Files
1hew-ComfyUI-1hewNodes/web/docs/ImageHLFreqTransform/en.md
T

2.5 KiB

Image HL Freq Transform

Node Function: The Image HL Freq Transform node implements advanced detail transfer technology, capable of transferring high-frequency information from detail images to generated images, with mask control and multiple frequency separation method support.

Inputs

Parameter Required Data Type Default Range Description
generate_image Required IMAGE - - Generated base image
detail_image Required IMAGE - - Detail reference image
method - COMBO[STRING] igbi rgb, hsv, igbi Frequency separation method
blur_radius - FLOAT 10.0 0.0-100.0 Gaussian blur radius
detail_mask Optional MASK - - Detail area mask, controls transfer regions

Outputs

Output Name Data Type Description
image IMAGE Final image after detail transfer
high_freq IMAGE Blended high-frequency layer
low_freq IMAGE Low-frequency layer of generated image

Features

Transfer Principles Details

RGB/HSV Mode

  • Processing flow:
    1. Perform frequency separation on detail image to get high-frequency layer
    2. Perform frequency separation on generated image to get high and low frequency layers
    3. Blend high-frequency layers through mask
    4. Recombine final image
  • Blur radius adjustment: Generated image uses 1.5x blur radius for better results
  • Characteristics: Traditional method, suitable for general detail transfer needs
  • Low-frequency layer: Gaussian blur result of the generated image
  • High-frequency layer calculation flow:
    1. Generated image processing: Invert → Gaussian blur → 50% blend with blurred image → Invert again
    2. Detail image processing: Invert → Gaussian blur → 50% blend with blurred image → Invert again
    3. Mask blending: Blend the two processed results through mask
  • Final composition: 65% high-frequency + 35% low-frequency, then apply levels adjustment (black point 83, white point 172)
  • Characteristics: Provides the most precise and natural detail transfer effect

Mask Control Mechanism

  • With Mask:
    • Detail transfer only in white areas of the mask
    • Black areas preserve original details of generated image
    • Supports grayscale masks for gradient transitions
  • Without Mask:
    • Detail transfer across the entire image
    • Automatically creates full white mask