ComfyUI_IPAdapter_plus

ComfyUI IPAdapter Plus is an extension for ComfyUI, a node-based tool for generating images with AI. The extension adds support for IPAdapter models, which let you use a reference image to influence how newly generated images look. Instead of describing a style in words, you provide a photo, and the model applies its visual character to the output. The README describes it as a one-image LoRA, meaning it works similarly to a style preset but driven entirely by an example image.

The extension supports a range of pre-trained model files covering different use cases: basic style transfer, stronger style transfer, portrait-focused generation that captures a specific person's face, and face identity models that preserve a subject's likeness across multiple generations. There are variants for both SD 1.5 and SDXL base models. All model files are downloaded separately from HuggingFace and placed in specific folders within the ComfyUI directory.

Installation involves cloning the repository into ComfyUI's custom_nodes folder or using the ComfyUI Manager. After installing, the extension adds new nodes to ComfyUI's visual graph editor. A Unified Model Loader node can automatically detect and load the correct models when files are named according to the naming convention described in the README. Older loader nodes still work with any file names but require manual selection.

The examples directory includes ready-made workflow files that demonstrate the full range of features. Several video tutorials are also linked in the README covering basic usage, style and composition transfer, attention masking, and animation features.

As of April 2025, the repository entered maintenance-only mode. The author no longer actively develops it, though critical pull requests may still be considered.