A filter puts dots on a picture. A halftone is made of them

Most halftones that look fake are a picture with a dot pattern laid over it. A real one is built from dots that know what they describe — merging in the shadows, vanishing in the light.

A white tee pinned flat to a plaster wall, printed with a coarse black halftone of a skateboarder mid-air, knees tucked over a tilted board, the figure built only from dots that merge in the shadows and thin to bare cotton on the lit edges.

A white tee pinned flat to a plaster wall like a poster. Across it, a skateboarder in mid-air — knees tucked, arms out, the board tilting under his feet. The dark jeans and the underside of the board are dots crowded almost solid; the lit shoulder and the top of the shirt are a few loose specks, then nothing.

There's no line anywhere in that figure. The dots are the drawing.

Why do some halftone designs look fake?

Because the dots are sitting on top of the picture instead of making it.

The quick version of the effect lays an even dot pattern over an image, or adds a dotted texture to a flat illustration. From far away it passes; up close the dots are the same size over light and dark alike, the outlines are still drawn in, and the shading underneath is still smooth. The dots are decoration. A real halftone has no smooth shading anywhere, because the size of the dots is the shading. Once you've seen the difference it's hard to unsee.

What does a real halftone do that a filter doesn't?

Three things, and all of them come from the dots following the light.

In the darkest shadows the dots grow until they touch and merge into solid ink. In the brightest highlights they shrink to nothing and leave bare paper — or bare shirt. And there's no outline: the edge of a shape exists only because the dots change size there. A figure drawn this way is modelled in light by the dots alone, which is why a good halftone reads as a photograph at a distance and as pure pattern up close.

Why does AI image generation often get it wrong?

Because halftone is ambiguous, and a model averaging every image tagged with the word lands on the ambiguous version: a picture with some dots on it.

The word covers newspaper photos, comic dot fills, pop art, dotted gradients in graphic design and a thousand retro overlay effects. Ask for "halftone" alone and you get the average of those. Name what the dots do and what era they come from, and the result sharpens.

What that sharpening looks like

At JustOG one line is enough — the design is expanded into full directions for you. If you want the wheel yourself, this is how the pictures on these pages were written. Each one names a source rather than a style: a newspaper wire photo, a 1980s skate magazine, a 70s surf poster. Each one describes the behaviour: dots merging into solid black in the shadows, shrinking to specks on the lit side, vanishing into the shirt where the light is brightest. And each one says no drawn outline, so the model can't fall back on drawing the shape and texturing it afterwards.

How coarse should the dots be?

Coarse enough to see from across a room, and no coarser than the subject can take.

A face needs enough dots across it for an eye and a mouth to survive; a big silhouette can take dots the size of peas. The rough test is whether you can still name the subject when you squint — if yes, the dots can get bigger; if the face turns into a blob, they've gone too far. On a shirt, coarse is also what the cloth wants: dots finer than the weave's own grain fill in and turn grey.

Does it matter what the picture is of?

Yes. The strongest halftones come from subjects with hard light.

A figure lit from one side, a face half in shadow, foam against dark water, a dog's coat in low sun: anything with a full range from deep dark to bright light gives the dots the most to say. Evenly lit, flat subjects give them nothing to do, and that's often when a halftone reads as a filter — because there's no light for the dots to follow, they fall into a uniform grid.

Is it all right to halftone someone else's image?

Turning an image into dots doesn't change who owns it.

A photograph belongs to whoever took it, and a halftone of it is still their photograph, only rendered differently. A famous shot, an artist's illustration or a recognisable person's face needs the permission it needed before. The technique itself is free — it was invented to print other people's pictures in newspapers, and nobody has ever owned it. Your own photos, or scenes described from nothing, are free on both counts.

Does it print well?

Yes — a real halftone is one of the most print-friendly images there is, because it was designed for printing.

Every dot is full-strength ink, so there are no pale in-between tones to go grey and patchy on fabric; the image is solid ink or bare cloth and nothing else. One colour is enough. Black on white gives the newspaper version; white on a dark shirt gives the night version, with the dots as the light. The filter version, by contrast, still has smooth shading underneath the dots, and that's the part that prints muddy.

How do I get one made?

Describe the subject and where the light is at JustOG — a coarse halftone, no outline, the dots merging in the shadows. Pick a direction, drag the crop frame, see it composited on the real garment, and it's made to order and shipped.

Designs other people have published are in the shop.

Take the dots away and there's no skater left. That's how you know they were doing the drawing.

Upload the image you already made, see it on the real garment, and get the physical piece. Made to order, no minimum.

Bring your image →

Related