1 | initial version |
Learned Lessons:
1.Dataset: Chessborad images, which were the ones I wanted to use, must be fully flat collected. I was using chessboard projected on an object. That is to be done afterwards, once calibration has been acomplished. So I changed the images for 'just' chessboard pattern on its own, with slightly degrees up&down, rigth&left. About 10.
After that, I got the calibration output I was looking for.