📐 Qwen Image Empty Latent Size Picker for ComfyUI

A ComfyUI node that creates empty latents with preset sizes for Qwen-Image, Z-Image, SDXL, Flux, and Flux2 models. Includes common resolutions for each model and allows custom size overrides.

Features

Two Node Types:

📐 Qwen Image Size Picker (Original)

• Pre-configured resolution lists for each model (Qwen-Image, Z-Image, SDXL, Flux, Flux2)
• Select specific resolutions from comprehensive dropdowns
• Custom size overrides (width/height)
• Batch size control
• Outputs: LATENT, width, height, batch_size

📐 Qwen Image Size Picker (Simple)

• Model type selection (Qwen-Image, Z-Image, SDXL, Flux, Flux2)
• Aspect ratio selection (1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3)
• Automatically selects the optimal resolution for your chosen model and aspect ratio
• Custom size overrides (width/height)
• Batch size control
• Outputs: LATENT, width, height, batch_size

Supported Models:

| Model | Description | Resolution Range | |-------|-------------|------------------| | Qwen-Image | Alibaba's Qwen image generation model | 928×1664 to 1664×928 | | Z-Image | Tongyi-MAI's flexible aspect ratio model | 720×1280 to 1600×900 | | SDXL | Stable Diffusion XL | 704×1408 to 1728×576 | | Flux | Black Forest Labs Flux | 512×512 to 1536×1536 | | Flux2 | Flux.1 Pro/Dev (2x SDXL) | 1408×2816 to 3456×1152 |

Installation

1. Clone into your ComfyUI custom_nodes folder:

cd ComfyUI/custom_nodes
git clone https://github.com/rzgarespo/ComfyUI-qwen-image-size-picker.git

2. Restart ComfyUI

Usage

Original Node

1. Add 📐 Qwen Image Size Picker to your workflow
2. Select your model type (e.g., Qwen-Image, SDXL, Flux)
3. Choose a specific resolution from the model's dropdown list
4. Adjust batch size if needed
5. Use width/height override to customize (optional)

Simple Node

1. Add 📐 Qwen Image Size Picker (Simple) to your workflow
2. Select your model type
3. Select an aspect ratio (e.g., 16:9 for landscape, 9:16 for portrait)
4. The node automatically picks the optimal resolution for your model
5. Adjust batch size if needed
6. Use width/height override to customize (optional)

Outputs

All nodes provide four outputs:

• LATENT - Empty latent tensor for image generation
• width - Width in pixels (INT)
• height - Height in pixels (INT)
• batch_size - Number of images (INT)

Z-Image Note

Z-Image is flexible with aspect ratios. The resolutions used are optimized for 1-2 megapixel range, staying within the official 512-2048 pixel guideline. Common aspect ratios are prioritized based on official examples and community recommendations.

License

MIT License