A sophisticated ComfyUI custom node engineered for advanced image background removal and precise segmentation of objects, faces, clothing, and fashion elements. This tool leverages a diverse array of models, including RMBG-2.0, INSPYRENET, BEN, BEN2, BiRefNet, SDMatte models, SAM, SAM2 and GroundingDINO, while also incorporating a new feature for real-time background replacement and enhanced edge detection for improved accuracy.
News & Updates
2026/01/01: Update ComfyUI-RMBG to v3.0.0 ( update.md )
2025/12/09: Update ComfyUI-RMBG to v2.9.6 ( update.md )
2025/03/13: Update ComfyUI-RMBG to v2.0.0 ( update.md )
Added Image and Mask Tools improved functionality.
Enhanced code structure and documentation for better usability.
Introduced a new category path: 🧪AILab/🛠️UTIL/🖼️IMAGE.
2025/02/24: Update ComfyUI-RMBG to v1.9.3 Clean up the code and fix the issue ( update.md )
2025/02/21: Update ComfyUI-RMBG to v1.9.2 with Fast Foreground Color Estimation ( update.md )
Added new foreground refinement feature for better transparency handling
Improved edge quality and detail preservation
Enhanced memory optimization
2025/02/20: Update ComfyUI-RMBG to v1.9.1 ( update.md )
Changed repository for model management to the new repository and Reorganized models files structure for better maintainability.
2025/02/19: Update ComfyUI-RMBG to v1.9.0 with BiRefNet model improvements ( update.md )
Enhanced BiRefNet model performance and stability
Improved memory management for large images
2025/02/07: Update ComfyUI-RMBG to v1.8.0 with new BiRefNet-HR model ( update.md )
Added a new custom node for BiRefNet-HR model.
Support high resolution image processing (up to 2048×2048)
2025/02/04: Update ComfyUI-RMBG to v1.7.0 with new BEN2 model ( update.md )
Added a new custom node for BEN2 model.
2025/01/22: Update ComfyUI-RMBG to v1.6.0 with new Face Segment custom node ( update.md )
Added a new custom node for face parsing and segmentation
Support for 19 facial feature categories (Skin, Nose, Eyes, Eyebrows, etc.)
Precise facial feature extraction and segmentation
Multiple feature selection for combined segmentation
Same parameter controls as other RMBG nodes
2025/01/05: Update ComfyUI-RMBG to v1.5.0 with new Fashion and accessories Segment custom node ( update.md )
Added a new custom node for fashion segmentation.
2025/01/02: Update ComfyUI-RMBG to v1.4.0 with new Clothes Segment node ( update.md )
Added intelligent clothes segmentation with 18 different categories
Support multiple item selection and combined segmentation
Same parameter controls as other RMBG nodes
2024/12/29: Update ComfyUI-RMBG to v1.3.2 with background handling ( update.md )
Enhanced background handling to support RGBA output when “Alpha” is selected.
Ensured RGB output for all other background color selections.
2024/12/25: Update ComfyUI-RMBG to v1.3.1 with bug fixes ( update.md )
Fixed an issue with mask processing when the model returns a list of masks.
Improved handling of image formats to prevent processing errors.
2024/12/23: Update ComfyUI-RMBG to v1.3.0 with new Segment node ( update.md )
Added text-prompted object segmentation
Support both tag-style (“cat, dog”) and natural language (“a person wearing red jacket”) prompts
Multiple models: SAM (vit_h/l/b) and GroundingDINO (SwinT/B) (as always model file will be downloaded automatically when first time using the specific model)
Windows desktop app: if the app crashes after install, set PYTHONUTF8=1 before installing requirements, then retry.
[!NOTE]
YOLO nodes require the optional ultralytics package. Install it only if you need YOLO to avoid dependency conflicts: ./ComfyUI/python_embeded/python -m pip install ultralytics --no-deps.
[!TIP]
Note: If your environment cannot install dependencies with the system Python, you can use ComfyUI’s embedded Python instead.
The model will be automatically downloaded to ComfyUI/models/RMBG/ when first time using the custom node.
Manually download the RMBG-2.0 model by visiting this link, then download the files and place them in the /ComfyUI/models/RMBG/RMBG-2.0 folder.
Manually download the INSPYRENET models by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/INSPYRENET folder.
Manually download the BEN model by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/BEN folder.
Manually download the BEN2 model by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/BEN2 folder.
Manually download the BiRefNet-HR by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/BiRefNet-HR folder.
Manually download the SAM models by visiting the link, then download the files and place them in the /ComfyUI/models/SAM folder.
Manually download the SAM2 models by visiting the link, then download the files (e.g., sam2.1_hiera_tiny.safetensors, sam2.1_hiera_small.safetensors, sam2.1_hiera_base_plus.safetensors, sam2.1_hiera_large.safetensors) and place them in the /ComfyUI/models/sam2 folder.
Manually download the GroundingDINO models by visiting the link, then download the files and place them in the /ComfyUI/models/grounding-dino folder.
Manually download the Clothes Segment model by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/segformer_clothes folder.
Manually download the Fashion Segment model by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/segformer_fashion folder.
Manually download BiRefNet models by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/BiRefNet folder.
Manually download SDMatte safetensors models by visiting the link, then download the files and place them in the /ComfyUI/models/RMBG/SDMatte folder.
| Sensitivity | Adjusts the strength of mask detection. Higher values result in stricter detection. | Default value is 0.5. Adjust based on image complexity; more complex images may require higher sensitivity. |
| Processing Resolution | Controls the processing resolution of the input image, affecting detail and memory usage. | Choose a value between 256 and 2048, with a default of 1024. Higher resolutions provide better detail but increase memory consumption. |
| Mask Blur | Controls the amount of blur applied to the mask edges, reducing jaggedness. | Default value is 0. Try setting it between 1 and 5 for smoother edge effects. |
| Mask Offset | Allows for expanding or shrinking the mask boundary. Positive values expand the boundary, while negative values shrink it. | Default value is 0. Adjust based on the specific image, typically fine-tuning between -10 and 10. |
| Background | Choose output background color | Alpha (transparent background) Black, White, Green, Blue, Red |
| Invert Output | Flip mask and image output | Invert both image and mask output |
| Refine Foreground | Use Fast Foreground Color Estimation to optimize transparent background | Enable for better edge quality and transparency handling |
| Performance Optimization | Properly setting options can enhance performance when processing multiple images. | If memory allows, consider increasing process_res and mask_blur values for better results, but be mindful of memory usage. |
Basic Usage
Load RMBG (Remove Background) node from the 🧪AILab/🧽RMBG category
Connect an image to the input
Select a model from the dropdown menu
select the parameters as needed (optional)
Get two outputs:
IMAGE: Processed image with transparent, black, white, green, blue, or red background
MASK: Binary mask of the foreground
Parameters
sensitivity: Controls the background removal sensitivity (0.0-1.0)
Load Segment (RMBG) node from the 🧪AILab/🧽RMBG category
Connect an image to the input
Enter text prompt (tag-style or natural language)
Select SAM and GroundingDINO models
Adjust parameters as needed:
Threshold: 0.25-0.35 for broad detection, 0.45-0.55 for precision
Mask blur and offset for edge refinement
Background color options
About Models
RMBG-2.0
RMBG-2.0 is is developed by BRIA AI and uses the BiRefNet architecture which includes:
High accuracy in complex environments
Precise edge detection and preservation
Excellent handling of fine details
Support for multiple objects in a single image
Output Comparison
Output with background
Batch output for video
The model is trained on a diverse dataset of over 15,000 high-quality images, ensuring:
Balanced representation across different image types
High accuracy in various scenarios
Robust performance with complex backgrounds
INSPYRENET
INSPYRENET is specialized in human portrait segmentation, offering:
Fast processing speed
Good edge detection capability
Ideal for portrait photos and human subjects
BEN
BEN is robust on various image types, offering:
Good balance between speed and accuracy
Effective on both simple and complex scenes
Suitable for batch processing
BEN2
BEN2 is a more advanced version of BEN, offering:
Improved accuracy and speed
Better handling of complex scenes
Support for more image types
Suitable for batch processing
BIREFNET MODELS
BIREFNET is a powerful model for image segmentation, offering:
BiRefNet-general purpose model (balanced performance)
BiRefNet_512x512 model (optimized for 512×512 resolution)
BiRefNet-portrait model (optimized for portrait/human matting)
BiRefNet-matting model (general purpose matting)
BiRefNet-HR model (high resolution up to 2560×2560)
BiRefNet-HR-matting model (high resolution matting)
BiRefNet_lite model (lightweight version for faster processing)
BiRefNet_lite-2K model (lightweight version for 2K resolution)
SAM
SAM is a powerful model for object detection and segmentation, offering:
High accuracy in complex environments
Precise edge detection and preservation
Excellent handling of fine details
Support for multiple objects in a single image
Output Comparison
Output with background
Batch output for video
SAM2
SAM2 is the latest segmentation model family designed for efficient, high-quality text-prompted segmentation:
Multiple sizes: Tiny, Small, Base+, Large
Optimized inference with strong accuracy
Automatic download on first use; manual placement supported in ComfyUI/models/sam2
GroundingDINO
GroundingDINO is a model for text-prompted object detection and segmentation, offering:
High accuracy in complex environments
Precise edge detection and preservation
Excellent handling of fine details
Support for multiple objects in a single image
Output Comparison
Output with background
Batch output for video
BiRefNet Models
BiRefNet-general purpose model (balanced performance)
BiRefNet_512x512 model (optimized for 512×512 resolution)
BiRefNet-portrait model (optimized for portrait/human matting)
BiRefNet-matting model (general purpose matting)
BiRefNet-HR model (high resolution up to 2560×2560)
BiRefNet-HR-matting model (high resolution matting)
BiRefNet_lite model (lightweight version for faster processing)
BiRefNet_lite-2K model (lightweight version for 2K resolution)
Requirements
ComfyUI
Python 3.10+
Required packages (automatically installed):
huggingface-hub>=0.19.0
transparent-background>=1.1.2
segment-anything>=1.0
groundingdino-py>=0.4.0
opencv-python>=4.7.0
onnxruntime>=1.15.0
onnxruntime-gpu>=1.15.0
protobuf>=3.20.2,<6.0.0
hydra-core>=1.3.0
omegaconf>=2.3.0
iopath>=0.1.9
SDMatte models (manual download)
Auto-download on first run to models/RMBG/SDMatte/
If network restricted, place weights manually:
models/RMBG/SDMatte/SDMatte.safetensors (standard) or SDMatte_plus.safetensors (plus)
Components (config files) are auto-downloaded; if needed, mirror the structure from the Hugging Face repo to models/RMBG/SDMatte/ (scheduler/, text_encoder/, tokenizer/, unet/, vae/)
Troubleshooting (short)
401 error when initializing GroundingDINO / missing models/sam2:
Delete %USERPROFILE%\.cache\huggingface\token (and %USERPROFILE%\.huggingface\token if present)
Ensure no HF_TOKEN/HUGGINGFACE_TOKEN env vars are set
Re-run; public repos download anonymously (no login required)
Preview shows “Required input is missing: images”:
Ensure image outputs are connected and upstream nodes ran successfully
