I’m working with some computer vision code on GitHub, and I’m struggling to understand just how the angles from RQDecomp3x3 map to pitch yaw and roll.
The code does some machine learning before eventually passing the data to solvePnP, then Rodrigues and finally RQDecomp3x3.
From enpirical testing, tilting the object up causes the x component to be more negative, yawing the object to the right causes the y to be negative, and rolling the object to the right causes the z component to be negative.
I’m trying to pass these angles into a graphics engine, and I have no idea what these angles correspond to, nor how to map them.