Sign error for point to plane

[AdamSturge]

I think I've found a sign error in the equations for point to plane (and point to point if you don't use the closed form I guess). Specifically in the matrix A.

My point to plane wasn't working right, and on a hunch I tried inverting my rotation matrix R, which worked. So I decided to go though the details of the derivation and I think all the signs on the X columns in A are reversed. I changed them in my code and I didn't need to invert my rotation anymore.

That being said, the signs look correct in that they are equivalent to the base problem. Maybe the signs in the small rotation matrix are wrong?

[AdamSturge]

Or maybe our minimizer is actually maximizing the distance? And therefore the inverse rotation is (at least approximate to) the minimizer?

[AdamSturge]

Looking here is seems like the signs might indeed be flipped https://en.wikipedia.org/wiki/Rotation_matrix#Basic_rotations

for example with Rx you'd have -alpha on the last row and alpha in the second row, which is the opposite of what is in the ReadMe

[rin-23]

I had the same problem. I think there is a mistake in the notes about computing closest rotation.
Notes have R = U sigmaV.transpose(). If you look at the notes i linked before U,V should be reversed.