An attempt to port AniFX to SD3.5M.
I find that it, just like SD3 and SD3.5L, is more suited towards artistry or a blend between realistic and anime (not 2.5D, example: cover image).
Use settings:
> USE TRIGGER WORDS AT THE BEGINNING OF YOUR PROMPT. Or you can choose to put it at the end but I never tested it that way. In theory it should also works.
> From my test, results seems better if you use full natural language a.k.a just yap in the prompt box since I trained it using full natural language caption.
> Strength ranges from 0.8 - 1, but I recommend 1 since it's moderately trained.
> Use Euler + normal, simple, or beta and/or DPM++2M + SGM Uniform. I never tested others, but if you're comfortable with certain samplers, you can definitely try with that.
> CFG ranges from 4 - 5.5, but so far 4 works well.
> You should go for 25 steps or higher if you could, unless you're using the Turbo checkpoint.
> You don't really need upscaler but you can use it if you want higher fidelity.
> Clip skip... I don't really know. I usually use 1, but 2 may work.
> Recommended for portrait images. I'm not talking about image orientation, but like, that portraits. I hope you understand this. But if you want to create something else then feel free to do it since I'm not really a good prompter and can't test it with good prompt.
Changelog is available in version info. If there's no changelog, that means that version is an initial release version or still under testing.
Any feedback and any type of roastery is appreciated! Hit me on the comments or Tensor discord server.
Thank you for using this model in your image generation.