Back in 2011, DARPA held a contest to see if someone could create a program to reconstruct shredded documents. I was wondering if Mocha Pro could be used in this scenario, maybe with some Python scripts. Has anyone attempted anything similar to this?
Here’s a link with some information about the contest.
Not without serious augmentation with additional Python modules. This is a significant machine learning and stitching challenge.
While Mocha could probably help with the stitching, blending and rendering, the actual algorithms to put together 10000+ separate pieces from a pile of possibly thousands of different papers would be difficult without a LOT of pre-trained data.
Quite a few modern research papers now exist on this exact approach, but it’s not really a task suited to our VFX tools.