Photo to Cross Stitch Pattern: How Many Colours Stay Stitchable

EFFECT EXPLAINERDESIGNER

Turning a photo into a cross stitch pattern is a grid and palette problem. Stitch-count maths, a colour-cap table and the failure modes that make a chart…

01/ ARTICLE

A photo becomes a cross stitch pattern in three moves: resize it to the stitch grid, cap the palette, then match each cell to the nearest thread colour. The palette cap decides whether the chart is stitchable, and about 30 colours is a sound starting point for a photo.

A cross stitch chart is a photo quantised to a fixed grid of cells and a fixed palette of thread colours. This is for the designer who wants the numbers: stitch-count arithmetic, a colour-cap table, three failure modes, and what kott can and can't do here.

What stitch count should you resize the photo to?

Resize to the stitch grid before you reduce colours, because the grid is the resolution of the finished piece. Stitches equal finished inches times the fabric count. Eight inches wide on 18-count fabric is 8 x 18 = 144 stitches across, and ten inches tall is 180 stitches down.

Finished width14-count16-count18-count22-count
6 in8496108132
8 in112128144176
10 in140160180220

Those four fabric counts, 14, 16, 18 and 22, are the ones one browser pattern maker offers. Its FAQ gives the same formula in reverse (stitch count divided by fabric count gives inches) and says to add 3 inches (7 to 8 cm) per side for framing. It also suggests 60 to 100 stitches across for a first project, and its width slider runs from 50 to 200.

The same grid logic applies to any photo reduced to coarse cells. If you have done it for pixel art, turning a photo into pixel art uses the same resize-first order.

kott shows the cost of a coarse grid. The cross-stitch effect has a cell size from 4 to 40 px with a default of 14, and cell size is the same knob as stitch count: fewer, bigger cells mean fewer stitches across. The grid below runs one bull photograph at 4, 8 and 16 px cells.

How does a photo get matched to thread colours?

Each cell is matched to a thread colour by finding the closest colour in the palette. Color quantization is usually described as two steps: choose a palette, then map the image onto it. With a fixed palette the second step is plain palette mapping, using the nearest colour by Euclidean distance in RGB, or in Lab when perceptual distance matters more.

Thread is the fixed-palette case. DMC makes 489 colours of six-strand floss, so you are not generating a palette from the photo. You are choosing a subset of the 489 and mapping to it. The same guide says basic converters compare colours with straight RGB distance, which does not track how people see colour difference, and recommends tools that work in CIELAB. Treat that as a recommendation rather than a measured result, since neither source publishes an error comparison.

If you want the palette-generation side, choosing a palette before you dither compares median cut, octree and k-means. For cross stitch the order flips: the thread colours exist first, so the cap on how many you use is the real decision.

Why does a cross stitch chart behave like an e-ink screen?

It has the same problem: a fixed grid, a short fixed colour list, and nothing in between. Thread count stands in for resolution, and the floss colours you buy stand in for the panel's colours. The 7-colour e-paper panel is the same constraint with 7 colours and a 600x448 grid. This comparison is kott's framing, not a claim from the pattern guides.

The consequence carries over. Error diffusion that reads as fine detail on a screen turns into a separate stitch for every speckle on cloth. You can try the screen version of the problem in the dither tool before you commit to a chart.

How many colours stay stitchable?

About 25 to 35 for a photo, and fewer if the chart will be sold or stitched by a beginner. The sources disagree a little, so here they are side by side.

Colour capWhere it comes fromWhat it suits
10 to 15FAQ of a browser pattern makerA first project
15 to 30StitchLark, on what buyers of sold patterns expectA pattern you sell
25 to 35StitchLarkA photo conversion, the stated sweet spot
30 to 50FAQ of the same pattern makerA photo-like result
Above 35StitchLarkBuyers start weighing floss cost; a 50-colour pattern can add $30 to $40 in thread

The working rule that fits all five rows: start at 30 and raise the cap only if the result looks blotchy. Limit the palette first, then choose specific DMC shades. Merge shades that look nearly identical when you squint, because they add stitches of a new colour without adding visible detail, and keep to a single thread brand, since sheen and hue differ slightly between DMC and Anchor.

What goes wrong when the palette is too big or too noisy?

Three things break a chart. A colour cap prevents the first two.

Confetti. Single stitches of a shade that appears nowhere else on the chart. StitchLark names this as the cost of keeping near-identical shades apart. You thread a needle for one stitch.

Noise turned into colours. A camera sensor leaves grain in flat areas such as skies and skin. With an unlimited cap, a converter can spend palette entries on that grain. This is our observation rather than a published test: if a flat area comes out speckled, lower the cap or blur the photo slightly before you convert.

Fractional stitches and dither. The pattern maker above offers smoothing that rounds diagonal edges with fractional stitches, and labels it harder to stitch. Its dithering option is Off or Floyd-Steinberg. On a small palette, every dithered pixel is its own colour change, so leave dither off unless a smooth gradient matters more than stitching time.

A clean-up pass that works is short. Cap the colours, merge near-duplicates, find isolated single stitches and fold them into their neighbours, then add backstitch to edges that blur together.

Can kott make a cross stitch chart?

No. kott does not map colours to DMC numbers or print a symbol chart. For the chart itself, use a dedicated pattern tool; they win here because they know the thread list and the fabric maths.

What kott does is preview how a photo survives a given grid. The cross-stitch effect, in the painterly category, draws an X of two diagonal threads in each cell over a faint aida-cloth weave, with the colour sampled at the cell centre and then quantised. It has three controls:

  • Cell size: 4 to 40 px, default 14.
  • Thread: 0 to 1, default 0.5, which sets how thick the strands are.
  • Contrast: 0.2 to 3, default 1.1, under advanced.

Run your photo at a few cell sizes before you spend a weekend on a chart. If the subject falls apart at 8 px cells, it will fall apart at the matching stitch count.

What should you do first?

Fix the grid, cap the palette at about 30, merge near-identical shades, then preview the result. Open the cross-stitch effect in kott with your own photo and try cell sizes 8, 14 and 24 before you commit to a stitch count.

Frequently Asked Questions

How many colours should a photo cross stitch pattern have?

Around 25 to 35 is the figure one pattern guide gives for photos. Another tool suggests 30 to 50 for a photo-like result and 10 to 15 for a first project. Start at about 30 and raise the cap only if the result looks blotchy.

How many stitches wide should a photo pattern be?

Multiply the finished width in inches by the fabric count. Eight inches on 18-count fabric is 144 stitches. One pattern tool suggests 60 to 100 stitches across for a first project.

Does kott export a DMC chart?

No. The cross-stitch effect in kott renders the look: an X of two diagonal threads per cell over an aida weave. It does not map colours to DMC thread numbers or print a symbol chart.

Should I use dithering on a cross stitch chart?

Usually not with a small palette, because every dithered speckle becomes a separate stitch to place. Pattern tools offer Floyd-Steinberg as an option; use it only where a smooth gradient matters more than stitching time.

02/ OUT

Try it in the studio

Every setting described above is a real control. Open your own image and sweep it.

All journal entries