Voronoi
Voronoi
Voronoi divides the image into seeded cells. Distance keeps the original smooth field. Cell walls produces a binary crack mask, Flat cells assigns one seeded value per region, and Sampled cells takes one RGB color from the optional Image input at each cell’s feature point.
Availability
This node requires Pro.
Common uses
- Use Cell walls and Wall width to make adjustable cracks directly.
- Use Flat cells as a clear input for Edge Detect or Flood Fill.
- Connect a color image to Sampled cells to assign one color per cell.
Input ports
-
image -
Image is optional and used by Sampled cells; other modes ignore it.
Output ports
-
out -
Distance retains the existing grayscale or multichannel result. Cell walls is grayscale; Flat cells can be grayscale or multichannel; Sampled cells is RGB.
Parameters
| Parameter | Type | Default | Range or values |
|---|---|---|---|
| Output | enum | distance | distance, walls, flat, sampled |
| Wall Width | float | 0.06 | 0.005 to 0.3 |
| Frequency | int | 10 | 1 to 128 |
| Randomness | float | 1 | 0 to 1 |
| Seed | int | 12345 | 0 to 999999 |
| Disorder | float | 0 | 0 to 1 |
| Distance | enum | euclidean | euclidean, manhattan, chebyshev |
| Invert | bool | false | - |
| Multichannel | bool | false | - |
| Non-Square Expansion | bool | true | - |
Behavior
- Noise nodes are deterministic for a given seed and parameter set.
- Different noise models create different value distributions and edge behavior.
- Disorder animates the underlying cell layout on a perfect loop, so 1.0 returns to the same result as 0.0.
Common mistakes
- Very high frequency values can alias at lower output resolutions.
- Sampled cells uses seeded colors when Image is disconnected.
- GPU rendering is available when supported; the CPU path remains the fallback.