There are several different technologies that are used to create these, mostly they are based on the same tools as the previous—Image—generation of AI generation (lol 2x). See my earlier post to get the hang of it.
So.... HOW does it work?
Essentially you get the neural network to make a bunch of images that are close together and then stitch them together into a GIF!
Also, you "might want" to use ControlNets to get the images to do what you want. Long story short, it's not to hard if you use
PS: There is probably a way to post this that does not use gifs (which slow down the website so much) but IDK how. If you know how then let me know! :)