Testing out the ability of Anima to generate realistic photos, using a very large data set, but more precise training variables and labels.
The only prompt needed is 'pantyhose', but it should be capable of a lot!
pantyhoseRealism_v1.safetensors