Sennke/RealCosplayAnima_All-IN-ONE_FullFinetuned

RealCosplay&Anime&3D

A full fine-tuned Anima model designed to bridge Anime, Cosplay, Realistic Photography, and 3DCG styles in a single checkpoint.

While preserving Anima's character knowledge and artist-style tag understanding, this model introduces a large amount of realistic visual knowledge to improve generalization

More Examples

Civitai: https://civitai.red/models/2671085/realcosplayandanimeand3d-all-in-one-animabase

Recommended Prompts

Positive:

masterpiece, best quality, newest, high resolution

For stronger realism:

photo (medium)

Recommended weight:

photo (medium): 0.5 ~ 3.0

Negative:

worst quality, low quality, lowres, signature, text, logo

Suggested Parameters

Setting	Recommendation
Sampler	ER SDE / Euler SDE / Heun / DPM++ 2M SDE
Scheduler	Beta / Simple
CFG	3 ~ 5
Realistic CFG	4.5
Shift	3

Artist Styles

Supports most Danbooru artist tags inherited from Anima.

artist:@artist_name

Notes

• solo is not recommended and may lead to more 2D-style outputs. Use solo focus instead.
• Redundant tags such as shirt, red shirt are unnecessary, as duplicate and synonymous tags were filtered during training.

Training

Full Fine-Tuning using:

https://github.com/wochenlong/lora-scripts-next

Special thanks to Chenkin (且听风吟) for the training framework and technical support.

License

This model is a derivative work based on Anima.

It follows the original Anima license from Circle Stone Labs. Please review the included LICENSE file before use.

Commercial use is subject to the terms of the original Anima license.