Files

T

Vito Sansevero ab23992c29 chore: project housekeeping and configuration updates

- Add code quality tools: flake8, mypy, black, pre-commit
- Add .gitattributes for line ending consistency
- Add .secrets.baseline for secret scanning
- Update GitHub workflows for better CI/CD
- Update documentation formatting and examples
- Add CLAUDE.md for AI assistant guidance
- Add scripts directory for automation tools
- Update project configuration in pyproject.toml
- Improve type hints and code formatting across all modules
- Update test configurations and fixtures

2025-08-01 08:35:08 -07:00

3.1 KiB

Raw Blame History

Resolution Calculator Tool

Overview

The Resolution Calculator is a ComfyUI node that calculates upscaled dimensions from image or latent inputs with proper scaling factors optimized for SDXL and FLUX models.

Inputs

Required

scale_factor (FLOAT): Scaling factor from 1.0 to 8.0
- Default: 2.0
- Common values: 1.2, 1.5, 2.0, 2.5, 3.0, 4.0
- Optimized for typical upscaling workflows

Optional (one required)

image (IMAGE): Input image tensor to calculate dimensions from
latent (LATENT): Input latent tensor to calculate dimensions from

Outputs

width (INT): Calculated target width
height (INT): Calculated target height

Features

Model Optimization

SDXL: Optimized for ~1 megapixel base resolution (1024×1024)
FLUX: Supports flexible resolutions from 0.2-2 megapixels

Automatic Constraints

Ensures all outputs are divisible by 8 (ComfyUI requirement)
Preserves aspect ratio during scaling
Rounds to nearest valid dimensions

Use Cases

1. Upscaler Preparation

Connect the width/height outputs to upscaler nodes to ensure proper target dimensions.

Image Loader → Resolution Calculator → Upscaler
             ↘ scale_factor: 2.0    ↗

2. Latent Upscaling

Calculate target dimensions from generated latents before upscaling.

VAE Encode → Resolution Calculator → Latent Upscaler
           ↘ scale_factor: 1.5    ↗

3. Memory Planning

Preview final dimensions to estimate memory usage before expensive operations.

Example Scenarios

SDXL Portrait Upscaling

Input: 832×1216 (SDXL portrait format)
Scale: 1.5x
Output: 1248×1824
Use: Perfect for 1.5x upscaling while maintaining SDXL ratios

FLUX Square Generation

Input: 1024×1024 (FLUX square format)
Scale: 2.0x
Output: 2048×2048
Use: Standard 2x upscaling for square outputs

Variable Scale Factors

The node supports any scale factor from 1.0 to 8.0, allowing for:

1.2x: Subtle enhancement
1.5x: Moderate upscaling
2.0x: Standard doubling
4.0x: Aggressive upscaling

Technical Details

Dimension Extraction

IMAGE tensors: [batch, height, width, channels] format
LATENT tensors: {"samples": [batch, channels, height/8, width/8]} format
Automatic detection and conversion between formats

Calculation Process

Extract original dimensions from input tensor
Apply scale factor: new_size = original_size * scale_factor
Round to nearest multiple of 8: ((size + 4) // 8) * 8
Return integer width and height

Error Handling

Validates that at least one input (image or latent) is provided
Ensures scale factor is within valid range (1.0-8.0)
Graceful error messages for invalid inputs

Integration

The Resolution Calculator integrates seamlessly with:

Standard ComfyUI image loaders
VAE encode/decode operations
Upscaler nodes (ESRGAN, Real-ESRGAN, etc.)
Custom latent processing workflows

3.1 KiB Raw Blame History Unescape Escape