ComfyUI Extension: ComfyUI-PaletteCanvas
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.
Control lighting in AI-generated images through color gradients and shapes.
Looking for a different extension?
Custom Nodes (0)
README
Palette Canvas
Control lighting in AI-generated images through color gradients and shapes.
<img width="450" height="374" alt="image" src="https://github.com/user-attachments/assets/95160a67-db3a-4487-b35a-5129822bdea7" />This ComfyUI custom node creates color canvases that serve as lighting maps for text-to-image generation workflows. By encoding these gradients through VAE and using them as initial latents with lowered denoise (0.87-0.95 instead of 1), you can precisely control the mood, atmosphere, and illumination of your generated images.
The Concept
Traditional text-to-image generation gives you limited control over lighting. This node solves that by letting you:
- Paint with Light - Create custom gradients that define how light flows across your scene
- Add Light Sources - Place bright shapes on dark backgrounds to create specific illumination points
- Guide the AI - Feed these light maps into your sampler to influence the final lighting without complex prompting
Use Cases
Atmospheric Lighting
- Create sunset gradients (orange → purple) for golden hour lighting
- Design dark-to-light gradients for cinematic rim lighting
- Build complex multi-color gradients for fantasy scenes
Specific Light Sources
- Place a bright circle on dark background → creates a spotlight effect
- Use rectangle shapes as window light sources
- Position triangles for dramatic angular lighting
Mood Control
- Cool gradients (blues) for night/calm scenes
- Warm gradients (reds/oranges) for energetic/warm scenes
- High contrast gradients for dramatic noir lighting
Installation
Method 1: ComfyUI Manager (Recommended)
- Open ComfyUI Manager
- Click "Install Custom Nodes"
- Search for "Palette Canvas"
- Click Install
Method 2: Git Clone
cd ComfyUI/custom_nodes git clone https://github.com/yourusername/ComfyUI-PaletteCanvasRGB.git
Method 3: Manual
- Download this repository as ZIP
- Extract to
ComfyUI/custom_nodes/ComfyUI-PaletteCanvasRGB/ - Restart ComfyUI
Technical Details
Why Denoise 0.87-0.95?
- 0.87-0.92: Gradient strongly influences the final image composition
- 0.93-0.95: Balanced - AI follows the lighting but creates natural details
- 0.96-0.99: Subtle influence - AI uses gradient as "mood suggestion"
Aspect Ratio Handling
The node automatically corrects for non-square aspect ratios. A circle remains circular whether your canvas is 512×512 or 896×1152.
Shape Distortion (scale_x/y)
Use independent X/Y scaling to create:
- Elliptical lights (stretched circle = wide light source)
- Anamorphic flares (compressed height = cinematic lens flares)
- Custom light patterns (rectangle stretched into line = neon strip)
Requirements
- ComfyUI (any recent version)
- Modern browser (for interactive widgets)
No additional Python packages required.
License
MIT License - See LICENSE file
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.