ComfyUI-GeekyRemB

ComfyUI-GeekyRemB
★ 49

背景去除抠图与遮罩动画与关键帧光照与合成
为ComfyUI提供基于rembg的图像处理节点:高精度背景去除、色键与遮罩修复,支持真实光照、投影和关键帧动画等高级合成功能。
💡 在ComfyUI中快速抠图并添加真实光照、投影和关键帧动画
🍴 5 Forks💻 Python🔄 2025-08-08
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/e58c8376a81b
📦 requirements.txt
#
GeekyRemB
v4.0
Requirements
#
=============================
#
Core
dependencies
numpy>=1.24.0
torch>=2.0.0
torchvision>=0.15.0
Pillow>=10.0.0
#
Computer
Vision
opencv-python>=4.8.0
scikit-image>=0.21.0
scipy>=1.11.0
#
Background
Removal
-
Latest
Models
rembg>=2.0.57
onnxruntime>=1.16.0
#
GPU
Acceleration
(uncomment
if
using
GPU)
onnxruntime-gpu>=1.16.0
#
Image
Processing
Advanced
kornia>=0.7.0
#
Advanced
computer
vision
operations
albumentations>=1.3.0
#
Image
augmentation
library
#
AI
Model
Integration
transformers>=4.35.0
huggingface-hub>=0.17.0
#
Performance
and
Utilities
tqdm>=4.66.0
requests>=2.31.0
psutil>=5.9.0
#
Memory
monitoring
#
Data
Handling
dataclasses-json>=0.6.0
#
Enhanced
dataclass
support
#
Optional:
Enhanced
Features
#
matplotlib>=3.8.0
#
For
debugging
visualizations
#
imageio>=2.31.0
#
Additional
image
I/O
formats
#
ffmpeg-python>=0.2.0
#
Video
processing
#
tensorboard>=2.14.0
#
Training
visualization
#
Development
Dependencies
(uncomment
for
development)
#
pytest>=7.4.0
#
black>=23.0.0
#
Code
formatting
#
flake8>=6.0.0
#
Code
linting
#
mypy>=1.6.0
#
Type
checking
#
Platform-specific
optimizations
#
Intel
processors:
#
intel-extension-for-pytorch>=2.0.0
#
AMD
ROCm
support:
#
torch-audio-rocm>=2.0.0
#
Apple
Silicon
optimization:
#
torch-audio>=2.0.0
#
Additional
AI
Models
(optional)
#
segment-anything>=1.0
#
detectron2>=0.6
#
Object
detection
#
timm>=0.9.0
#
Vision
models
#
Note:
Some
packages
may
require
specific
versions
based
on
your
system
#
For
best
compatibility,
use
virtual
environments:
#
python
-m
venv
geeky_remb_env
#
source
geeky_remb_env/bin/activate
#
Linux/Mac
#
geeky_remb_env\Scripts\activate
#
Windows
#
pip
install
-r
requirements.txt
📄 README

GeekyRemB v4.0: Ultimate AI Background Removal & Video Layering Node for ComfyUI

GeekyRemB v4.0 is a revolutionary image and video processing node suite that brings state-of-the-art AI models, professional compositing, advanced animation, and cutting-edge multimodal capabilities to ComfyUI. Built on the latest AI architectures including RMBG v2.0 with BiRefNet, it represents the pinnacle of background removal and video layering technology.

🚀 What’s New in v4.0

  • Latest AI Models: RMBG v2.0 with BiRefNet architecture for superior background removal
  • Advanced Multimodal Processing: Cross-modal analysis and generation capabilities
  • Professional Animation System: 20+ animation types with advanced easing functions
  • Physics-Aware Generation: Realistic simulation for next-generation content
  • Professional Blending Modes: 15+ industry-standard compositing modes
  • Self-Optimizing Workflows: AI-powered workflow optimization and generation
  • GPU Acceleration: Optimized performance with intelligent memory management
  • Video Batch Processing: Frame-consistent processing for video sequences
  • Table of Contents

  • Installation
  • Core Features
  • AI Models & Methods
  • Professional Animation System
  • Advanced Positioning & Blending
  • Multimodal Processing
  • Workflow Examples
  • Performance Optimization
  • API Reference
  • Troubleshooting

  • Installation

    Prerequisites

  • ComfyUI (latest version recommended)
  • Python 3.8+
  • PyTorch 2.0+
  • CUDA-capable GPU (recommended for optimal performance)
  • Quick Installation

  • Clone the repository:
  • “`bash

    cd ComfyUI/custom_nodes

    git clone https://github.com/GeekyGhost/ComfyUI-GeekyRemB.git

    cd ComfyUI-GeekyRemB

    “`

  • Install dependencies:
  • “`bash

    pip install -r requirements.txt

    “`

  • GPU Acceleration Setup (Optional but recommended):
  • “`bash

    # For NVIDIA GPUs

    pip install onnxruntime-gpu>=1.16.0

    # For AMD ROCm

    pip install torch-audio-rocm>=2.0.0

    “`

  • Restart ComfyUI to load the new nodes.
  • Advanced Installation Options

    For development or advanced users:

    # Create virtual environment
    python -m venv geeky_remb_env
    source geeky_remb_env/bin/activate  # Linux/Mac
    # or: geeky_remb_env\Scripts\activate  # Windows
    
    # Install with optional features
    pip install -r requirements.txt
    pip install matplotlib imageio ffmpeg-python  # Optional enhancements


    Core Features

    🎯 State-of-the-Art Background Removal

    Supported AI Models:

  • RMBG v2.0 (BiRefNet) – Latest architecture for superior quality
  • RMBG v1.4 (BRIA AI) – Professional-grade removal
  • BiRefNet Direct – High-performance segmentation
  • Advanced Chroma Key – Professional green/blue screen with spill suppression
  • Hybrid AI+Chroma – Combines AI and chroma key for optimal results
  • Advanced Features:

  • Adaptive model selection based on content analysis
  • Real-time edge refinement with alpha matting
  • Intelligent mask processing with small object removal
  • Professional spill suppression and edge feathering
  • Batch processing with frame consistency
  • 🎬 Professional Video Processing

  • Multi-frame Processing: Optimized for video sequences with frame consistency
  • Thread Pool Optimization: Parallel processing for maximum performance
  • Advanced Caching: LRU cache system reduces redundant computations
  • Memory Management: Intelligent VRAM handling from 1GB to 64GB+ systems
  • Export Options: Multiple video formats with quality optimization
  • 🎨 Advanced Compositing Engine

  • 15+ Blend Modes: Industry-standard modes including Multiply, Screen, Overlay, Soft Light
  • Professional Positioning: 11 positioning modes from absolute pixels to custom anchors
  • Transform Controls: Scale, rotation, flip with sub-pixel precision
  • Layer Effects: Drop shadows, perspective shadows, lighting effects
  • Alpha Channel Management: Sophisticated transparency handling

  • AI Models & Methods

    Model Selection Guide

    | Model | Best For | Performance | Quality | GPU Memory |

    |——-|———-|————-|———|————|

    | RMBG v2.0 (BiRefNet) | General use, highest quality | Medium | Excellent | 4-8GB |

    | RMBG v1.4 (BRIA AI) | Professional workflows | Fast | Very Good | 2-4GB |

    | BiRefNet Direct | Complex subjects, fine details | Slow | Excellent | 6-12GB |

    | Advanced Chroma Key | Studio environments | Very Fast | Good | <1GB |

    | Hybrid AI+Chroma | Mixed environments | Medium | Excellent | 4-6GB |

    Configuration Examples

    # High-quality portrait processing
    removal_config = BackgroundRemovalConfig(
        method=RemovalMethod.RMBG_V2,
        model_precision="fp16",
        processing_resolution=1024,
        edge_feathering=3.0,
        remove_small_objects=True,
        anti_aliasing=True
    )
    
    # Speed-optimized processing  
    removal_config = BackgroundRemovalConfig(
        method=RemovalMethod.RMBG_V1_4,
        processing_resolution=512,
        batch_size=4,
        use_gpu_acceleration=True
    )
    
    # Professional chroma key
    removal_config = BackgroundRemovalConfig(
        method=RemovalMethod.CHROMA_KEY,
        chroma_color="green",
        tolerance=0.15,
        spill_suppression=0.9,
        edge_feathering=2.0
    )


    Professional Animation System

    20+ Animation Types

    Movement Animations:

  • slide_left, slide_right, slide_up, slide_down, slide_diagonal
  • orbit_circular, spiral_in, bounce_in, bounce_out
  • Transform Animations:

  • scale_in, scale_out, scale_pulse
  • rotate_clockwise, rotate_counter_clockwise
  • flip_horizontal, flip_vertical
  • Special Effects:

  • fade_in, fade_out, fade_pulse
  • elastic_in, elastic_out
  • shake_subtle, wobble
  • Advanced Easing Functions

    Professional easing for natural motion:

  • Linear: Constant speed movement
  • Quadratic: ease_in_quad, ease_out_quad, ease_in_out_quad
  • Cubic: Smooth acceleration/deceleration
  • Sine: Natural sinusoidal motion
  • Exponential: Dramatic speed changes
  • Elastic: Spring-like bouncing effects
  • Bounce: Realistic bouncing motion
  • Back: Overshoot and settle effects
  • Animation Configuration

    # Smooth bounce animation
    animation_config = AnimationConfig(
        animation_type=AnimationType.BOUNCE_IN,
        duration=2.0,
        easing=EasingFunction.EASE_OUT_BOUNCE,
        amplitude=150.0,
        frequency=1.0,
        loop_count=3,
        ping_pong=True
    )
    
    # Complex orbital motion
    animation_config = AnimationConfig(
        animation_type=AnimationType.ORBIT_CIRCULAR,
        duration=4.0,
        easing=EasingFunction.SINE_IN_OUT,
        amplitude=200.0,
        frequency=2.0,
        start_delay=0.5
    )


    Advanced Positioning & Blending

    11 Positioning Modes

  • Absolute Pixels – Precise pixel positioning
  • Relative Percent – Percentage-based positioning
  • Center – Automatic centering
  • Corner Positions – Top/bottom + left/center/right combinations
  • Custom Anchor – User-defined anchor points with offset
  • Professional Blend Modes

    | Blend Mode | Effect | Best For |

    |————|——–|———-|

    | Normal | Standard overlay | General compositing |

    | Multiply | Darkening effect | Creating shadows, depth |

    | Screen | Lightening effect | Adding highlights, glow |

    | Overlay | Contrast enhancement | Dramatic effects |

    | Soft Light | Subtle lighting | Natural lighting effects |

    | Hard Light | Strong contrast | Bold, dramatic looks |

    | Color Dodge/Burn | Extreme lighting | Special effects |

    | Darken/Lighten | Selective blending | Texture blending |

    | Difference | Inversion effects | Abstract, artistic looks |

    Advanced Layer Effects

    # Professional drop shadow
    blend_config = BlendingConfig(
        blend_mode=BlendMode.NORMAL,
        opacity=0.9,
        drop_shadow=True,
        shadow_offset_x=8.0,
        shadow_offset_y=12.0,
        shadow_blur=15.0,
        shadow_opacity=0.6,
        shadow_color=(0, 0, 0)
    )
    
    # Cinematic lighting effect
    blend_config = BlendingConfig(
        blend_mode=BlendMode.SOFT_LIGHT,
        opacity=0.8,
        preserve_luminosity=True,
        knock_out=False
    )


    Multimodal Processing

    Cross-Modal Analysis

    GeekyRemB v4.0 introduces cutting-edge multimodal capabilities:

  • Vision-Audio Synchronization: Automatic sync detection for video content
  • Text-to-Visual Generation: AI-powered visual generation from descriptions
  • Cross-Modal Search: Find content across different modalities
  • Multimodal Reasoning: Advanced AI reasoning across image, audio, and text
  • Usage Examples

    # Multimodal content analysis
    result = multimodal_processor.process_multimodal(
        processing_mode="analysis",
        image=input_image,
        audio_path="audio_track.wav",
        text_prompt="Analyze the emotional content",
        modality_weights={"vision": 0.4, "audio": 0.4, "text": 0.2}
    )
    
    # Cross-modal generation
    result = multimodal_processor.process_multimodal(
        processing_mode="generation",
        text_prompt="A serene sunset over calm waters",
        reference_style=style_image,
        quality_target="cinema",
        output_modalities="all"
    )


    Workflow Examples

    1. Basic Professional Background Removal

    # Input: Portrait photo with complex background
    # Output: Clean subject with transparent background
    
    workflow = {
        "nodes": {
            "1": {
                "class_type": "GeekyRemB",
                "inputs": {
                    "foreground": ["load_image", 0],
                    "skip_background_removal": False,
                    "removal_method": "rmbg_v2_birefnet",
                    "processing_resolution": 1024,
                    "mask_blur": 1.5,
                    "edge_feathering": 3.0,
                    "remove_small_objects": True
                }
            }
        }
    }

    2. Cinematic Character Animation

    # Create a cinematic character entrance with lighting
    workflow = {
        "character_removal": {
            "removal_method": "birefnet_zhengpeng7",
            "quality_target": "cinema",
            "edge_feathering": 2.0
        },
        "animation": {
            "animation_type": "slide_right",
            "duration": 3.0,
            "easing": "ease_out_cubic",
            "amplitude": 400.0
        },
        "lighting": {
            "enable_lighting": True,
            "light_direction_x": -100,
            "light_direction_y": -150,
            "light_intensity": 0.8,
            "enable_shadow": True,
            "shadow_opacity": 0.4
        },
        "composition": {
            "blend_mode": "soft_light",
            "position_mode": "center",
            "scale_x": 1.2,
            "scale_y": 1.2
        }
    }

    3. Multi-Subject Video Processing

    # Process video with multiple subjects and complex backgrounds
    workflow = {
        "batch_processing": {
            "total_frames": 120,
            "video_fps": 24.0,
            "removal_method": "hybrid_ai_chroma",
            "batch_processing": True
        },
        "optimization": {
            "use_gpu_acceleration": True,
            "processing_resolution": 720,
            "thread_count": 8
        },
        "output": {
            "video_format": "mp4",
            "quality": "high",
            "maintain_frame_consistency": True
        }
    }

    4. Advanced Keyframe Animation

    # Create complex keyframe-based animation sequence
    keyframes = [
        {
            "frame": 0,
            "x_position": -200,
            "y_position": 0,
            "scale": 0.8,
            "rotation": 0,
            "opacity": 0.0
        },
        {
            "frame": 30,
            "x_position": 0,
            "y_position": -50,
            "scale": 1.0,
            "rotation": 15,
            "opacity": 1.0
        },
        {
            "frame": 60,
            "x_position": 150,
            "y_position": 0,
            "scale": 1.1,
            "rotation": 0,
            "opacity": 0.9
        },
        {
            "frame": 90,
            "x_position": 0,
            "y_position": 0,
            "scale": 1.0,
            "rotation": 0,
            "opacity": 1.0
        }
    ]
    
    animation_config = {
        "use_keyframes": True,
        "easing_function": "ease_in_out_cubic",
        "total_frames": 90,
        "fps": 30
    }


    Performance Optimization

    GPU Memory Management

    GeekyRemB v4.0 includes intelligent memory management for various GPU configurations:

    | GPU Memory | Recommended Settings | Performance |

    |————|———————|————-|

    | 1-2GB | LOW_VRAM mode, 512px resolution | Basic processing |

    | 4-6GB | NORMAL_VRAM mode, 1024px resolution | Good performance |

    | 8-12GB | HIGH_VRAM mode, 1536px resolution | Optimal performance |

    | 16GB+ | Maximum settings, 2048px+ resolution | Maximum quality |

    Optimization Settings

    # For 4GB GPU
    optimization_config = {
        "processing_resolution": 1024,
        "batch_size": 2,
        "use_gpu_acceleration": True,
        "model_precision": "fp16",
        "memory_management": "NORMAL_VRAM"
    }
    
    # For 16GB+ GPU  
    optimization_config = {
        "processing_resolution": 2048,
        "batch_size": 8,
        "use_gpu_acceleration": True,
        "model_precision": "fp32",
        "memory_management": "HIGH_VRAM",
        "enable_model_caching": True
    }
    
    # CPU-optimized (no GPU)
    optimization_config = {
        "processing_resolution": 512,
        "batch_size": 1,
        "use_gpu_acceleration": False,
        "thread_count": 8,
        "memory_management": "CPU_ONLY"
    }

    Performance Tips

  • Model Selection: Use lighter models (RMBG v1.4) for real-time applications
  • Resolution: Process at lower resolution, then upscale if needed
  • Batch Processing: Group similar operations for efficiency
  • Caching: Enable model and result caching for repeated operations
  • Threading: Adjust thread count based on CPU cores (recommended: CPU cores – 1)

  • API Reference

    Core Classes

    GeekyRemBv4

    The main processing node with comprehensive background removal and compositing capabilities.

    Key Parameters:

  • removal_method: Choose AI model or chroma key method
  • processing_resolution: Balance quality vs. performance
  • animation_type: Select from 20+ animation types
  • blend_mode: Professional compositing modes
  • position_mode: Flexible positioning system
  • BackgroundRemovalConfig

    Advanced configuration for background removal algorithms.

    config = BackgroundRemovalConfig(
        method=RemovalMethod.RMBG_V2,
        model_precision="fp16",
        batch_size=4,
        chroma_color="green",
        tolerance=0.15,
        spill_suppression=0.9,
        edge_feathering=3.0,
        processing_resolution=1024,
        anti_aliasing=True
    )

    AnimationConfig

    Professional animation control system.

    config = AnimationConfig(
        animation_type=AnimationType.BOUNCE_IN,
        duration=2.0,
        easing=EasingFunction.EASE_OUT_BOUNCE,
        amplitude=100.0,
        frequency=1.0,
        loop_count=3,
        ping_pong=True
    )

    PositionConfig

    Comprehensive positioning and transform system.

    config = PositionConfig(
        mode=PositionMode.CUSTOM_ANCHOR,
        x_offset=100.0,
        y_offset=-50.0,
        anchor_x=0.5,
        anchor_y=0.3,
        scale_x=1.2,
        scale_y=1.0,
        rotation=15.0,
        flip_horizontal=False
    )

    Input/Output Types

    Supported Input Types:

  • IMAGE: Standard ComfyUI image tensors
  • STRING: Text prompts and file paths
  • FLOAT: Numerical parameters with validation
  • INT: Integer values with range checking
  • BOOLEAN: Toggle switches for features
  • Output Types:

  • IMAGE: Processed image sequences
  • MASK: Generated alpha masks
  • STRING: Processing metadata and logs
  • FLOAT: Confidence scores and metrics

  • Troubleshooting

    Common Issues & Solutions

    1. Background Removal Quality Issues

    Problem: Poor edge quality or missed details

    Solutions:

  • Try RMBG v2.0 (BiRefNet) for best quality
  • Enable alpha_matting for fine details like hair
  • Adjust edge_feathering (2.0-5.0 for complex edges)
  • Increase processing_resolution to 1024 or higher
  • Enable anti_aliasing for smoother edges
  • 2. Performance & Memory Issues

    Problem: Slow processing or out-of-memory errors

    Solutions:

  • Reduce processing_resolution (try 512px)
  • Switch to lighter model (RMBG v1.4)
  • Enable LOW_VRAM mode in settings
  • Reduce batch_size to 1-2
  • Close other GPU-intensive applications
  • Use CPU mode for very limited systems
  • 3. Animation Issues

    Problem: Jerky or incorrect animations

    Solutions:

  • Check fps setting matches your target frame rate
  • Verify keyframe frame_number values are in correct sequence
  • Ensure easing_function is appropriate for animation type
  • Check that total_frames encompasses all keyframes
  • Verify node connections are correct
  • 4. Video Processing Issues

    Problem: Frame inconsistency or export failures

    Solutions:

  • Enable batch_processing for video sequences
  • Check that all frames have consistent dimensions
  • Verify video_fps matches source material
  • Ensure sufficient disk space for output
  • Try different output formats (mp4, webm)
  • 5. Chroma Key Issues

    Problem: Color spill or incomplete removal

    Solutions:

  • Adjust tolerance (start with 0.15, increase gradually)
  • Enable spill_suppression (0.7-0.9)
  • Increase edge_feathering (2.0-5.0)
  • Check lighting conditions in source material
  • Consider hybrid_ai_chroma method for difficult scenarios
  • Debug Mode

    Enable debug logging for detailed troubleshooting:

    import logging
    logging.basicConfig(level=logging.DEBUG)
    logger = logging.getLogger(__name__)

    System Requirements Verification

    Minimum Requirements:

  • Python 3.8+
  • PyTorch 2.0+
  • 4GB RAM
  • 2GB storage space
  • Recommended:

  • Python 3.10+
  • PyTorch 2.1+
  • CUDA-capable GPU with 8GB+ VRAM
  • 16GB+ system RAM
  • SSD storage
  • Optimal:

  • Latest Python 3.11+
  • PyTorch 2.2+ with CUDA 12+
  • RTX 4080/4090 or equivalent
  • 32GB+ system RAM
  • NVMe SSD storage

  • Contributing

    We welcome contributions from the community! Please see our Contributing Guidelines for details on:

  • Code style and standards
  • Testing requirements
  • Documentation guidelines
  • Pull request process
  • Development Setup

    # Clone repository
    git clone https://github.com/GeekyGhost/ComfyUI-GeekyRemB.git
    cd ComfyUI-GeekyRemB
    
    # Create development environment
    python -m venv dev_env
    source dev_env/bin/activate
    
    # Install development dependencies
    pip install -r requirements.txt
    pip install pytest black flake8 mypy
    
    # Run tests
    pytest tests/
    
    # Code formatting
    black --check scripts/
    flake8 scripts/
    mypy scripts/


    License & Acknowledgements

    License

    This project is licensed under the MIT License. See LICENSE for details.

    Note: Some AI models may have separate licensing requirements for commercial use. Please review individual model licenses.

    Acknowledgements

    GeekyRemB v4.0 builds upon exceptional open-source projects:

  • RMBG – BiRefNet architecture for background removal
  • BRIA AI – Professional-grade RMBG models
  • ComfyUI – The incredible node-based UI framework
  • OpenCV – Computer vision processing
  • PyTorch – Deep learning framework
  • Special Thanks

  • ComfyUI Community – Feedback, testing, and feature requests
  • AI Researchers – Advancing the state-of-the-art in background removal
  • Open Source Contributors – Making this technology accessible to everyone
  • Beta Testers – Helping refine v4.0 before release

  • Support & Community

  • Issues: GitHub Issues
  • Discussions: GitHub Discussions
  • Documentation: Wiki

  • GeekyRemB v4.0 – Pushing the boundaries of what’s possible in AI-powered image and video processing.

    *Made with ❤️ by GeekyGhost and the open-source community.*