You might want to look at this : https://textual-inversion.github.io/ https://textual-inversion.github.io/ It allows you to find a string that corresponds to

You might want to look at this : https://textual-inversion.github.io/ https://textual-inversion.github.io/ It allows you to find a string that corresponds to some "thing" you have pictures of. Then you can use a text to image model containing a reference to your "thing". Looks pretty amazing! However it needs 20gb of VRAM so I couldn't try it out.