Houdini 21.0 Nodes Geometry nodes

Labs Spectral Feature Extract 1.0 geometry node

Extracts frequency-based features from scalar or vector attributes.

Spectral feature extraction separates an input signal into components based on spatial frequency. Low-frequency components represent broad, smooth variations, while high-frequency components capture fine detail and noise.

This node computes these components using either a diffusion partial differential equation (PDE) approach or a smoothing-based approximation. The extracted results can be written as float or vector attributes and optionally visualized for debugging and look development.

Parameters

Input Attribute

The attribute to analyze.

Input Type

The data type of the input attribute.

Vector Input

Extract

Defines how vector frequency content is extracted.

1D Frequencies in Projection Directions: Projects vectors onto a direction before extracting frequencies.

3D Frequencies: Extracts frequency content independently in all three vector components.

Projection Direction

Attribute defining the projection direction when using 1D extraction.

Output Float Attribute

Name of the output attribute when extracting scalar results.

Output Vector Attribute

Name of the output attribute when extracting vector results.

Frequency Spectrum Extraction

Method

The algorithm used to extract frequency components.

Diffusion (PDE): Uses a diffusion process to separate frequency bands.

Smooth: Uses smoothing operations as an approximation

Diffusion PDE

Diffusion Times

Higher-End Time

Diffusion time used to extract high-frequency components.

Lower-End Time

Diffusion time used to extract low-frequency components.

Diffusion Speed Attributes

Higher-End Attribute

Attribute controlling diffusion speed for high-frequency extraction.

Lower-End Attribute

Attribute controlling diffusion speed for low-frequency extraction.

Smooth

Smooth Strengths

Higher-End Strength

Smoothing strength used to extract high-frequency components.

Lower-End Strength

Smoothing strength used to extract low-frequency components.

Smooth Filter Qualities

Higher-End Filter

Filter quality used for high-frequency smoothing.

Lower-End Filter

Filter quality used for low-frequency smoothing.

Smooth Models

Higher-End Model

Model used to guide high-frequency smoothing.

Uniform: Applies uniform smoothing.

Scale-Dominant: Biases smoothing based on scale.

Curvature-Dominant: Biases smoothing based on curvature.

Lower-End Model

Model used to guide low-frequency smoothing.

Uniform: Applies uniform smoothing.

Scale-Dominant: Biases smoothing based on scale.

Curvature-Dominant: Biases smoothing based on curvature.

Advanced

Time Scale

Global scale applied to diffusion times.

Strength Scale

Global scale applied to smoothing strengths.

Invert Extraction Direction

Swaps the interpretation of high- and low-frequency extraction.

Constrained Points

Constrain Boundary Points

Constrains boundary points during spectral extraction.

Higher-End Points

Point group defining constraints for high-frequency extraction.

Lower-End Points

Point group defining constraints for low-frequency extraction.

Visualization

Recompute Output Normals

Recomputes normals after extraction for visualization purposes.

Visualize Extracted Spectrum

Enables visualization of the extracted frequency components.

Visualization Controls

Float Color Gradient

Ramp used to visualize scalar output values.

Vector Color Gradient

Controls how vector values are visualized.

Scaled Original Vector Components: Colors based on scaled original vectors.

Remapped Vector Components: Colors based on remapped extracted vectors.

Visualized Vector Scale

Scale applied to vectors during visualization.

Geometry nodes