Image To Grid

Create printable mixed-from-paints grids, closest-from-palette maps, and pixel-style artwork

Click to upload or drag & drop

Supports: JPG, PNG, WEBP, BMP, GIF, HEIC, HEIF, ITG (project)

Color Method

Color Set

Output Format

Grid Detail: 60

Quantization Block Size: 1

Averages NxN pixel blocks before choosing a Color Set colour. Larger values produce fewer, bigger colour regions. 1 = per-pixel.

Advanced Processing

Interface Style

Changes the complete interface layout. Studio Workspace keeps the light/dark icon in the banner.

Show help icons

Shows or hides the small ? buttons that open contextual help.

Always start guided tour

Debug helper: starts the guided tour on every page load until this option is turned off.

Recipe display mode

Controls how paint mixing recipes are displayed in the legend: as parts, percentages, or both.

Recipe accuracy

Controls how exhaustively mixed-from-paints recipes search for the closest mix. Fast is the current default for quick experimentation; Balanced matches the previous solver behavior; High is slower but tests more paint combinations and ratios.

Enable Triangular Formats

Enables the "Triangular" and "Triangular Region" output formats. When unchecked, the options are hidden and no triangular processing is performed.

Enable Color Contour Map

Enables the "Color Contour Map" output format. When unchecked, the option is hidden and no contour processing is performed.

Show Quantization Block Size

Shows the Quantization Block Size slider in the main settings when Color Contour Map is selected.

Merge tiny contour colours

Joins tiny Color Contour Map colours to a neighbouring colour instead of hiding them.

Merge criterion

Hide tiny contour colours

Used only when tiny-colour merging is disabled. Hides Color Contour Map colours whose visible area is below the threshold.

Tiny colour threshold (%)

Color Contour Map colours below this visible area are considered tiny.

Contour border smoothing

Smooths jagged colour boundaries in the contour map with a spatial majority filter.

Smoothing radius

Neighbourhood radius for the majority filter (1–4). Higher values produce smoother boundaries but may lose fine detail.

Smoothing passes

Number of majority-filter passes. More passes further smooth boundaries.

Suppress weak contour boundaries

Hides contour boundaries when the original image gradient is weak and the two neighbouring Color Set colours are very similar. Useful for smooth backgrounds such as walls, skies, or skin shading.

Remove small contour islands

Merges tiny isolated contour regions into the surrounding colour independently from contour smoothing.

Island max size (px)

Connected colour regions up to this size are merged into the surrounding colour. 0 = disabled.

Island connectivity

Controls whether diagonal pixels count as the same island.

Merge target

Chooses which neighbouring colour receives a small island.

Preserve high-contrast details

Keeps deliberate small marks when their colour is very different from the merge target.

Detail contrast threshold

Higher values preserve fewer details.

Smooth contour lines

Draws contour boundaries as smooth anti-aliased curves (Chaikin subdivision) instead of jagged pixel-level dots.

Smoothing method

Choose how contour polylines are converted into smooth curves.

Chaikin iterations

Higher values produce rounder contour curves, but can soften small corners and thin details.

Post-process mode

Raw keeps pixel staircase. Simplified reduces points then fits an interpolating spline. Approximation smooths the average path without passing through every pixel step. No colours, Color Set entries, or regions are changed.

Simplification tolerance

RDP epsilon. Higher removes more staircase points before the spline is fit.

Target point ratio

Cap on remaining points as a fraction of the original contour (e.g. 0.3 = at most 30%).

Smoothing strength

0 = barely smoothed, 1 = strongly smoothed average path (does not pass through every pixel step).

Simplify before smoothing

Reduces approximation input points before averaging the contour path.

Approx. simplification tolerance

Higher values remove more input staircase points before approximation smoothing.

Approx. target point ratio

Cap on approximation input points as a fraction of the original contour.

Min contour length

Contour fragments shorter than this (in pixel-edge units) are handled by the small-contour policy.

Small contour policy

How tiny contour fragments are rendered.

Closed contour handling

Override the detected open/closed state of each contour fragment.

Minimum region area

Borders touching smaller regions stay raw so paintable details remain readable.

Export mode

Original exports the contour as shown on screen. Preserve Shape and Vector-Like apply geometric re-smoothing to the final label-map contours; Vector-Like produces cleaner, more flowing outlines.

Export format

Choose the file format for downloads and printing. SVG produces scalable vector files. PNG produces high-resolution raster images that render faster in print preview.

Label placement

How large contour zones are split for label placement. Grid: regular rectangular cells. K-Means: spatial clustering. Bisection: recursive halving along the longest axis.

Show original vs closest

Hover color inspector

Enable this option, then move the mouse over a cell in the visual preview to see both the original extracted color and the closest Color Set color in RGB and HSV.

Show white-cell debug overlay

Shows the detection result directly in the visual preview: transparent = blue, outside white = red, inside white = green.

Compute inside/outside white cells

Classifies white cells as exterior or interior. Exterior white cells are converted to transparent; interior white cells can then be shown or hidden separately.

White detection method

White detection percentile

White-region connectivity

Fine white detection

Uses a finer hidden grid derived from the original image to classify white cells before projecting the result back to the user grid.

Fine scale

Interior subcell threshold (%)

Detection time: n/a

Show interior white cells

When inside/outside computation is enabled, enclosed white cells inside the drawing can stay visible or be hidden on demand. Exterior white cells remain transparent.

White-edge erosion

After processing, white cells adjacent to transparent cells are converted to transparent, removing stray white edge artifacts around the subject boundary.

Iterative erosion Near-white threshold

When enabled, the erosion repeats until no white cells with transparent neighbours remain. When disabled, only a single pass is applied. The near-white threshold sets how far from pure white a colour can be (Euclidean RGB distance) and still be treated as white for erosion (0 = exact white only).

Unanimous neighbour recolor

Cells whose neighbours agree on a single colour (or are all transparent) are converted to match. Min neighbours = how many must agree (6 = all hex neighbours).

Min neighbours

Iterative recolor

When iterative is enabled, the recolor repeats until no more cells change.

Cell Color Rule

Quantize Percentile

Grid Width

Grid Height

Resample Kernel

Quantization Mode

Color clustering method

Number of color clusters

Shadow / tone preservation

Reserves clustered color slots for meaningful darker tone bands inside large color families, preserving shadows without changing contour geometry.

Recipe display

Mixing order strategy

Production order for mixing paints. Only the stage-based strategy is active; others are planned.

Application order strategy

Application order for filling cells. Only the dark-to-light strategy is active; others are planned.

Color Distance Metric

Family mismatch penalty

Neutral saturation cutoff

Skin max saturation

Brown max saturation

White Snap Luminance

Black Snap Luminance

Majority / mode filter

Recolor a cell when a different neighboring color clearly dominates around it.

Min dominant neighbors

Max same-color neighbors

Iterations

Remove tiny color islands

Merge very small connected same-color regions into their dominant surrounding color.

Max island size

Connectivity

Edge-aware smoothing

Smooth low-contrast areas while avoiding strong local edges from the extracted source colors.

Max local contrast

Min neighbor support

Iterations

Pre-quantization denoise

Filter extracted cell colors before Color Set matching to suppress noisy single-cell variations.

Method

Radius

Passes

Neighborhood-aware quantization

Bias Color Set selection toward colors that stay consistent with the surrounding cells.

Neighbor penalty

Iterations