BiRefNet

BiRefNet

Background-removal model files re-packed by Comfy-Org so ComfyUI can load them. They make a mask that separates a subject from its background. For ComfyUI users who cut subjects out of images.

Made by Comfy-Org on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This repository holds BiRefNet model weights that Comfy-Org re-packed so ComfyUI can load them. BiRefNet makes a mask that separates the subject of an image from its background. It lists two files, birefnet.safetensors and lucida.safetensors, and names ZhengPeng7/BiRefNet and egeorcun/lucida as the original repositories.

Who it suits: It suits people who build image workflows in ComfyUI and want a ready way to cut a subject out of a picture. It is a set of model files, not an app.

Title: Who uses it
Industry | Who uses it | How they use it | Example
Image editing | Designers | Cut a subject out of a photo inside ComfyUI | Remove the backdrop from a product photo
E-commerce | Shop owners | Prepare clean product pictures | Get a cut-out of an item to place on a new background
AI art | Workflow builders | Add a mask step to a ComfyUI graph | Mask the subject before another model edits the rest

People who do not use ComfyUI can skip it, since the card gives no other way to run it.

Where to use it:

• ComfyUI, the only place the card describes
• Windows, macOS or Linux with Python, to download the files
• The ComfyUI workflow template named Remove Background: BiRefNet

Title: Top jobs
Use case | Who | What they get back
Background removal | Designers | A subject separated from its background
Mask generation | Workflow builders | A mask other nodes can use
Product pictures | Shop owners | A clean cut-out of an item

How to set it up:

• ComfyUI installed on your computer
• Python, for the download
• The huggingface_hub package

1. Install the Hugging Face hub package.
2. Download the model files to your computer.
3. Copy birefnet.safetensors and lucida.safetensors into the background_removal folder, inside ComfyUI's models folder.
4. Restart ComfyUI so it sees the new files.

pip install -U huggingface_hub
hf download Comfy-Org/BiRefNet

How to use it: In ComfyUI, open the workflow template named Remove Background: BiRefNet from the Comfy-Org workflow templates, then run it on your image. The card gives no commands beyond the download.

Example: You open the Remove Background: BiRefNet template and run it on a photo of a product. The workflow returns the subject without its background. The card shows a before and after thumbnail pair, so check the result on your own picture.

For newcomers and experts: If you are new, a mask is a map that marks which parts of a picture are the subject. The first thing to try is the card's template with one photo you know well.

For experts, the record lists the task as mask generation and the library as diffusion-single-file. The weights are single safetensors files that ComfyUI reads from its models folder, so no extra code is needed. The card does not say which of the two files is better.

Download the two safetensors files
Copy them into the models folder
Open the BiRefNet workflow
Run it to get the mask

Good to know:

• Repackaged files: the original work comes from the repositories the card names.
• License: MIT.
• Reach: it runs in ComfyUI on your computer.
• Cost to run: none named.
• Data it reaches: it downloads files from Hugging Face, and the source names nothing else.
• Not for: chat, writing or any use outside ComfyUI.

On Hugging Face, 13 November 2026

• Task: mask generation
• Library: diffusion-single-file
• Downloads in the last 30 days: 232,004
• Likes: 64
• License: mit
• Based on: ZhengPeng7/BiRefNet
• Tags: diffusion-single-file, comfyui, mask-generation

Links

• Model: https://huggingface.co/Comfy-Org/BiRefNet
• Files: https://huggingface.co/Comfy-Org/BiRefNet/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download Comfy-Org/BiRefNet

Then follow the usage in its model card. Nothing in this kit runs by itself.