Type: huggingface.image_to_image.StableDiffusionLatentUpscaler
Namespace: huggingface.image_to_image
Description
Upscales Stable Diffusion latents (x2) using the SD Latent Upscaler pipeline. tensor, upscaling, stable-diffusion, latent, SD
Input and output are tensors for chaining with latent-based workflows.
Properties
| Property | Type | Description | Default |
|---|---|---|---|
| prompt | str |
The prompt for upscaling guidance. | `` |
| negative_prompt | str |
The negative prompt to guide what should not appear in the result. | `` |
| num_inference_steps | int |
Number of upscaling denoising steps. | 10 |
| guidance_scale | float |
Guidance scale for upscaling. 0 preserves content strongly. | 0.0 |
| seed | int |
Seed for the random number generator. Use -1 for a random seed. | -1 |
| latents | torch_tensor |
Low-resolution latents tensor to upscale. | {'type': 'torch_tensor', 'value': None, 'dtype': '<i8', 'shape': [1]} |
Outputs
| Output | Type | Description |
|---|---|---|
| output | torch_tensor |
Metadata
Related Nodes
Browse other nodes in the huggingface.image_to_image namespace.