Reverse engineered prompt

I want a simple Raspberry Pi touch table demo built from this repo, with a desktop test mode so I can try it without the full hardware first.

Please make the main app open a big 1280 by 800 black window that acts like the projected table, then have a second small 320 by 240 sensor window that sends touch or object data to it over TCP using the JSON format in the README. When I click, drag, and release in the small window, I should see clear visual reactions in the big window. Keep the flow easy to run on one computer first, but structure it so it can also work later with two Raspberry Pi devices, one for sensing and one for display.

If possible, also wire up the camera based sensor mode using OpenCV, but the mouse driven client should be the main thing working end to end. Please make setup and run steps straightforward, fix anything outdated, and look up current docs online if you need to.

Want more depth? Deep Reverse