Documentation Index
Fetch the complete documentation index at: https://docs.comfy.org/llms.txt
Use this file to discover all available pages before exploring further.
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
Overview
This node creates an empty latent image in pixel space, specifically designed for the HiDream-O1-Image model. It generates a blank tensor of zeros that serves as the starting point for image generation, with dimensions defined by the width, height, and batch size inputs.Inputs
| Parameter | Data Type | Required | Range | Description |
|---|---|---|---|---|
width | INT | Yes | 64 to 4096 (step: 32) | The width of the latent image in pixels (default: 2048). The model was trained at ~4 megapixels; lower resolutions go off-distribution and quality regresses noticeably. |
height | INT | Yes | 64 to 4096 (step: 32) | The height of the latent image in pixels (default: 2048). The model was trained at ~4 megapixels; lower resolutions go off-distribution and quality regresses noticeably. |
batch_size | INT | No | 1 to 64 | The number of latent images to generate in a single batch (default: 1). |
Outputs
| Output Name | Data Type | Description |
|---|---|---|
samples | LATENT | A tensor filled with zeros representing the empty latent image, with shape (batch_size, 3, height, width). |
Notes
- The HiDream-O1-Image model was trained at approximately 4 megapixels. Using significantly lower resolutions may result in reduced image quality.
- Trained resolutions include: 2048x2048, 2304x1728, 1728x2304, 2560x1440, 1440x2560, 2496x1664, 1664x2496, 3104x1312, 1312x3104, 2304x1792, 1792x2304.
Source fingerprint (SHA-256):
fca32bbeddf120b4a7f9a9b88814f5345db133b35252c4d86079397be350c15e