Files
facok-ComfyUI-DiversityBoost/pyproject.toml
T
facok 951bfa6526 Initial release: frequency-domain composition diversity for distilled models
Injects seed-dependent low-frequency phase from initial noise into the
model's denoised prediction at step 0 via post-CFG hook.  Different seeds
produce different compositions instead of identical layouts.

Key features:
- Slerp phase rotation with shared rotation field (zero color artifacts)
- 6th-order Butterworth LPF (steep composition/object frequency separation)
- Per-bin tanh rotation budget (decouples diversity from tail risk)
- High-frequency attenuation (prevents frequency shearing on distilled models)
- Optional energy compensation
- LTXAV video format compatible
2026-04-14 04:12:53 +08:00

10 lines
308 B
TOML

[project]
name = "comfyui-diversityboost"
description = "Restore composition diversity for distilled diffusion models. Training-free frequency-domain phase injection."
version = "1.0.0"
license = "MIT"
requires-python = ">=3.10"
[project.urls]
Repository = "https://github.com/facok/ComfyUI-DiversityBoost"