Do not confuse this with pov_peephole, it is different.
Trained with a dataset of 150 varied images of poses, positions and scenarios.
Compatibility: Trained mostly with anime images, and only tested on anime generations.
Recommended weight: 0.8 - 1.0
Keyword: fisheye. (You can combine it with angles of view e.g. from above,from behind,from below,from side)