https://us.kandaovr.com/products/qoocam-ego
I did some experiments with manually converting anime images to stereograms by adding a depth map but I had trouble because there are unseen pixels that you have to make something up for. I know a grad student who has been making 3-d models based on images from the Pathfinder rover and he is struggling with this problem. I'd guess an A.I. model could do a good job here because you only need to fill maybe 5-10% of pixels in the scene.
I wrote some software which works like Disney's multiplane camera
https://en.wikipedia.org/wiki/Multiplane_camera
which can take a bunch of images and then compose them to make a stereo image, I badly want to take some of the characters from Pokemon Sun and Moon and make 3-d pictures of them in their environments. Back before I had the MQ3 I was making red-cyan anaglyphs that did not work reliably, I ought to go back and give it another crack.