luxdelux7/ComfyUI-Forbidden-Vision

Integrates three custom-trained models (YOLO11 detection, segmentation, and a neural corrector) that perform learned tone and color adjustment alongside face restoration within ComfyUI's node workflow. The modular architecture chains detection→segmentation→context-aware inpainting with automatic color correction and feathered blending, while the Refiner and Builder nodes provide pre- and post-processing passes with self-correction and integrated VAE decoding. Models auto-download from HuggingFace on first use.