About gooblen.ro
gooblen.ro turns a photograph into everything you need to stitch it: a counted chart, a photorealistic simulation of the finished piece, a colour list with real thread codes, and a calculation of how much thread to buy. There is no account, no subscription and nothing to install.
Why it exists
Converting an image to cross-stitch looks like an easy problem: divide the photo into a grid and give each square a colour. The hard part starts after that. Thread is not sold in arbitrary shades but in a fixed palette, printed on a shade card. A good chart is not the one that sits closest to the original pixels, it is the one that looks right once stitched, with a colour count a person is actually willing to work through and can buy at a needlework shop.
This project is written from scratch around that constraint, rather than as a front end over someone else's tool.
How the colours are chosen
The distance between two colours is not measured by comparing hex codes. Two blues can be numerically close and visibly different, while two greens can be numerically far apart and practically indistinguishable by eye. The system uses CIEDE2000, the published standard for perceptual colour difference, to estimate how different two shades actually look to a human viewer.
On top of that standard sits the part that is this project's own: choosing the palette. The system does not simply count which threads would be used most and keep the top few. That approach has a subtle flaw: once a pixel loses the vote, its error stops mattering at all, no matter how far it ends up from anything in the palette. Small, high-contrast details, an eye or a highlight, can never out-vote a large flat background, so those were exactly what got dropped.
Instead, the palette is built by repeatedly adding whichever thread removes the most remaining error, weighted by how many stitches it affects. Measured on a portrait from the gallery at 15 colours, the change brought mean error down from 6.54 to 5.38 and the worst case from 24.1 to 16.9. In practice that is the difference between a grey face and one with skin tones.
The method has two useful properties beyond the quality of the result. It is deterministic: the same image yields the same chart and the same symbols on every run. And it is nested: the palette for 20 colours is exactly the palette for 21 without its last colour, so comparing variants does not show you a palette completely reshuffled from one step to the next.
How many colours are enough
More colours mean more detail, but also more work, more skeins to buy and more needle changes. Past a certain point, the extra colours simply stop being visible.
So the system proposes a number itself, computed from your image: it builds the quality curve across possible palettes and picks the smallest one that lands within 15% of what a very rich palette would achieve. That is where the wide spread in the gallery comes from, with a flat graphic served by a dozen colours where a portrait needs fifty. The proposal stays a proposal: the slider is yours, and before deciding you can generate previews at your chosen count and at slightly lower and higher ones, to see the difference for yourself.
The upper limit is not arbitrary: every colour needs its own printable, legible symbol on the chart, and the symbol alphabet holds 207 of them. The most-used colour deliberately gets the blank symbol, because it is the one you will repeat most often and there is no point filling the page with an endlessly repeated mark.
The threads are real
Matching is done against the published palettes of existing brands, not an abstract list of colours. There are 52 ranges available, from the classic DMC, Anchor and Madeira cottons to tapestry wools, silks, hand-dyed threads and beads. You can browse the full palette of each one from the palettes page without starting a project.
Keep one thing in mind that no software can fix, though: the colour on your screen is not the colour of the thread. A screen emits light, thread reflects it, and the result depends on the fabric and on the light in the room you stitch in. The codes are a very good starting point, not a substitute for the physical shade card.
How much material you need
The shopping list is not a rough guess but a calculation from explicit parameters: a density of 10 stitches per centimetre, roughly 0.685 cm of thread consumed per stitch, and 800 cm skeins of 6 strands, of which 2 are used at a time. From these follow, for each colour, the stitch count, the thread length and the number of skeins to buy, ready to take to the shop.
They remain estimates, though. How tightly you stitch, how much thread you lose at the ends and which fabric you use all change real consumption, so it is worth buying a reasonable margin on the dominant colours.
What you get
The result is an archive containing a photorealistic simulation of the finished piece, the counted chart in both a colour version and a black-and-white one for economical printing, the colour chart listing each thread's code, name, stitch count and requirement, a colour reference with large swatches for matching by eye, and the skein list in a spreadsheet-ready format. Large charts are tiled automatically across A4, A3 or A2 sheets with grid coordinates; when that happens, you also get an assembly map showing how the sheets fit together.
The files stay available for 14 days, then delete themselves automatically. Download them and keep them locally.
Who makes it
The project is built in Romania by Valer Bocan, PhD, CSSLP (bocan.ro). It is under active development: if you find a problem, if a chart does not come out as you expected, or if a thread brand you use is missing, write to contact@gooblen.ro.