Testing out an idea I had.
Chop up A, Chop up B. Use the parts of B that are like A to rebuild A using only parts from B. 3D convolution. In audio the technique is called Convolution reverb. In Machine learning the technique is used to teach the computer to recognize objects.
Here in my test it's used to glue a bunch of soggy corn flakes to the poor beleaguered Stanford terra cotta bunny.
More explanation and demo videos to come. Questions? Comments? Has someone already done this but better?