Flex Mask Voronoi Scheduled ⚡🅡🅞🅣🅘

Voronoi diagrams - the cracked-glass, cell-boundary pattern you get from scattering points and coloring each pixel by which point it's closest to - are a classic procedural-texture building block, and this is the most parameter-dense node in the pack's mask family. It's not modifying your input mask so much as generating a whole cellular pattern from scratch, with nearly every knob available for reactive control.

How it works

scale, detail (cell count), randomness, and seed control the base cell layout - more detail means more, smaller cells; higher randomness moves cell centers away from a perfectly regular grid toward something organic; seed picks which specific random layout you get. x_offset/y_offset pan the whole pattern. distance_metric is where this node gets genuinely unusual: alongside the standard euclidean, manhattan, chebyshev, and minkowski distance functions, it also ships five decorative "kaleidoscope" metrics (kaleidoscope_star, kaleidoscope_wave, and three kaleidoscope_radiation variants) that warp the cell shapes into starburst and radial patterns instead of plain polygons - someone clearly had fun with this one, and it's worth actually trying a few before settling on euclidean out of habit.

The formula/a/b trio is specific to this node: formula (Linear, Quadratic, Cubic, Sinusoidal, Exponential) shapes the curve used to map your feature's value onto whichever parameter feature_param is targeting, with a and b as the two tunable constants for that curve. That gives you seven things a FEATURE can drive - scale, detail, randomness, seed, x_offset, y_offset, or none - each shaped by a non-linear response curve instead of a flat linear mapping, which is more control than any other node in this batch exposes.

Inputs and outputs that matter

• masks (required, MASK) - the source mask this pattern is generated against.
• distance_metric - 10 options, from standard Euclidean/Manhattan through the kaleidoscope variants.
• scale, detail, randomness, seed, x_offset, y_offset - the pattern's layout controls.
• formula, a, b - the response curve shaping any feature modulation.
• opt_feature (optional, FEATURE) - drives whichever parameter feature_param selects.
• Output - a single MASK.

Installing it

Via ComfyUI Manager: search "RyanOnTheInside," install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/ryanontheinside/ComfyUI_RyanOnTheInside.git
cd ComfyUI_RyanOnTheInside
pip install -r requirements.txt

then restart. Procedural pattern generation - no model downloads.

Common issues & troubleshooting

Pattern looks like noise, not clean cells. High randomness values (up to 5.0) push cell placement well past "organic" into genuinely chaotic - if you wanted a recognizable Voronoi look rather than static, pull it back toward 1.0 or lower.

Kaleidoscope metrics produce unexpected shapes. That's expected - they deliberately distort cell boundaries away from plain polygons into star/wave/radial forms. If you want a standard cracked-glass look, stick with euclidean; reach for the kaleidoscope options when you specifically want that decorative distortion.

Feature modulation feels too extreme or too subtle. Before touching strength or feature_threshold, check formula/a/b - an Exponential or Cubic curve with aggressive a/b values will make small feature changes swing a parameter hard near one end of its range, which can look "broken" when it's actually the curve shape doing exactly what you set it to do.