ComfyUI Aspect Ratio Presets

Two custom ComfyUI nodes to kick-start your latent workflows:

• Empty Latent (Aspect Ratio Preset) Choose from a curated list of popular model resolutions and aspect ratios.

• Empty Latent (Aspect Ratio by Axis) Specify one dimension and an X:Y ratio; the node computes the other side automatically.

---

Features

• Preset Mode: Dropdown of Standard Diffusion, SDXL, Flux/HiD, HiDream resolutions.
• Axis Mode: Enter a “primary dimension” (width or height) plus any common aspect ratio (1:1, 3:2, 16:9, 21:9, etc.).
• Validation: Enforces positive values divisible by 8 to match VAE down-sampling.
• Batch Support: Produce multiple blank latents in one go.
• Tooltips: In-UI guidance for each parameter.

---

Installation

1. Clone into your ComfyUI custom_nodes/ directory:

   git clone https://github.com/yourusername/comfyui-aspect-ratio-presets.git
   

2. (Optional) Install dependencies:

   pip install -r custom_nodes/comfyui-aspect-ratio-presets/requirements.txt
   

3. Restart ComfyUI. The new nodes appear under ComfyUI Aspect Ratio Preset.

---

Usage

1. Empty Latent (Aspect Ratio Preset)

   • Open the node and pick a resolution like SDXL 1152×768 (3:2 Landscape).
   • Set your batch size (default 1).
   • Connect the output into your generation graph.

2. Empty Latent (Aspect Ratio by Axis)

   • Enter a primary value (e.g. 1024) and choose whether that’s your Width or Height.
   • Select an aspect ratio label (e.g. 16:9 Landscape).
   • Adjust batch size if needed.
   • Use the output tensor as a blank latent canvas.

---

Presets

All built-in presets live in presets.py and follow the format:

(Model name, "X:Y Orientation", width, height)

Feel free to extend or override this list in your fork.

---

License

This project is released under the MIT License. Feel free to use, modify, and redistribute!