1 | initial version |
Yes I can confirm having the same problem using the iterative solvePnP approach on a set (about 30 on a flat marker) of coplanar points. The flip is always the same so not some random error produced by outliers.
Did you ever manage to fix it?