Which Photos Turn Into a Good Paint by Numbers?
A paint by numbers canvas is your photo reduced to twenty-odd flat colours with hard edges between them. That reduction is brutal. Some photos survive it and come out better than the original. Others turn into an unrecognisable smear, and no amount of extra colours saves them.
The good news: you can tell which is which in about ten seconds, once you know what to look for.
The one test that predicts everything
Squint at the photo until it goes blurry. If you can still tell what it is — where the face is, where the light falls, what is in front and what is behind — it will make a good painting. If it dissolves into one grey mush, it will not.
That is not a party trick. Squinting throws away exactly what the conversion throws away: fine detail and subtle tonal steps. What survives your squint is what survives the pattern.
What makes a photo work
Light from one side
Side light creates a light half and a shadow half, and that difference is what turns into shape on the canvas. A face lit from the side gets a nose, a jawline, cheekbones. The same face under flat overhead light or a direct flash turns into a pale oval with features drawn on it.
The subject fills the frame
Whatever you care about should take up most of the photo. A dog fifteen metres away in a field is a brown blob however good the camera was. Crop in before you upload — the closer the crop, the more areas get spent on what matters.
A background that stays quiet
A busy background does not just look busy, it steals colours. The palette is divided over the whole image, so a cluttered kitchen behind your subject eats colours that the face needed. Plain walls, open sky, grass, water and blurred backgrounds all work well.
Real contrast between subject and background
A black dog against a dark sofa is one shape to the converter, not two. Dark on light, or light on dark — the separation has to exist in the photo before it can exist on the canvas.
What does not work
- Group photos. Four faces across one canvas means each face gets a quarter of the detail. One person, or two at most.
- Direct flash. It flattens every shadow, and shadows are what the pattern uses to build form.
- Screenshots and heavily compressed images. The blocky artefacts become their own areas with their own numbers.
- Very dark or very bright photos. If the detail is lost in the shadows on your screen, it is lost for good — the converter cannot invent it.
- Heavy filters. Anything that already crushed the tones leaves less for the pattern to work with.
Resolution matters less than you would think
A modern phone photo is plenty, even for a 60x80 cm canvas. The pattern reduces the image to a few dozen colour areas, so it is not looking for pixel-level detail. A photo that is sharp on your phone screen is sharp enough.
What does matter: do not use a photo that has been sent through a chat app several times. Each round of re-compression eats edges, and edges are exactly what the converter looks for.
Portraits deserve their own paragraph
Faces are the hardest subject and the most popular one. Two things decide whether a portrait works:
The eyes need to be big enough on the canvas. An eye needs several distinct areas — iris, pupil, white, lid shadow — and each of those needs to be at least a couple of millimetres wide or a brush cannot reach it. On a 30x40 cm canvas with a half-body portrait, the eyes are simply too small. Crop closer, or go bigger.
Skin tones need enough colours. Skin is a gradual gradient, which is the hardest thing to represent in flat areas. Twenty-four colours gives a poster-like face. Thirty-six is where it starts looking like the person. See how many colours your painting needs.
Just try it
All of this is a rule of thumb, and the generator is free and unlimited. Upload the photo, look at the preview, and judge it on what you see instead of on what should theoretically work. Two minutes of trying beats twenty minutes of guessing — and if the preview looks wrong, try a tighter crop before you give up on the photo.