"How to turn a photo into a cross stitch pattern"

Most people expect a miracle from a photo-to-pattern converter: upload a snapshot, get a perfect chart. In practice the output usually looks like blurred pixel noise, and the algorithm takes the blame. Any automatic conversion is just a tool. To make it produce something good, you have to meet it halfway. A clean chart starts long before you click "Generate". Four simple steps turn a photo into a piece of needlework worth framing.

The quality of a photo-to-pattern conversion is driven by four factors: chart size in stitches, palette colour count, contrast of the source photo, and the quantization algorithm. Large files and dense colour don't help — the smaller each stitch and the more lookalike shades collapse into one averaged group, the muddier the result. A good chart starts not with the upload, but with prep: cropping to the final frame ratio, boosting contrast, simplifying the background.

Why the result comes out muddy (and where colour confetti comes from)

Picture this: you take a huge phone photo and squeeze it down to a tiny piece of aida. What does the software do? It takes hundreds of pixels and tries to average them into a single stitch.

A bright highlight in the eye, a dark eyelash and a patch of skin all blend together — and out comes a plain beige cross. Sharp edges blur, smooth gradients turn into stripes.

The obvious fix seems to be: add more colours! That is the main trap. Ask the program for 100 colours on a portrait and it will find dozens of nearly identical greyish-pink shades. Neighbouring cells will start flickering between them. To the eye it reads as visual noise; to the stitcher it means a hundred skeins of floss and a twitchy eye from constant thread changes.

The algorithm doesn't know which part of the photo is the eye and which is the background. Everything is equally important to it. So the job of placing the accents is on you.

Step 1: A good photo is 80% of the result

The most common complaint is "the face came out unreadable". The cause is almost always the source image itself.

Focus and sharpness. The software can't imagine missing detail. If the photo is slightly blurred, the chart will be twice as blurred. Pick the sharpest frames.

Light. Soft daylight (a window) is ideal. Harsh sun shadows turn into solid black or dark-brown patches on the chart, eating detail.

Tight crop. A full-body group photo is the worst candidate. With 30 stitches per face, there is physically no room for eyes or a smile. Choose chest-up portraits or tighter.

Drop the busy background. A patterned carpet or a detailed street scene will steal colours from the palette. The plainer the background, the more shades are left for the subject.

Step 2: Pick a size (so you don't stitch for years)

Chart size in stitches is like screen resolution. Too few and detail collapses into blocks. Too many and the project drags on forever. How do you find the balance?

Anchor the decision to two things: the target physical size and the subject itself.

From the frame size. If the end goal is fixed — say, 30 × 40 cm on a bedroom wall — convert it to stitches via aida count. On 14ct aida that's about 165 × 220 stitches; on finer 18ct, 212 × 283.

From the subject. For an adult face to be recognisable, the face itself (chin to crown) should take roughly 80–90 stitches in height. A landscape with houses needs at least 200 stitches across. A clean cat silhouette or a single flower can read well at 100 × 100.

The golden rule: 1 pixel = 1 stitch. This is the trick experienced stitchers use. Before uploading a phone photo (millions of pixels) into the converter, open it in any photo editor and downscale the photo itself to the chart's resolution. Want a 200 × 150 chart? Resize the source to 200 × 150 pixels (and re-sharpen). You decide which detail survives and which disappears, instead of leaving it to a soulless algorithm.

The universal ceiling. Charts bigger than 300 × 400 stitches turn into a year-long marathon. For a gift, aim at 150–250 stitches on the long side — the finished size stays sensible, and so does the stitching time.

Step 3: Don't get greedy with colours

The DMC palette has around 500 shades; your photo has millions. Don't try to keep them all. Cross stitch is closer to watercolour or mosaic: expressive blocks of colour read better than a million smooth transitions. There's more on working with colour in the DMC palette guide.

Realistic numbers to start from:

Subject	Sweet spot
Human portrait	30–50
Landscape	25–40
Floral motif	15–25
Silhouette / minimalism	8–15

A good converter lets you "lock" a few key shades so they aren't averaged into their neighbours — the white of an eye, a highlight, a saturated lip accent. If yours doesn't, you'll have to repaint those points by hand after generation.

Remember: an 80-colour portrait chart almost always looks worse and noisier than a 35-colour one. Save your floss budget and your sanity for actually stitching.

Step 4: The prep magic (10 minutes before upload)

Before handing the photo to the converter, spend a couple of minutes in a phone or desktop editor. This saves hours of cleanup later.

Crop tight. Lock the aspect ratio of the future chart up front.
Boost contrast (+10–20%). Conversion always softens and averages colours. Make the photo a touch brighter and punchier than reality — on the chart it lands just right.
Lift the shadows. Pull the Shadows slider up a bit so dark hair or clothing doesn't collapse into one black blob.
Blur a busy background. If the backdrop carries no meaning, soften it — the algorithm will spend the saved shades on the subject.
Erase the background. For a clean portrait on plain aida, remove the background in the editor before uploading.

Step 5: Final touches (the manual pass)

No converter in the world produces a perfect chart on the first try. Treat the software's output as a strong draft that needs grooming.

What's worth doing almost every time:

Sharpen the contours. Walk through the key lines (brows, lip contour) and tighten them where the algorithm blurred them out.
Remove the orphans. Spotted a single green stitch in the middle of a cheek? That's an algorithm glitch (a grass reflection). Paint it over with skin tone, no hesitation.
Merge near-duplicate colours. If the chart has two reds and one of them covers only 20 stitches across the whole piece, replace it with the other. Nobody will notice, and you save a whole skein.
Check readability. Compare the chart preview against a thumbnail of the original at the same size. If it's still recognisable at that scale, the chart is done. If not, go back to Step 1 — the photo needs different prep.

How to turn a photo into a chart: step by step

Prep the source. Crop to the final frame ratio, lift contrast and saturation by 10–20%, blur or remove a complex background if needed.
Decide the size in stitches. Either from the target physical size and aida count, or from the subject (at least 80–90 stitches for face height, 200 across for a landscape).
Downscale the photo to chart resolution. Planning 200 × 150 stitches? Don't upload a 6000 × 4000 photo — resize and re-sharpen in a photo editor first.
Run the conversion. Load the prepped file into the photo-to-pattern converter; start with the DMC palette and the colour count from the table above.
Choose the colour count by comparison. Generate a few variants (e.g. 25 / 35 / 50 colours) and pick the most recognisable one, not the most "detailed".
Clean up by hand. Repaint key contours, kill single noise stitches, merge near-duplicate shades with low stitch counts.
Verify readability. Compare the chart preview to a downscaled copy of the original at the same height. If it's still recognisable, export the PDF and order your floss.

Frequently asked

Can any photo be turned into a cross stitch chart?

Technically yes, but results depend on the source. Tight portraits, macro shots (flowers, details) and high-contrast landscapes work best. Full-body group photos, dark frames or out-of-focus shots will not become a good chart.

How many colours should I use for a portrait?

Sweet spot: 30 to 50 DMC shades. Below 25 the face looks patchy, like an old video-game sprite. Above 60 you get "colour noise" — clusters of single-stitch oddities.

What's the minimum chart size for a recognisable human face?

The face itself (without hair or background) needs to occupy at least 80–90 stitches in height. If the whole chart is 150 × 200 stitches and the face fills only a quarter of the height, you're left with about 50 stitches for the features — and recognisability almost certainly suffers. For portraits, go in tight.

Why is the chart still blurry?

Most likely the software is trying to squeeze a huge photo into a small chart. Crop the photo in advance, push contrast up, and drop the colour count.

Do I need to remove the background before converting?

The background is a colour thief. If it's a patterned carpet, the software will spend 15 colours on it and leave 10 for the subject. Blur it, or erase it entirely, so all the attention (and all the floss) goes to the main object.

Why use a browser-based photo-to-pattern converter instead of a desktop app?

A browser tool needs no install, no licence, no OS compatibility — open a tab, drop in the photo. Modern web converters (ixo included) already handle the DMC palette, background removal and on-fabric preview right in the browser. Desktop apps may offer finer controls, but the cost of entry is higher in both money and time.

In ixo: the photo is a starting point, not the finish line

We at ixo (our browser-based editor) know this pain well. So in ixo, building a chart from a photo isn't a black box, it's a workflow.

In ixo, converting a photo isn't a single step — it's a chain. You upload a photo and, in the same place, can remove the background, pick the DMC palette and dial the colour count. More importantly, the output opens straight in the full editor. No jumping between tools: you can immediately wipe a stray stitch, tidy a lip contour, merge a redundant shade and export the PDF you actually want to stitch from.

The takeaway: the software does half the work. Picking the right photo and the final hand-polish are yours.