Hi
I've got cross eyed stereo setup which seems to produce sensible disparity and when I convert the disparity to 3D using repojectImageTo3D the 3D points look almost right but there seems to be some distortion. To investigate this I set disparity to zero and repojectImageTo3D produces a perfect flat plane when used with Q matrix from stereoRectify. Now this does not feel right, I would have expected that in a cross eyed case zero disparity would/should produce a warped plane?! Or am I mistaken?
If this is the case is there something in OpenCV that would produce the correct disparity to 3D projection or do I have to work this out by myself?
br Kusti