ComfyUI-LatentClamp

ComfyUI-LatentClamp
★ 2

潜变量处理数值钳制img2img优化噪声补偿
对 samples 中超出 min 与 max 的潜变量按 outside_multiplier 缩放,并可对区间内值与额外噪声做实验性调整,帮助控制极值、提升 img2img 的提示遵从性。
💡 在 img2img 流程中钳制潜空间值以增强提示遵从性。
🍴 2 Forks💻 Python🔄 2025-04-15
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/a9fb3a59e10c
📄 README

ComfyUI-LatentClamp

A node for ComfyUI that modifies the values in the samples input that fall outside of a min and max range using a multiplier.

Additional controls are available for experimental purposes.

In img2img operations, latent clamping has been observed to improve prompt adherence. Issues with image fidelity resulting from the clamp can be rectified with a quick second diffusion pass.

Inputs

  • min: Values in samples below this value will be multiplied by outside_multiplier.
  • max: Values in samples above this value will be multiplied by outside_multiplier.
  • inside_multiplier: Values in samples between min and max will be multiplied by this value.
  • outside_multiplier: Values in samples outside of min and max will be multiplied by this value.
  • extra_noise: Add compensatory noise to the values in samples.