Brainime
Version 3 Update
I have recently migrated to ComfyUI and CivitAI currently doesn't extract the metadata so unfortunately, I can't provide the prompts for the pictures shared...I'm sorry
Will produce almost anything (as long as it's anime) without too much of a prompt, see Notes below.
This model prefers natural language than long comma-separated chunks. It will still work with it, just not as well!
Your feedback is essential to improve the future updates, please consider leaving a comment or a review 🙏
Recommended Settings:
Sampler
DDIM
DPM++ variants
Euler A
CFG Scale
3~7 (I leave mine around 5 usually)
Steps
With Hi-Res: 20~30
Without Hi-Res: >=50
Hi-Res highly recommended
Recipe
This is my first merge, I've sort of lost the recipe at this point but it's using mostly
Notes
Highly recommended to use EasyNegative
I'm still new to this so I'll add more details on here as I get feedback from the community.
Description
Brainime v2.0 + InPainting V2.0
Following the release of V1 I've been annoyed by weird or incoherent backgrounds and so-so contrast. Version 2 mostly addresses this thanks to the user Jemnite on Huggingface.
Brainme V2 uses GyoZanMixR, GyoZanMixX and TangYuanV2 weights finely adjusted using a mix of RING, WRAP and OUT.
Recommended settings are same as V1.
For V3 I'd like to get rid of the EasyNegative embedding requirement if possible....will be in touch!
Make sure to leave a review or a comment as your feedback is what drives this project forward 🙏
FAQ
Comments (10)
How is the fp32 model 2gb 🤔
The filenames are the same. I think the fp32 link is wrong.
It's the Inpainting model and it is indeed fp16.
Civit has a limitation on files you can upload for one "version" of a model. It wont let you put two Checkpoints with the same settings so I put the "Inpainting" model as fp32 to hack my way around this limitation but it could be confusing..
@braintacles ╰（‵□′）╯
@braintacles That is actually very bad idea, if i had just downloaded normally without checking i would have downloaded inpainting model as i have changed the setting to always give me fp32 version. ( ˘︹˘ )
@Potatovision I wasn't aware of that feature, I'll make sure to make the change!
Fixed
Strong background, difficult model for props (tobacco and pipe).
The model is strong against the background, so it is beautiful even if the aspect ratio is changed from a full-length portrait of a vertical character. It is twice as tasty.
We expect the model to continue to be strong in the background and to be able to handle small objects as well.
can't find the InPainting V2.0 ？
I initially added the InPainting a a second file on the main V2.0 version but it confused other users as well. My apologies, InPainting V2.0 is now it's own version 🙏