ComfyUI Extension: ComfyUI Deforum-X-Flux Nodes
Professional video animation nodes for ComfyUI based on Deforum-X-Flux research
Custom Nodes (0)
README
ComfyUI Deforum-X-Flux Nodes
Professional video animation nodes for ComfyUI, based on the original XmYx/deforum-comfy-nodes with FLUX model integration.
🎬 Features
- Original Deforum Architecture: Follows the exact node structure and workflow patterns from XmYx/deforum-comfy-nodes
- FLUX Model Integration: Enhanced for high-quality generation using state-of-the-art FLUX models
- Mathematical Motion Control: Use expressions like
0:(1.0), 30:(1.5*sin(t/10))
for dynamic animations - Parameter Chaining: Modular parameter nodes that chain together like the original
- Frame-by-Frame Generation: Iterator-based system with caching for smooth animations
- Professional Video Output: Export to MP4, GIF, WebM with FFmpeg integration
🚀 Installation
Method 1: ComfyUI Manager (Recommended)
- Open ComfyUI Manager
- Search for "Deforum-X-Flux"
- Click Install
- Restart ComfyUI
Method 2: Manual Installation
# Navigate to ComfyUI custom nodes directory
cd ComfyUI/custom_nodes/
# Clone the repository
git clone https://github.com/A043-studios/comfyui-deforum-x-flux-nodes.git
# Install dependencies
cd comfyui-deforum-x-flux-nodes
pip install -r requirements.txt
# Restart ComfyUI
Method 3: Download and Extract
- Download the latest release ZIP
- Extract to
ComfyUI/custom_nodes/comfyui-deforum-x-flux-nodes/
- Install dependencies:
pip install -r requirements.txt
- Restart ComfyUI
📋 Requirements
System Requirements
- Python: 3.9 or higher
- GPU: NVIDIA GPU with 8GB+ VRAM (12GB+ recommended)
- RAM: 16GB+ (32GB recommended for large animations)
- Storage: 10GB+ free space for models and outputs
Software Dependencies
- ComfyUI: Latest version
- PyTorch: 2.0.0 or higher
- FFmpeg: For video output (install separately)
Hardware Recommendations
| Quality Level | GPU | VRAM | RAM | Performance | |---------------|-----|------|-----|-------------| | Basic | RTX 3060 | 12GB | 16GB | 2-5 sec/frame | | Standard | RTX 4070 | 16GB | 32GB | 1-3 sec/frame | | Professional | RTX 4090 | 24GB | 64GB | <1 sec/frame |
🎮 Quick Start
Basic 2D Animation Workflow
-
🎬 Add Animation Setup Node
- Set animation mode to "2D"
- Configure 60 frames at 1024x1024
- Set FPS to 12
-
📐 Add Motion Controller Node
- Connect to Animation Setup
- Set angle:
0:(0), 30:(180), 60:(360)
- Set zoom:
0:(1.0), 60:(1.2)
-
🎭 Add Keyframe Manager Node
- Connect to Animation Setup
- Set prompts:
0:(beautiful landscape), 30:(sunset landscape), 60:(night landscape)
-
🎨 Add Deforum Renderer Node
- Connect all previous nodes
- Connect your FLUX model, VAE, and CLIP
- Generate animation frames
-
📹 Add Video Output Node
- Connect rendered images
- Choose MP4 format
- Set output path and quality
Example Mathematical Expressions
# Smooth camera movement
translation_x = "0:(sin(t*0.08)*100 + cos(t*0.03)*50)"
translation_y = "0:(cos(t*0.08)*100 + sin(t*0.03)*50)"
# Dynamic zoom with breathing effect
zoom = "0:(1.0 + sin(t*0.05)*0.3 + cos(t*0.02)*0.1)"
# Complex rotation patterns
angle = "0:(sin(t*0.1)*45 + t*0.5)"
# 3D camera movements
rotation_3d_y = "0:(sin(t*0.06)*15 + t*1.5)"
translation_z = "0:(sin(t*0.04)*30)"
📚 Node Reference
🎬 Deforum Animation Setup
Purpose: Core animation configuration and setup
Inputs:
animation_mode
: 2D, 3D, Video Input, Interpolationmax_frames
: Number of frames (1-10000)width/height
: Output dimensions (64-4096, multiples of 64)fps
: Frame rate (1-60)border_mode
: replicate, wrapuse_depth_warping
: Enable 3D depth processingcolor_coherence
: Frame-to-frame color consistency
Outputs:
animation_config
: Configuration for other nodes
Example:
Mode: 2D
Frames: 120
Size: 1024x1024
FPS: 24
📐 Deforum Motion Controller
Purpose: Mathematical motion parameter control
Inputs:
animation_config
: From Animation Setupangle
: Rotation keyframes (degrees)zoom
: Zoom keyframes (scale factor)translation_x/y
: 2D movement keyframes (pixels)translation_z
: 3D depth movement (units)rotation_3d_x/y/z
: 3D rotation keyframes (degrees)noise_schedule
: Noise amount over time (0.0-1.0)strength_schedule
: Generation strength (0.0-1.0)
Mathematical Variables:
t
: Current frame number- Standard functions:
sin
,cos
,tan
,log
,exp
,sqrt
Example Expressions:
angle = "0:(0), 60:(sin(t*0.1)*180), 120:(360)"
zoom = "0:(1.0), 120:(1.0 + sin(t*0.05)*0.5)"
translation_x = "0:(sin(t*0.08)*200)"
🎭 Deforum Keyframe Manager
Purpose: Prompt and parameter scheduling
Inputs:
animation_config
: Configurationpositive_prompts
: Keyframed positive promptsnegative_prompts
: Keyframed negative promptsguidance_scale
: CFG scale (1.0-20.0)steps
: Generation steps (1-100)
Prompt Format:
0:(first prompt), 30:(transition prompt), 60:(final prompt)
Example:
Positive: 0:(serene mountain lake), 40:(lake at golden hour), 80:(lake under starry sky)
Negative: 0:(blurry, low quality, distorted)
🎨 Deforum Renderer
Purpose: Main FLUX-based rendering engine
Inputs:
animation_config
: Configurationmotion_params
: Motion datakeyframes
: Prompt schedulingflux_model
: FLUX modelflux_vae
: VAE for encoding/decodingclip
: CLIP for text encodinginit_image
: Optional starting imagedepth_model
: Optional depth model for 3D
Outputs:
images
: Generated animation framesanimation_data
: Metadata and frame info
Features:
- Frame-by-frame generation with motion transformation
- Color coherence across frames
- Preview mode for testing
- Memory-efficient processing
🌊 Deforum Depth Warping
Purpose: 3D depth-aware transformation controller
Inputs:
animation_config
: Configurationdepth_model_name
: midas, adabins, dptnear_plane/far_plane
: 3D camera clipping (200-50000)fov
: Field of view (10-120 degrees)
Features:
- MiDaS depth estimation
- 3D perspective transformations
- Depth map visualization
- Camera parameter control
🎞️ Deforum Video Input
Purpose: Video processing and hybrid composition
Inputs:
animation_config
: Configurationvideo_path
: Input video file pathextract_nth_frame
: Frame extraction ratehybrid_motion
: None, Optical Flow, Perspective, Affinehybrid_flow_method
: DenseRLOF, DIS Medium, Farneback, SF
Features:
- Automatic frame extraction
- Optical flow analysis
- Hybrid video composition
- Multiple flow algorithms
🔄 Deforum Interpolation
Purpose: Frame interpolation and smoothing
Inputs:
animation_data
: From rendererinterpolation_method
: linear, cubic, optical_flowinterpolation_frames
: Frames to insert (1-10)smooth_transitions
: Enable smoothingpreserve_details
: Detail preservation
Features:
- Multiple interpolation algorithms
- Configurable frame insertion
- Smooth transition generation
- Detail preservation options
📹 Deforum Video Output
Purpose: Video compilation and export
Inputs:
images
: Animation framesanimation_config
: Configurationoutput_format
: mp4, gif, webm, movoutput_path
: File save locationfps
: Output frame ratequality
: low, medium, high, losslesscodec
: h264, h265, vp9, proresaudio_path
: Optional audio track
Features:
- Multiple output formats
- Quality control
- Codec selection
- Audio track support
- FFmpeg integration
🎯 Example Workflows
Workflow 1: Simple 2D Animation
AnimationSetup(2D, 60 frames) →
MotionController(rotation + zoom) →
KeyframeManager(prompt transitions) →
Renderer(FLUX generation) →
VideoOutput(MP4)
Workflow 2: 3D Depth Animation
AnimationSetup(3D, 120 frames) →
DepthWarping(MiDaS) →
MotionController(3D movement) →
KeyframeManager(complex prompts) →
Renderer(depth-aware) →
Interpolation(smooth) →
VideoOutput(high quality)
Workflow 3: Hybrid Video Composition
AnimationSetup(Video Input) →
VideoInput(optical flow) →
MotionController(style transfer) →
KeyframeManager(artistic styles) →
Renderer(hybrid composition) →
VideoOutput(artistic video)
⚡ Performance Optimization
Memory Management
- Use lower resolutions for testing (512x512)
- Enable model offloading in ComfyUI settings
- Process long animations in segments
- Clear GPU cache between generations
Speed Optimization
- Use preview mode for workflow testing
- Reduce steps for draft generations
- Use efficient mathematical expressions
- Batch process multiple short animations
Quality Settings
- Draft: 512x512, 20 steps, 2D mode
- Preview: 1024x1024, 30 steps, basic 3D
- Production: 1024x1024+, 40+ steps, full 3D + interpolation
🐛 Troubleshooting
Common Issues
"FFmpeg not found"
Problem: Video output fails with FFmpeg error Solution:
# Windows: Download from https://ffmpeg.org/
# macOS:
brew install ffmpeg
# Linux:
sudo apt install ffmpeg
"Out of memory" errors
Problem: GPU runs out of VRAM during generation Solutions:
- Reduce resolution (1024→512)
- Lower batch size in ComfyUI settings
- Enable model offloading
- Use CPU for some operations
- Close other GPU applications
"Invalid keyframe format"
Problem: Mathematical expressions not parsing correctly Solutions:
- Use format:
0:(value), 10:(value2)
- Ensure parentheses around values
- Check mathematical syntax
- Use
t
for frame variable - Avoid unsupported functions
"No frames generated"
Problem: Renderer produces no output Solutions:
- Check max_frames > 0
- Verify FLUX model is loaded
- Ensure prompts are not empty
- Check motion parameters are valid
- Verify all connections are made
Slow generation speed
Problem: Very slow frame generation Solutions:
- Use preview mode for testing
- Reduce image resolution
- Lower generation steps
- Use 2D mode instead of 3D
- Check GPU utilization
Performance Issues
High memory usage
- Monitor VRAM usage in Task Manager/Activity Monitor
- Use smaller models if available
- Process in smaller batches
- Enable gradient checkpointing
Slow mathematical evaluation
- Simplify complex expressions
- Avoid nested functions
- Use efficient mathematical operations
- Pre-calculate constants
Quality Issues
Inconsistent frames
- Enable color coherence
- Use appropriate strength values (0.6-0.8)
- Ensure smooth motion parameters
- Check keyframe transitions
Artifacts in video
- Increase generation steps
- Use higher quality settings
- Check motion parameter smoothness
- Verify model compatibility
📄 License
MIT License - See LICENSE file for details.
🤝 Contributing
Contributions welcome! Please read CONTRIBUTING.md for guidelines.
🙏 Acknowledgments
- Deforum Team: Original animation framework
- XLabs-AI: FLUX model integration and Deforum-X-Flux research
- ComfyUI Community: Node development platform
- MCP Framework: Automated development system
📞 Support
- Issues: GitHub Issues
- Discussions: GitHub Discussions
- Discord: ComfyUI Discord
Transform your creative vision into stunning animations with mathematical precision and AI power! 🎬✨