This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHubThe StableCascade_EmptyLatentImage node creates empty latent tensors for Stable Cascade models. It generates two separate latent representations - one for stage C and another for stage B - with appropriate dimensions based on the input resolution and compression settings. This node provides the starting point for the Stable Cascade generation pipeline.
Inputs
| Parameter | Data Type | Required | Range | Description |
|---|---|---|---|---|
width | INT | Yes | 256 to MAX_RESOLUTION | The width of the output image in pixels (default: 1024, step: 8) |
height | INT | Yes | 256 to MAX_RESOLUTION | The height of the output image in pixels (default: 1024, step: 8) |
compression | INT | Yes | 4 to 128 | The compression factor that determines the latent dimensions for stage C (default: 42, step: 1) |
batch_size | INT | No | 1 to 4096 | The number of latent samples to generate in a batch (default: 1) |
Outputs
| Output Name | Data Type | Description |
|---|---|---|
stage_c | LATENT | The stage C latent tensor with dimensions [batch_size, 16, height//compression, width//compression] |
stage_b | LATENT | The stage B latent tensor with dimensions [batch_size, 4, height//4, width//4] |