let me know what you think guys.
For games u need consistency between images, which is hard to do with current AI tech. For example, after making a base image for the girl, next u would want to put some clothes on her, but she should still be the same girl. After that u would want to add facial expressions too and here comes another problem: exponential growth of images. For example u have 10 outfits for a girl and then want to add 5 facial expressions: it's 10 x 5 = 50 pics already. To solve this a sprite system is typically used (u add expressions on top of the original image, so instead of 10 x 5 u get 10 + 5 = 15 images).
But this means that first 10 images with different outfits should have
absolutely the same girl in it, pixel to pixel. And then this should be done to 600 girls that are already in the game (+ there will be other variations like cuter looks after the surgery, bugger breasts, fatter body, piercings, scars, tattoos, etc), meaning u can't manually edit all those pics in photoshop. Good luck coming with the solution to this.
Edit: On top of that there are sex scenes: it's super hard to make those with AI, but this game needs those for each girl too. Even trying to generate anal sex is a pain bc every AI model was trained on danbooru-type tagged images that simply don't have anal sex tag.