Painting-style colour reduction

A two-pass local average establishes broad colour regions. Each channel is then rounded to a chosen number of levels, while a luminance gradient controls how much strong boundaries are darkened. The result resembles a simplified painted study because it suppresses fine texture and preserves large transitions.

Brush radius

The neighbourhood averaged around each pixel.

Colour levels

The allowed steps remaining in each channel.

Edge weight

How strongly measured luminance gradients darken.

A local average is not an artist's brush. At a high radius it crosses a face edge and muddies it; at a low radius the photograph's texture remains and the result looks merely posterised. Edge weight can make hair and foliage turn into dark noise when those regions contain many gradients.

The preview, draft and export all run the same kernel. Parameters measured in pixels are scaled with the source, so a chosen tile, band or radius does not become unexpectedly finer in the downloaded file.

Is this an AI painting generator?

No. It is deterministic pixel arithmetic: separable local averaging, channel quantisation and gradient-based edge weighting. It invents no objects or brush strokes.

Why do small details disappear?

The brush radius averages a neighbourhood before colours are reduced. Detail smaller than that neighbourhood becomes part of its surrounding region.

What does edge weight darken?

It measures the local luminance gradient and reduces colour near strong transitions. It is a contour emphasis, not a semantic outline.

• PosteriseContinuous tone reduced to measured channel levels, with optional pre-dither.
• MosaicMean-colour tiles with editable grout width and colour.
• DuotoneTwo colours and the ramp between them, rebuilt on every change to either end.
• The effects labAll six effects, chosen from your own photograph rather than from a list of names.

Rendered on your computer, not on a server.