Anima Prompt Architect v0.2.1

Image-to-Anima prompt generation using PixAI Tagger v1.0 BF16 and MiniCPM-V 4.7 1B Heretic native INT8 ConvRot.

• One self-contained safetensors file per model, including tokenizer and configuration assets.
• Base, Aesthetic and Turbo profiles, confidence/fidelity controls, positive/negative strings and structured/raw analysis.
• Native ComfyUI model ownership, sequential unloading and real Kitchen integer execution.
• Example visual/API workflows, conversion recipes and integrity/correctness tests.

Download both models from Hugging Face. Put them in ComfyUI/models/anima_prompt_architect/, install this node under custom_nodes, and restart ComfyUI. Follow the README for runtime requirements.

Portable hosted tests and local native CUDA checks passed. INT8 is not guaranteed faster than equally optimized BF16, and captions can contain inaccurate details. See the validation documentation for scope and limitations.