ComfyUI Image Size Tool

A professional resolution calculator node pack for ComfyUI that provides model-specific, constraint-aware image dimensions for optimal AI generation quality.

Why This Tool Exists

Most existing resolution tools either:

• Force you to manually enter width/height (error-prone)
• Ignore model-specific training constraints
• Provide arbitrary resolutions that hurt generation quality
• Break your ComfyUI installation with complex dependencies

This tool solves the resolution calculation problem by providing curated, model-optimized presets based on actual training specifications and bucket systems.

Features

✅ Zero Dependencies - No external libraries, no conflicts, no breakage
✅ Model-Aware - Respects VAE constraints (div by 8/64) and training buckets
✅ Quality Focused - Uses actual SDXL bucket resolutions, not arbitrary calculations
✅ Plug & Play - Simple dropdowns, clean outputs, works immediately
✅ Professional - Semantic versioning, proper documentation, stable API

Included Nodes

SD 1.5 Resolution

• 8 optimized presets from 512×512 to 912×512
• Divisible by 8 constraint validation (VAE requirement)
• Outputs: width, height, aspect_ratio

SDXL Resolution

• 13 bucket-trained presets targeting ~1 megapixel
• Divisible by 64 constraint validation
• Based on the 43-bucket training system for optimal quality
• Outputs: width, height, aspect_ratio, total_pixels

Flux.1 Dev Resolution

• 11 practical presets from 1024×1024 to 2560×1440
• No strict constraints - more flexible than SD models
• Hardware-optimized for common use cases
• Outputs: width, height, aspect_ratio

Installation

Method 1: Git Clone (Recommended)

cd ComfyUI/custom_nodes
git clone https://github.com/TheLustriVA/ComfyUI-Image-Size-Tool.git

Method 2: ComfyUI Manager

1. Open ComfyUI Manager
2. Install from Git URL: https://github.com/TheLustriVA/ComfyUI-Image-Size-Tool

No additional dependencies to install - just restart ComfyUI.

Usage

1. Add Node: Search for "Image Size Tool" in the node menu
2. Select Resolution: Choose from the dropdown (e.g., "1920×1080 (16:9) - Full HD")
3. Connect Outputs: Wire width and height to your Empty Latent Image node
4. Generate: Enjoy optimal quality with proper model constraints

Example Workflow

[Flux.1 Dev Resolution] → [Empty Latent Image] → [KSampler] → [VAE Decode] → [Save Image]
     ↓ width=1920
     ↓ height=1080

Why Model-Specific Resolutions Matter

SD 1.5

• Trained at 512×512 - quality degrades significantly when deviating
• VAE constraint - dimensions must be divisible by 8 or generation fails

SDXL

• Bucket training system - uses 43 specific resolutions targeting 1MP
• Quality buckets - arbitrary resolutions produce inferior results
• VAE constraint - dimensions must be divisible by 64

Flux.1 Dev

• Flexible architecture - handles 0.2 to 2 megapixels effectively
• Hardware optimized - common resolutions for practical generation

Troubleshooting

Nodes Don't Appear

cd ComfyUI/custom_nodes/ComfyUI-Image-Size-Tool
git pull
find . -name "__pycache__" -exec rm -rf {} +

Then restart ComfyUI.

Import Errors (If Any)

This tool uses zero external dependencies and a single-file architecture specifically to avoid import issues. If you encounter problems, please open an issue.

Technical Details

• Architecture: Single-file design with embedded constants
• Dependencies: None (Python stdlib only)
• Compatibility: ComfyUI (all recent versions)
• Performance: Instant loading, zero overhead

Contributing

Issues and pull requests welcome! This tool aims to be the definitive resolution calculator for ComfyUI.

License

MIT License - see LICENSE file for details.

---

Generated with professional AI development practices
No more resolution guesswork - just optimal image generation