E-stick/anima-aesthetic-v1.1-GGUF

-- # Anima Aesthetic v1.1 — Q8_0 GGUF

- - Unofficial GGUF quantization of anima-aesthetic-v1.1 from circlestone-labs/Anima, converted to Q8_0 format using llama-quantize.

- - ## Model Info

- - | Field | Value | - |-------|-------| - | Architecture | Anima (Cosmos-based DiT) | - | Parameters | ~2B | - | Quantization | Q8_0 | - | File Size | ~2.1 GB | - | Original Precision | BF16/F16 |

- - ## Compatibility

- -

• —ComfyUI — Load via ComfyUI-GGUF plugin with the UnetLoaderGGUF node -
• —stable-diffusion.cpp — Native support, load text encoder with --llm

- - ## Recommended Settings

- - | Parameter | Value | - |-----------|-------| - | Sampler | euler / euler a | - | Scheduler | normal | - | Steps | 30–50 | - | CFG | 4–5 | - | Resolution | 512×768 / 768×512 |

- - ## Required Companion Models

- -

• —Text Encoder: qwen_3_06b_base.safetensors -
• —VAE: qwen_image_vae.safetensors

- - ## License

- - -This quantized model is subject to the CircleStone Labs Non-Commercial License. Non-commercial use only. - - - -Generated images may be used commercially. Model weights may not be used commercially without a separate license (contact tdrussell@circlestone.ai). - - +This quantized model is subject to the CircleStone Labs Non-Commercial License. Non-commercial use only. Generated images may be used commercially. Model weights may not be used commercially without a separate license (contact tdrussell@circlestone.ai).

The original model is based on NVIDIA Cosmos and is also subject to the NVIDIA Open Model License Agreement.

- - ## Credits

- -

• —Original model: circlestone-labs/Anima