Halftone & Dither Generator

About the Halftone & Dither Generator

Turn a photo or a drawing into halftone dots or dithered pixels at print resolution. For a halftone, choose the screen ruling in lines per inch, the screen angle and the dot — round, elliptical, square, diamond, lines or a cross — and the generator draws one dot per cell, sized so that its area is the tone of that spot, as a printing screen does. Make a one-ink screen for screen printing, DTF transfers, risograph or laser engraving, or a CMYK separation preview with four plates at their own angles. For dithered art, choose Floyd–Steinberg, Atkinson, Jarvis–Judice–Ninke, Stucki, Burkes, Sierra Lite or ordered Bayer dithering to 1 bit, to grey levels, or to a palette of your own inks or of colours taken from the picture.

The result is made at the size and DPI you print at — a 200 mm print at 300 DPI is 2,362 pixels wide — with dot-gain compensation and the smallest and largest dot your press can hold, and you can inspect any spot at print resolution. With a Pro pass it downloads as a transparent PNG with the DPI written in the file, or as a vector SVG with one shape per dot; without one, the free preview shows the result at preview size with a watermark and one detail at print resolution. Everything happens on your device: the picture is never uploaded.

Common uses

• Halftone films and DTF transfers for screen-printed T-shirts, posters and merchandise.
• Photos for laser engraving and laser marking, dithered to one burn spot per pixel.
• Retro newspaper, comic and pop-art looks for posters, album covers and social posts.
• Risograph and two-ink prints dithered to the inks you have.
• Vector halftone patterns for vinyl cutting, stencils and design files.

How the halftone is made

A halftone screen lays a grid of cells over the picture — as many cells per inch as the screen ruling says, turned by the screen angle — and paints one dot in each cell. The dot’s area is the share of ink the picture needs there: a 30% grey gets a dot covering 30% of its cell. Round dots touch their neighbours at about 79% and then fill the gaps between them; elliptical dots join into chains along their long side at about 47%; diamond dots meet at their corners at 50%, like a checkerboard; square dots, lined up with the grid, fill their cell only at 100%; lines get thicker; crosses grow to the size of their cell and then thicken their arms.

The output pixels are the printer’s dots, so the screen ruling and the resolution decide how many tones a cell can show: a cell of n pixels shows n + 1 levels (Adobe PostScript Language Reference, section 7.4). At 300 DPI a 45 LPI screen has cells of 6.7 pixels, about 45 tones; a 100 LPI screen only 3 pixels and 10 tones, which the page warns about. Sharp edges decide each pixel on its own (1-bit, as a film or plate needs). On a turned screen the dots meet the pixels at a different place in every cell, so their area keeps the tone; on a screen at 0° or 90°, whose cells line up with the pixels, each pixel of a cell switches on at its own step of the tone, as in a printer’s threshold array, so a cell of n pixels shows all of its n + 1 tones. Smooth edges are anti-aliased for pictures shown on screens.

Screen angles and moiré

A single ink is usually screened at 45°, where the eye notices a regular pattern least. When screens of several inks lie on top of each other, their grids interfere and can form moiré; turning them apart makes the dots form small rosettes instead. The defaults are the traditional angles — cyan 15°, magenta 75°, yellow 0° and black 45° — which keep the three strong inks 30° apart and put yellow, the least visible ink, 15° from its neighbours. The page warns when two plates share an angle (or differ by 90°, which is the same for a square grid).

Dot gain, smallest and largest dot

Ink spreads: on paper, mesh and fabric a printed dot comes out larger than on film, most in the middle tones. Dot gain at 50% compensates for it in advance with a simple parabolic model of tone value increase that peaks at the 50% tone, TVI(a) = 4 × gain × a × (1 − a): each dot is made smaller so that it prints at the intended size. Measure or ask your printer for the value; 0 leaves the dots as they are.

Smallest dot drops dots too small to hold on a screen or plate: a tone that needs one becomes paper or the smallest dot, whichever is nearer, so the lightest highlights print as paper. Largest dot does the same in the shadows, where dots close to solid fill in on press anyway.

The dithering methods

• Floyd–Steinberg passes each pixel’s rounding error to its unvisited neighbours — 7/16 to the right, 3/16 below-left, 5/16 below, 1/16 below-right — so areas keep their average tone; the classic all-rounder.
• Atkinson (from the early Macintosh) passes on only 6/8 of the error, to six neighbours: flat areas stay cleaner and contrast goes up, at the cost of some detail in the darkest and lightest parts.
• Jarvis–Judice–Ninke, Stucki and Burkes spread the error over 12 or 7 neighbours: smoother, less wormy gradients, good for engraving photos.
• Sierra Lite uses three neighbours: light and fast.
• Bayer ordered dithering adds a fixed 2 × 2, 4 × 4 or 8 × 8 threshold pattern instead: a regular cross-hatch that looks retro and repeats exactly, with no error carried between pixels.

Serpentine scanning runs alternate rows right to left, which avoids diagonal streaks. Linear light dithers in linear light rather than in the picture’s encoded values, so a dithered area looks as bright as the original on a screen.

The CMYK separation preview

For CMYK the picture is separated with a simple device-independent conversion: black takes the share of grey you choose (Black generation, K = share × (1 − the largest of R, G, B)) and cyan, magenta and yellow the rest (C = (1 − R − K) ÷ (1 − K), and so on). Each plate is screened at its own angle and the plates are shown on their own and printed on top of each other in screen approximations of the process inks. It is a preview and a creative effect, not a colour-managed separation: it uses no ICC profile, ink limit or press characterisation, so use your printer’s RIP for colour-critical jobs.

The files

The PNG is made at the full output size, with the resolution stored in the file (its pHYs chunk), so layout programs, RIPs and laser software place it at the right physical size. Sharp halftones and 1-bit dithers are saved as 1-bit PNGs, CMYK composites and palettes as indexed colour, so even large files stay small; with a transparent background the paper is transparent and only the ink is solid. The SVG has one circle, ellipse, square, diamond, cross or line outline per dot, placed in the screen’s own turned grid, with the print size in millimetres and a clip path at the picture’s edges.

Limitations

• Outputs are limited to 16,000 pixels on a side and 40 megapixels (an A2 poster at 300 DPI is about 35); anti-aliased CMYK to 30 megapixels.
• Pictures larger than 4,096 pixels or 12 megapixels are reduced to that working size first, which is still more detail than a screen of cells can show.
• Tones come from the picture’s grey values: a 50% grey becomes a 50% dot. Use Midtones and Dot gain to match your press.
• The CMYK plates are shown for checking; the download is the composite. SVG is made for one-ink halftones of up to 1,500,000 dots.
• In the free preview the result is shown at preview size (up to 640 pixels) with a watermark, and one 256-pixel detail at print resolution; the downloads stay locked until the result is unlocked.

Privacy

Everything happens in your browser. What you enter or open here is not uploaded or stored by MySmartCoPilot.