Anybody could tell me the best settings to train a Embedding for an anime character?
-How many images to use
-Settings: steps, learning rate, batch
-Textual inversion template (style, subjetc, etc)
Here I have seen excellent "embeddings" but when I try to do mine I get horrible images