Inputs
Note: The schedule is computed from the image sequence length, which is derived from
width and height as (width * height) / 256, reflecting the model’s 16x latent downsampling. Larger images produce longer sequences, which shifts the noise schedule accordingly.
Outputs
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
Source fingerprint (SHA-256):
9606177f37f7bc03aef524623f03b7f24bcdc3d9327dcdf74863fe2befeb2b65