FFF_ANIMA-29B

A full fine-tune of OOO_ANIMA-29B v11 base (Anima 2.9B lineage).

One trigger word brings up a consistent soft-shading style, and faces do not blush unless you ask for it.

OOO_ANIMA-29B v11 base を Full Fine-Tune したモデルです。trigger を 1 語置くだけで柔らかい塗りの画風が出て、指示しない限り赤面しません。

Hugging Face: https://huggingface.co/flyfront/FFF_ANIMA-29B (published 2026-09-13)

### Files

- FFF_ANIMA-29B-v1.safetensors — bf16, 5.84 GB. Reference weights, for further training and merging.

- FFF_ANIMA-29B-v1_int8_convrot.safetensors — INT8 ConvRot, 3.16 GB. For everyday generation in ComfyUI or Forge Neo. Output is practically identical.

Both contain the DiT only, so you also need the VAE and text encoder, the same ones OOO_ANIMA-29B uses:

- VAE: OOO_Anima_VAE_v3.5.safetensors (from the OOO_ANIMA-29B page) or qwen_image_vae.safetensors

- Text encoder: qwen_3_06b_base.safetensors (from circlestone-labs/Anima)

SHA256:

- bf16: 63920103cd6d2b1ef5859c1715d3a81443f0fb42d84c07686231c6cd8bb18742

- int8: 86431a4c3aa9dad494335fd18503123cf5f0c0d53249a4c9bc3fd42e348651c6

### Prompt

```

<your tags>, @fyfnt, masterpiece, best quality, high quality, score_9, newest, <rating>

```

<rating> is one of the Anima rating tags; the samples use safe, some weighted as (safe:2.0).

The trigger can sit at the head of the prompt instead; both were checked on the same prompts and seeds and the style comes out the same.

The official Anima card puts the quality tags at the head of the prompt, but this model was trained with them at the end of the caption, so the end is the intended position (the head still works).

<rating> には Anima の rating タグsafe など)を入れます。trigger は先頭に置いても構いません(同じプロンプトと seed で両方を比べてあり、画風は変わりません)。品質タグは公式 Anima では先頭ですが、このモデルは末尾に置いて学習しているので末尾が想定の位置です(先頭でも動きます)。

### Negative prompt

```

worst quality, low quality, score_1, score_2, score_3, early, old, realistic, 3d, photorealistic, sketch, blush, embarrassed, empty eyes, ringed eyes, light particles, artifacts, copyright name, japanese text, page number, bad anatomy, bad hands, border

```

Optional negatives for keeping outfits more covered are listed on the Hugging Face card (linked above).

### Settings

ER SDE or Euler, 25-30 steps, CFG 4, Shift 3. 1536x1536 is the training resolution and the safest choice; 1344x1792 (portrait) and 1792x1344 (landscape) work as well and the samples use those two. Going much above that area makes the style unstable.

### Notes

- Blush: about 20% of images on a fixed 30-image benchmark when nothing in the prompt asks for it (the base model on the same prompts: 13%). Write blush when you want it.

- Rating tags behave as in the base model.

- The trigger of the OOO_ANIMA refined model has no effect here.

### Lineage and license

Anima (CircleStone Labs) → Anima-2.9B (Gazingstars123) → OOO_ANIMA-29B v11 base (oron1208380) → FFF_ANIMA-29B v1

This model is a Derivative under the CircleStone Labs Non-Commercial License v1.2, inherited from the models above. Non-commercial use only: no paid generation services, no selling access, no revenue from hosting the model. You may share and modify it, but keep the license and this attribution with any redistribution. Generated images are yours.

Please do not re-upload this model to generation services that charge users. Re-uploads can be checked against the SHA256 above.

上流から継承した CircleStone Labs Non-Commercial License v1.2 に従います。非商用に限ります(有料の生成サービスでの提供、アクセスの販売、ホスティングによる収益は不可)。再配布と改変はできますが、ライセンスとこの帰属表示を一緒に付けてください。生成した画像は利用者のものです。利用者に課金する生成サービスへの転載はしないでください。転載は上の SHA256 で照合できます。