My cousin wanted to turn a photo of her cat's face into a perler bead ornament. She took a snapshot with her phone, sent it to me and asked how many beads to buy and which color codes to use. I stared at the photo for a long time before answering: a photo is continuous color, while beads are separate pieces, so there has to be a step in between that spreads the colors across a grid and then reduces them to a limited set of codes. Skip that step and none of the three questions, cell count, color codes or quantity, can be answered.
Set the finished size first, then the cell count
Beginners usually work in the opposite order: they import any image, get 100x100 cells, decide it looks detailed, and then discover that the board they own cannot hold it, or that the finished piece is twice the size they expected with nowhere to put it.
The relationship is simple: the side length of the finished piece is roughly the cell count multiplied by the center-to-center spacing of one bead. The spacing comes from the bead size, and medium and small are the two common sizes with different nominal values. I calculate from the figure printed on the package rather than by feel. The numbers below are my estimate at the scale of a medium bead around 5mm each; they are only meant to build intuition, so redo them for your own beads.
| Canvas (cells wide x high) | Total cells | Estimated finished side | What I use it for |
|---|---|---|---|
| 24 x 24 | 576 | about 12cm | Coasters and keychains, done on one board in a single pass |
| 40 x 40 | 1600 | about 20cm | Portrait ornament; this is the size I used |
| 52 x 52 | 2704 | about 26cm | Near the limit of one board; anything larger needs joining |
| 72 x 72 | 5184 | about 36cm | Must be joined; lining up pegs across boards shifts the whole design |
Joining boards is where I got burned. Once the pattern is cut by a board edge, every row afterwards has to line up with the pegs, and one cell off tilts the whole piece. So my rule now is to ask where it will sit and how big it should be, work backward to the cell count, and only then talk about the image.
Quantizing to a palette always shifts the color
A photo has millions of colors, while a bead palette is a few dozen codes. All the tool can do is map each cell to the nearest code, and the word nearest is where the color shift comes from. It will not give you a color that was not there; it gives you the closest substitute it can.
The three places where the shift is most obvious, in my experience, are large gradients (blurred backgrounds, the highlight transition in fur), low saturation areas (gray-white fur, skin) and shadows. Flat backgrounds and patterns with clear outlines hardly shift at all.
The controls sit at the input and at the edit stage:
- Choose a photo with even lighting and avoid backlight and colored light from one side, or the whole image picks up a cast in that direction.
- Crop away everything irrelevant first and keep only the subject. A background that goes into the quantization is wasted cells.
- More colors is not better. When I cut from 12 colors to 7, the scattered colors that occupied only a dozen beads each merged into neighboring main colors, the pattern looked more unified, and I needed two fewer codes in the shopping list.
- Set colors by hand on the critical parts. In the generated chart I change the cells for the eyes and the nose one by one, clicking a cell to recolor it, because those few cells decide whether it looks like the cat.
I keep the grid on while building the chart and turn it off to judge the result: the grid lets me line up pegs and count cells, and only with it off can I see whether the whole thing is that cat.
The color count is for buying supplies, not for looking at
The tool reports the quantity of each color, and for handwork that is the most concrete output: knowing how many codes and how many beads of each is what lets you work out how many bags to buy.
My method with the numbers is to group by color first, then add a margin on top of each group. Losses come from test ironing, from pieces that have to be pulled apart after a failed fuse, and from the few beads that fall on the floor and vanish. That 40x40 cat face ended up with 7 codes; the main color, a light gray for the fur, took about 900 beads, and the rest ranged from a few dozen to two hundred. Divide by the nominal count per bag and the number of bags is obvious.
The other use is judging feasibility in advance. If the top two or three colors already cover eighty percent of the cells, the image is actually simple and the cell count can come down. If instead the number of colors approaches half the cell count, the job does not suit handwork, and you will want to quit around bead 300.
One more thing I do before importing
Phone photos are often 4000 pixels wide, while all I need is a 40x40 sample. The cell count is the sampling limit; extra resolution adds nothing to the finished piece, but I want to see detail while previewing and choosing, so I usually crop and scale the image to the target proportions first, so the composition is already right inside the grid, and then move on to quantization. I do that step in Image Resizer, mainly to lock the aspect ratio of the finished piece so the face is not squashed when it comes out.
The actual conversion happens in Perler Bead Pattern Maker: upload the image, set the width and height in cells, match the built-in palette automatically, get the color counts and export the chart image. I usually print a paper copy and keep it beside me for counting cells.
When the preview colors do not match the real beads, I do not trust the screen. I read the colors of a few key cells off the chart as hex values and check them against the codes on the palette, which I do in Color Converter. Brand naming systems do not talk to each other; the same light gray can have completely different numbers in two product lines, and hex is the only thing that can be compared.
Q: Does a sharper photo produce a better pattern?
No. The final pattern is determined only by the cell count; sharpness affects how comfortable you are while judging and cropping. To get more detail, the only effective move is more cells, at the cost of a larger finished piece and more beads.
Q: Do more colors make it look more like the original?
Closer, but the limit is clear: the palette does not contain that color, so adding codes cannot reproduce it. And once a transition area is split into too many colors, the handwork difficulty and cost jump. I usually reduce colors first and add later if needed.
Q: What if the fused piece looks very different from the chart?
I iron a small test patch first. After fusing, the beads merge into each other, and the look is not exactly the screen preview, especially for light colors that let light through. Once I have seen the test, I decide whether to swap a code.
Q: Are empty cells included in the usage count?
No. The count covers cells that actually hold a bead; empty cells are simply unused. Conversely, if the color total is small but the cell count is large, the subject occupies most of the frame and the design suits a solid ornament.
Q: Can I just buy by the color names printed on the chart?
I would not. Every brand has its own naming, and the mistake I made was buying by name and finding the beads a shade darker. Checking hex values, or reverse-matching against the palette in your own hands, is more reliable.
My cousin ended up with a 40x40 chart in 7 colors, an estimated side of 20cm, about 900 beads of the main color and dozens to two hundred of each other color, and she bought three bags plus two extra codes based on the count. The answer to her original question of how many beads to buy is not a number you look up: it comes from the cell count, the number of colors and a margin, all three together. Fix those and the quantity falls out on its own.