Concept video here

The key enabling components are the capture mechanism, print mechanism, and tracking mechanism.

The device can use commodity optical flow sensors from computer mice to track its orientation and position on the surface, whether in capture (copy) or print (paste) mode. There's a nice breakout board by Jkicklighter on Tindie for the ADNS-9800 sensor from a laser mouse.

For capture, I hope to use the same optical flow sensors; they are essentially cameras, and some of them can output the raw image pixels along with the tracking information. If not the ADNS-9800 specifically, perhaps some other part.

The print mechanism is the most difficult part. Ideally, we want something high-resolution and full-color that is well-documented and drive-able from a microcontroller. It seems like modern printheads are proprietary (not documented). Nicolas Lewis has created the InkShield, which uses a relatively old black inkjet printhead and cartridge, and should suffice for the proof-of-concept.