An attempt to create a realistic ANIMA model.
Version 2.0
Version 2.0 was primarily tuned for the "er_sde" sampler, steps 12-16.
Run: Euler/simple (Beta), CFG 1, Steps 8-16
Text encoder: https://huggingface.co/circlestone-labs/Anima/tree/main/split_files/text_encoders
VAE: https://huggingface.co/circlestone-labs/Anima/tree/main/split_files/vae
Description
FAQ
Comments (10)
I love your models! Please keep developing for Anima! ^^
Why turbo?
Because the turbo, in addition to acceleration, also equalizes the weights.
Damn, shame it's for CFG 1 when Negative Prompts are actually really useful on Anima. Even specifically designed and trained for it.
This is the best IRL tune I've seen so far. It maintains a lot of concept knowledge from the base compared to others I've tried.
If you don't mind sharing, what tool/method are you using to distill the cfg?
This is a good question, I don’t have any particular method, I just do a lot of experiments until I am satisfied with the final result. ))
This is a good question. He just grab other people's model, and merge them in ComfyUI.
and in order to hide the metadata saved by ComfyUI so nobody would know what model he used, he saved the merged model again using a wrapper workflow.
👍
@reakaakasky Well, what's the problem? If the end result is satisfactory, then what difference does it make how it was done?