“Not ML” as in “not learning from data to apply in new situations” but rather they do “mathematical optimisation”. The data they optimise over is just the imag

“Not ML” as in “not learning from data to apply in new situations” but rather they do “mathematical optimisation”. The data they optimise over is just the images of the current camera trajectory (as far as I understand)