Which Photos Make Good Cross-Stitch Patterns?

The most common reason a photo-generated chart disappoints is not the software. It is the photo. A chart is a drastically reduced version of your image - a few hundred stitches across, in thirty-odd colours - and that reduction is unforgiving. This is how to tell, before you start, whether a photo will survive.

The one test that predicts everything

Open the photo and shrink it on screen until it is about the width of your thumb. Then look at it.

That is roughly what a chart sees. If at thumbnail size you can still tell what the subject is, where its edges are and where the light is coming from, the photo will convert. If it turns into an indistinct blob, so will the chart - because the generator has no more information than you do at that size.

What actually matters

Tonal contrast, not colour contrast

Stitchers learn this the hard way with black pets. A black dog against a dark green sofa has lots of colour difference and almost no brightness difference. Convert the photo to greyscale in any photo app: if the subject vanishes into the background in greyscale, it will vanish in the chart too, no matter how many threads you throw at it.

Fix: reshoot against a lighter background, or choose a frame where light rims the subject's edge.

A tight crop

Every stitch you spend on background is a stitch not spent on the face. Crop so the subject fills most of the frame before uploading. If you want an oval or circular finish, crop to that shape early - it changes where the interesting detail sits.

Even, directional light

Soft daylight from a window, slightly to one side, is close to ideal: it gives shape without hard-edged shadows. Direct flash flattens everything into a mask. Strong backlight makes a silhouette - occasionally a great effect, usually not what you wanted.

Sharpness where it counts

Eyes and lettering must be sharp. A soft background is fine and often helps separate the subject. Motion blur across the subject cannot be recovered.

A clean original file

Use the original photo, not a screenshot of it, not a copy re-sent through a messaging app. Each round trip adds compression artefacts, and compression noise converts into confetti stitches. Low-light phone photos carry heavy sensor noise for the same reason - the grain becomes stray single stitches.

Subject by subject

Pets. The best-behaved subject there is. Fur reads as texture, so small inaccuracies disappear. Shoot at the animal's eye level, get a catchlight in the eye, and crop to head and shoulders. Dark animals need rim light.

People. The hardest, because we are exquisitely sensitive to faces. Use a front-lit portrait, avoid wide-angle phone lenses close to the face, and allow more stitches - 180 wide is a floor for a recognisable adult face, and 220+ is safer. Group photos rarely work at any size.

Landscapes. Deceptive. They look detailed but are mostly gradient, and gradients turn into banded stripes. Choose scenes with strong shapes - a lone tree, a lighthouse, a skyline - rather than a beautiful but formless sunset.

Logos, lettering and line art. These convert beautifully, because they are already flat colour with hard edges. Go smaller than you would for a photo, and lean on backstitch for the outlines.

Flowers and still life. Usually excellent: bold shapes, strong colours, a naturally out-of-focus background.

Three problems and their fixes

"The background is busy." Crop tighter first. If the subject still competes with clutter, remove the background before converting - many editors, Stitcherry included, can cut the subject out automatically, which also saves hundreds of stitches of wallpaper.

"Skin tones look wrong." Usually a colour-matching problem rather than a photo problem. Nearest-thread matching done in RGB pushes skin towards green or grey; perceptual matching in CIELAB fixes it. If your tool has no such setting, hand-correcting a dozen stitches on the cheeks does a surprising amount.

"It looks flat." Too few colours, or a flash-lit source. Raise the colour count slightly and add backstitch along the main edges; if it is still flat, the photo has no tonal range to work with.

Before you upload - quick checklist

• Shrunk to thumbnail size, the subject is still readable.
• In greyscale, the subject separates from the background.
• Eyes (or lettering) are sharp.
• The subject fills most of the frame.
• It is the original file, not a re-shared copy.

Got one that passes? Drop it into the editor - generating and previewing is free, so it costs nothing to see how it charts.