Ahh, I love this article. I'm among those that dislike all my photos with small objects that don't look as I see them. I guess building photo optics that do not

Ahh, I love this article. I'm among those that dislike all my photos with small objects that don't look as I see them. I guess building photo optics that do not use linear perspective is out of the question. I haven't got time to read all the papers they are referencing (currently searching for a new computer vision job, and there is so much in that area that I can't really read everything), but I wonder how many techniques to computationally change the perspective of the image depend on LIDAR depth information, how many on guessing depth via some other means (e.g. neural networks), and how many on neither?