README.md

metadata

title: Anima-2.9B
emoji: ⚡
colorFrom: indigo
colorTo: indigo
sdk: gradio
sdk_version: 6.25.0
app_file: run.py
pinned: false
hf_oauth: true
short_description: Anime/illustration text-to-image generator with Anima-2.9B
python_version: '3.12'
startup_duration_timeout: 30m

Anima-2.9B — Anime/Illustration Text-to-Image

A Hugging Face Space demo for Gazingstars123/Anima-2.9B, a 2.9B-parameter anime/illustration diffusion model (fine-tune + layer expansion of circlestone-labs/Anima, built on the Cosmos Predict2 backbone).

Prompting tips: The more detailed the prompt, the better. Use quality tags, year/period tags, @artist tags, character counts (1girl, 1boy), Danbooru/Gelbooru tags, series/copyrights, and base appearance.

Recommended generation settings:

• Sampler: Euler / Res-multistep / Er-sde
• Scheduler: sgm-uniform / beta / beta57 / linear-quadratic
• Resolution: 812×1216, 1152×1536, 1536×1536
• Steps: 28–50
• CFG: 3.5–5

Note: Model weights are released under the CircleStone Labs Non-Commercial License (derivative model category). For non-commercial use only.

Built on:

• nvidia/Cosmos-Predict2-2B-Text2Image
• circlestone-labs/Anima
• LLaMA Pro: Progressive LLaMA with Block Expansion