Guides

Photo to Bead Pattern: Image, Size, Colour Count

Updated 2026-10-12

Converting a photo into a bead pattern usually fails at "it does not look like the original". This is not about which buttons to press, but about four decisions: whether the image qualifies, how large to go, how far to cap colours, and what to fix by hand afterwards.

Start with a squint test

Save the image you want to use, then cover the colours and look only at the light-and-dark shapes — can you still tell what it is?

This one test filters out most failures. A bead pattern quantises an image into a limited grid where each cell can hold exactly one colour. If the original carries its information mainly through colour and gradient, quantisation will tear it apart.

Converts wellFalls apart
Line art, icons, cartoon characters, pixel artCandid photos with busy backgrounds
Subject-focused still life on a clean backgroundLarge gradients (sunsets, seascapes)
Pets, front-on, with clear coat contrastFine texture (grass, foliage, fur)
A single person, not a group shotGroup photos and wide landscapes

If the image fails, you have two options: use a different one, or crop first — isolate the subject and drop the busy background. Cropping is usually the more effective move.

Step 2: Pick a size — this decides whether you finish it

Size is the most underrated decision. Many people assume bigger means clearer, end up with a several-hundred-cell project spanning a dozen colour families, and abandon it halfway.

Some real-world reference points:

The right question is not "which looks better" but how many evenings you intend to spend on it. Fuse beading is handwork; a pattern is worth something when it gets finished.

Step 3: Cap the colour count — more is not better

Colour count drives two things: how hard it is to buy and how hard it is to assemble.

A 50×50 pattern using 40 colours means buying 40 codes — and most codes are sold by the pack, while you might use only a few dozen beads of each. Assembly also means constant colour switching, which raises the error rate sharply.

Comfortable ranges in practice:

Capping colours loses some detail, but buys you a pattern you can actually buy and finish. That trade is usually worth it.

Step 4: Three things you must fix after converting

Automatic conversion returns the nearest quantisation, not a finished pattern. Check at least three things:

  1. Subject edges — algorithms tend to scatter stray cells around outlines; fixing a handful cleans up the whole figure
  2. Isolated single cells — one cell in a contrasting colour surrounded by something else is usually noise; delete it
  3. Key features — eyes, mouths, small logos and text blur first and need hand work

Typically this is a dozen cells of editing, and it changes how the finished piece reads.

Three common failures

Failure 1: a family photo at 50×50. Each face ends up a few pixels across and nobody is recognisable. Group shots need a very large grid to keep facial features — convert one person instead.

Failure 2: chasing 40 colours for fidelity. It does look accurate, but half the codes turn out to be out of stock and the project stalls. Confirm you can buy the codes before you chase fidelity.

Failure 3: a sky background. The gradient quantises into visible bands and looks like a rendering fault. Crop the background out and it is fine immediately.

A short checklist

Run those four and you will not produce an unfinishable pattern. For how the whole process fits together, see the getting-started guide; for colour-code specifics, see the colour guide.

Start making a pattern →