Yi Jiang
Yi Jiang
I can't think of any situations when people want the rotation center to be somewhere else other than the center of the matrix. I'd even automatically reset the rotation center...
Another paper on this topic: https://doi.org/10.1364/OL.389492
Perpahs all we need is to add a resizing step in /+GPU/private/gradient_position_solver.m
The Adorym paper (https://arxiv.org/abs/2012.12686) also mentioned that having regularization such as L1 or TV might be useful for multi-slice ptycho. It's worth testing them in the future.
The double DIP code is public now: https://github.com/mdw771/ddip4ptycho