What Is a Halftone? How the Print Effect Works, How to Make One

The short answer

A halftone fakes continuous tone with ink that is either there or not. The image is divided into a grid of small cells, and each cell gets one dot whose area matches how dark that spot of the image is: a 20% grey becomes a dot covering about a fifth of its cell, a 90% shadow becomes a dot so large it nearly fills the cell. From reading distance the eye averages dots and paper into grey. Colour printing does the same thing with four grids, one each for cyan, magenta, yellow and black, rotated to different angles.

You can try every setting described below in the free halftone generator on this site, which runs in the browser and exports PNG or SVG.

Where the halftone came from

The core problem was simple: a printing press puts down solid ink, and a photograph is made of continuous greys. Line drawings could be engraved and printed; photographs could not, until someone found a way to turn grey into a pattern.

• 1852. William Henry Fox Talbot’s patent for photographic engraving proposed “photographic screens or veils” to break up the image on an intaglio plate. The V&A holds a Talbot photoglyphic engraving of ferns from about 1858, made through a fabric screen (an inscription on it mentions a “crape veil”); a past gallery label describes Talbot laying loosely woven fabric over the image so that solid areas broke into small ink-holding ones.
• March 4, 1880. The Daily Graphic in New York printed a halftone, rather than an engraving, of a news photograph, “A Scene in Shantytown”. It is usually cited as the first halftone news photograph in a US newspaper.
• 1881–1882. Frederic Ives patented a commercially successful dot process in 1881, and Georg Meisenbach patented his “autotype” process in Germany in 1882; it used a single-line screen turned during exposure to get a cross-lined result. Ives then worked with Louis and Max Levy on the invention and commercial production of cross-lined screens.

No single inventor owns the process. Cornell, where Ives worked in the photographic laboratory from 1874 to 1878, credits him with inventing it there; other histories give credit to Talbot, Meisenbach and the Levys, and to earlier experimenters such as William Leggo in Canada, whose halftone of Prince Arthur was printed in 1869. The principle is still used to print photographs today.

How an AM halftone works: cells and dots

The classic method is called AM (amplitude-modulated) screening: dot spacing is fixed, dot size changes. Picture a square grid laid over the image. Each square is a halftone cell. For each cell the screen reads the average tone underneath and draws one dot in the centre, sized so the inked area equals that tone. Light areas get pinpoints, midtones get dots that almost touch, shadows get dots so large that the remaining paper shows as small light holes.

Round dots touch their neighbours at about 78.5% coverage (the area of a circle inscribed in a square is π/4 of it). Above that point, dots merge into a dark field with holes, which is why the shadow end of a halftone looks different from the highlight end.

LPI, DPI and PPI

Three numbers get confused constantly:

A halftone dot is built out of device dots. If a platesetter images at 2400 dpi and the screen is 150 lpi, each cell is 2400 ÷ 150 = 16 device dots wide, so 16 × 16 = 256 device dots can be switched on, giving 257 possible tone levels including blank paper. Drop the device to 1200 dpi at the same ruling and each cell is only 8 × 8, which gives 65 levels, and smooth gradients can start to show steps. That trade-off between ruling and tone steps is the reason fine screens need high-resolution output.

Higher LPI means finer detail and less visible dots, but only if the paper and press can hold the dots. The CUNY City Tech resolution guide notes that newspaper dots are spaced far apart because cheap paper soaks up and spreads the ink.

Dot gain

Printed dots come out larger than the dots on the file or plate. The printing term is dot gain or tone value increase: a 30% dot that prints as 50% has 20 points of gain. Part of it is physical (ink spreads into the paper and under press pressure), and part is optical: light scatters inside the paper around each dot, so dots absorb more light than their size suggests, an effect known as Yule–Nielsen.

Gain is largest in the midtones, roughly 40–60%, where dots have the most total edge for ink to spread from. Above that, dots touch and the free edge shrinks. Prepress software compensates with tone curves built for each press and paper, so a file prepared for one setup can print too dark on another.

Screen angles, moiré and rosettes

When two regular patterns overlap at nearly the same angle or frequency, they produce a third, larger pattern: moiré. It is the same beating effect you see when two window screens overlap. In four-colour printing there are four grids on top of each other, so their angles have to be chosen carefully.

A widely used set, quoted in an Eastman Kodak printing patent, is:

The logic behind it:

• 30° between the strong inks. Cyan, magenta and black sit 30° apart. At that separation the interference turns into a small, regular flower-like cluster called a rosette, fine enough to read as texture instead of a visible wave. The Kodak patent calls this a “visually pleasing rosette structure”.
• Only three fit. A square grid repeats every 90°, so a 30° spacing leaves room for exactly three inks. The fourth has to sit closer to one of them.
• Yellow takes the bad slot. Yellow is the lightest ink, so the patent places it at 0°, 15° from its neighbours, where its moiré is hardest to see.
• Black gets 45°. The same patent calls 0° “a highly noticeable angle”, which is why it goes to the lightest ink. The darkest ink goes on the 45° diagonal, conventionally treated as the least conspicuous angle.

Because a square grid looks the same after a 90° turn, 15° and 105° describe the same screen, which is why different software lists different numbers for what is effectively the same set. Registration matters too: the Kodak patent notes that misregistration can make the yellow screen’s moiré visible in certain composite colours. Your printer or RIP decides the final angles; designers rarely need to set them for process work.

FM and stochastic screening

FM (frequency-modulated) or stochastic screening turns the idea around: in classic (first-order) FM every dot is the same small size, and tone is controlled by how many dots appear in an area. The dots are placed pseudo-randomly, not on a grid.

• There are no screen angles, so there is no screening moiré and there are no rosettes.
• Misregistration between plates is still there but is less visible.
• The dots are tiny. The CUNY guide gives a range of 10–60 microns, roughly comparable to 480–120 lpi in AM terms.
• Tiny dots are harder to hold on plate and press. Wasatch’s RIP documentation notes that first-order stochastic screens are limited by dot gain, and that second-order screens grow the dot size through the midtones to flatten the dot gain curve.

Inkjet printing works in a similar way; the CUNY guide calls it virtually identical to FM screening, although many inkjets also vary droplet size. When people say “halftone effect” as a design style, though, they almost always mean the visible AM grid of round dots, not FM.

Halftone for print versus for screen

The same word covers two different jobs.

As a print process, the halftone is invisible infrastructure. You hand over a continuous-tone image at the right resolution, and the printer’s RIP screens it at the press’s ruling. You do not add dots yourself.

As a visual effect, the dots are the point. You make them deliberately large so they read as dots, then deliver the result as artwork. That works on screens and in print, with two caveats:

• On screen, every dot is drawn with pixels. If the browser or a social platform resamples the image, the dot grid beats against the pixel grid and moiré appears. Export at the exact display size, or use SVG so the dots stay geometric at any size.
• In print, a raster file full of grey anti-aliased dot edges gets screened again by the printer, so the RIP draws its own fine halftone inside and around your large one. Clean results come from dots that are solid ink: vector (SVG or PDF), or a 1-bit bitmap at the final output size, and a dot pattern far coarser than the press screen.

Starting points for designers

These are starting points to test and adjust, not measurements and not standards. Your printer’s specification always overrides them.

The print rulings in the first rows come from the typical-resolution table in Wikipedia’s halftone article, which itself is marked as needing citation; treat them as a sanity check before talking to the printer.

Common mistakes

1. Treating DPI, PPI and LPI as the same number. A 300 ppi image does not print at “300 lines”. Ask for the ruling, then size the image for it.
2. Halftoning the image, then scaling it. Resampling dots blurs their edges and creates moiré. Set the final size first, then apply the effect.
3. Saving the result as JPEG. JPEG compression smears hard dot edges into blocky noise. Use PNG for raster, SVG or PDF for vector.
4. Ignoring dot gain. A halftone that looks balanced on screen can print heavy in the midtones, especially on uncoated stock. Lighten midtones or use a coarser screen.
5. Placing two screens at the same or nearly the same angle. That produces broad moiré waves. Keep 30° between strong colours unless the wave is the effect you want.
6. Scanning printed photos without descreening. The scanner’s pixel grid interferes with the printed dots. Use the scanner’s descreen option or a source that is not already printed.
7. Going finer than the process. A 150 lpi file sent to a screen printer or a desktop printer can fill in. Match the ruling to the output.

How the generator on this site builds the effect

The halftone generator is a WebGL tool: a fragment shader runs once for every output pixel on the GPU. The approach is the standard one for a shader-based AM halftone:

1. Rotate into screen space. The pixel’s position is rotated by the screen angle, so the cell grid can sit at 45°, 15° or any other angle while the image stays upright.
2. Find the cell and sample its tone. The shader works out which cell the pixel falls in, rotates that cell’s centre back into image space, and reads the image there. Colour is converted to luminance with a weighted sum of red, green and blue, where green counts the most and blue the least, and ink coverage is one minus that luminance after contrast and brightness adjustments.
3. Size the dot by area. For round dots the radius is chosen so the dot’s area equals the ink coverage, r = cell × √(t ÷ π), until neighbouring dots touch at about 78.5%. Past that point the dot keeps growing towards the cell diagonal so that 100% becomes solid ink.
4. Check the neighbours. Dots in the shadows grow past their own cell, so for each pixel the shader evaluates the four nearest cell centres. That lets dots merge the way ink does instead of being clipped at cell borders. Edges get a narrow anti-aliased band so they stay smooth on screen.
5. Choose a shape. Round, square and diamond dots use the same sampling with a different distance function. The line screen interpolates stroke thickness between neighbouring cell centres, which gives engraving-like strokes.

The second ink works like a second printing plate. It samples the same image on its own grid with its own angle, by default 30° away from the first, its own coverage, and a pixel offset in X and Y. That offset imitates misregistration, the slight shift between plates on a press. On a real press misregistration is a defect; here it is a controlled effect, and the same offset-plate idea runs through the visual identity of iakzhin.com. The iakzhin.com case study shows how that print language is used across the site.

One conversion helps when moving from the tool to print: the cell size is set in image pixels, so the equivalent ruling depends on output resolution. At 300 ppi, an 8-pixel cell is 300 ÷ 8 = 37.5 lpi, which is a coarse, clearly visible screen.

If a project needs a halftone treatment planned for a specific press, paper or brand system, that is part of visual design work.