of a phone “looking” at the embedded device and seeing into it’s display. A window into the soul of a device
I guess it should be possible to develop a “real” display driver and capture and transmit graphic primitives, but that’s beyond my current abilities.