Rotation invariant representation for 2-D shapes about cubes HOT 3 OPEN

I found that there is a new repo for further developments and I moved this issue there.

from cubes.

Hi Spacha,
Thanks for letting me know about the new repo. I feel like your algo is interesting. If I wanted, I could use your algo to quickly check uniqueness before doing a longer check. Unfortunately, I don't think it's perfect. As the size of the grid increases, the number of solution grows something like 2^(n^2) while the number of combos in your algo seems to only grow at this kind of rate: (2^10)*n^2. For large n, 2^(n^2) will win, but I still like it!

from cubes.

Yes, this one was a miss unfortunately. As pointed out in the other repo, this has another issue which makes this useless. I've been trying to find a better way of encoding the relative positions but it gets difficult for n > 5.

I still feel like there is a way of representing all of the rotations by a single hash but it isn't easy.

from cubes.