ComfyUI EcomID (Native Support)

Native SDXL-EcomID support for ComfyUI.

This extension doesn't use diffusers but instead implements EcomID natively and it fully integrates with ComfyUI.

Show Cases

EcomID enhances portrait representation, delivering a more authentic and aesthetically pleasing appearance while ensuring semantic consistency and greater internal ID similarity (i.e., traits that do not vary with age, hairstyle, glasses, or other physical changes).

You can see more showcases in the SDXL-EcomID repository.

Basic Workflow

In the examples directory, you'll find both the basic workflow and the dedicated face-swapping workflow.

Installation

Upgrade ComfyUI to the latest version!

Download or git clone this repository into the ComfyUI/custom_nodes/ directory or use the Manager.

EcomID requires insightface, you need to add it to your libraries together with onnxruntime and onnxruntime-gpu.

Models:

• PuLID pre-trained model goes in ComfyUI/models/pulid/ (thanks to Chenlei Hu for converting them into IPAdapter format)
• The EVA CLIP is EVA02-CLIP-L-14-336, but should be downloaded automatically (will be located in the huggingface directory).
• facexlib dependency needs to be installed, the models are downloaded at first use
• You also need InsightFace with AntelopeV2, the unzipped models should be placed in ComfyUI/models/insightface/models/antelopev2
• The ip_adapter model of InstantID can be downloaded from HuggingFace and should be placed into the ComfyUI/models/instantid directory. (Note that the model is called ip_adapter as it is based on the IPAdapter).
• You also needs a ControlNet trained on 2M real human images. Please place it in the ComfyUI controlnet directory.

Various Resolutions

Feel free to generate images in various resolutions, as we have trained the controlnet on 2 million high-quality images.

Normal CFG

You can set a higher CFG value to achieve even better semantic consistency. Values from 1 to 7 are acceptable, with 6 being recommended.

Face keypoints

The pose of the person is aligned with the keypoints generated from the reference image, just like in InstantID. You can achieve a different pose by sending an image to the image_kps input.

Advanced Node

There's an EcomID advanced node available, at the moment the only difference with the standard one is that you can set the weights for the ip-adapter (PulID) models and the controlnet separately.

Other notes

• Referenced the following repositories: ComfyUI_InstantID and PuLID_ComfyUI.

• Remember at the moment this is only compatible with SDXL-based models, such as EcomXL, leosams-helloworld-xl, dreamshaper-xl, stable-diffusion-xl-base-1.0 and so on.

• It works very well with SDXL Turbo/Lighting, EcomXL-Inpainting-ControlNet and EcomXL-Softedge-ControlNet.

The results of the Multi-ControlNet are shown below: