What is a halftone? How dots print a photograph
A halftone prints a photo with dots of one ink that change in size. How the screen works, what frequency, angle and dot shape mean, and who invented it.
Published
A halftone is a picture made of dots of a single ink, arranged in a regular grid, where the size of each dot follows the darkness of the picture at that point. Large dots that nearly touch read as dark, small dots with paper between them read as light, and from reading distance the eye blends them into greys that are not really there. It is how a press, which can only print ink or no ink, prints a photograph.
GIMP’s manual gives the general definition: halftoning is “the process of rendering an image with multiple levels of gray or color (i.e. a continuous tone image) on a device with fewer tones; often a bi-level device such as a printer or typesetter” (GIMP Newsprint filter).
10%
30%
50%
70%
90%
How the dots make a tone
Every dot is the same full-strength ink. What changes is how much of the paper the ink covers, which printers call the tone value and give as a percentage. A 30% tone has ink on 30% of the area.
For round dots on a square grid the sums are short. A dot of radius r in a cell of width p covers π × r² ÷ p² of the cell:
- a 10% tone needs a radius of 0.18 of the cell width
- a 50% tone needs 0.40 of the cell width
- at half the cell width the dots touch their neighbours, which happens at π ÷ 4, or 78.5%
Above that the dots merge and the white paper is left as small pinched shapes, as in the last square of the figure. Because area goes with the square of the radius, a dot twice as wide covers four times the area.
The three settings of a screen
The grid of dots is called the screen. Three numbers describe it.
Frequency is how many rows of dots fit in an inch, in lines per inch (lpi), or in a centimetre. The printing standards call it screen ruling or screen frequency. ISO 12647-2 (2004 edition) gives 45 to 80 lines per centimetre for four-colour offset printing, which is 114 to 203 lpi, and ISO 12647-5 gives 20 to 40 per centimetre, 51 to 102 lpi, for screen printing. The paper matters: ISO 12647-1 notes that “rough-print substrates require coarser screens than smoothly coated ones”. At 85 lpi a cell is 25.4 ÷ 85 = 0.30 mm wide; at 150 lpi it is 0.17 mm.
Angle is how far the grid is turned. Stephen Horgan, who worked on newspaper halftones from the 1870s, wrote in 1913 that the 45° alignment of the dots “is selected because it was found at that angle they were least objectionable. Were they arranged, for instance, horizontally and vertically, they would be most obvious” (Horgan’s Half-tone and Photomechanical Processes). In colour printing each ink gets its own angle: CMYK halftone goes through the rule.
Dot shape is usually round, square or elliptical. Those three are the shapes that ISO 12647-2, the standard for offset printing, allows in its 2004 edition. The ISO texts quoted on this page were read as adopted, word for word, by the Bureau of Indian Standards: IS 15963-1, IS 15963-2 and IS 15963-5.
How much detail a halftone holds
A halftone trades detail for tone. GIMP’s manual puts it plainly: each cell of the grid holds a single spot, and “a large cell size results in a heavy loss in resolution”. That sets how many pixels the photo needs. DPI vs PPI gives the usual rule and goes through the units.
There is a second limit. A printer builds each halftone dot from its own much smaller machine dots, and the more of them fit in a cell, the more sizes the dot can take. ISO 12647-2 asks for enough resolution “to permit the reproduction of at least 100 tone-value steps”, and gives the example of a screen of 70 lines per centimetre, which needs at least 700 dots per centimetre (1,800 dpi). That is ten machine dots across a cell, and 10 × 10 = 100 in its area. By the same sum a 600 dpi printer has ten dots across a cell at 60 lpi.
Dot gain
Ink spreads when it meets paper, so a printed dot is larger than the dot that was asked for. The standards call this tone value increase. ISO 12647-1 gives the example of a patch that is 40% on the film and measures 55% on the print: an increase of 15%. The same standard notes that it was “formerly known as dot gain”. Halftones for screen printing covers how screen printers allow for it.
Who invented it
The account here is Horgan’s, from the 1913 book quoted above; he was one of the people involved, and the claims about his own work are his.
The idea of putting the pattern of a screen into a photographic printing plate goes back to the 1850s. Horgan records that William Henry Fox Talbot’s patent of 29 October 1852 impressed “the image of a piece of gauze” on the gelatin coating of the plate before the picture, and counts it as the principle behind rotary photogravure.
Horgan himself began experimenting in 1876 and, by his own account, “devised a practical half-tone process which was first shown in the New York Daily Graphic, of March 4, 1880, with a picture titled ‘Shanty Town’”. His first screens were perforated cardboard.
The cross-line screen in use when Horgan wrote came a few years later. He credits it to “Mr. Frederick E. Ives, who in the winter of 1885-6 sealed two single-line screens together, and to Mr. Max Levy, who in 1893 perfected the manufacture”. Horgan describes it as an opaque surface pierced by uniform square openings, made by cementing two ruled glass screens together at right angles. The photograph was copied in a camera through this grid, and the light passing through the openings formed, in his words, “dots of different shapes and areas” on the plate.
Today software works out the dots, either in the printer’s raster image processor or in an image editor.
Halftone, dithering and stippling
All three show tone with marks of one colour. They differ in what changes.
| Method | Where the dots sit | What follows the tone |
|---|---|---|
| Halftone | On a regular grid | The size of each dot |
| Dithering | On the pixel grid | How many pixels are on |
| Stippling | Anywhere, no grid | How close the dots are |
What is dithering? covers the second and stippling the third. The line between them is not sharp: the GIMP 2.8 manual says its halftone filter “uses a clustered-dot ordered dither”.
The halftone effect in software
On a screen, a halftone is no longer needed to show a photo, so it has become a look: large, obvious dots that call up newsprint and comics. Most image editors can make it.
- GIMP has Filters > Distorts > Newsprint.
- Krita has a Halftone filter that, in its manual’s words, uses “simple shapes that vary in size”.
- Affinity Photo 2 has a Halftone filter with Monochrome, Color, Line and Circular screens.
- LightBurn has Newsprint and Halftone image modes for lasers, described in LightBurn image modes explained.
Plotcaster’s editor does not draw this kind of halftone, where dot size changes on a fixed grid. Its Dither style is the nearest: marks of one size on a grid, more of them where the photo is darker, saved as an SVG in which each dot is a small pen-drawn loop, or as a 1-bit PNG. The Bayer algorithms, such as Bayer 8×8, give a regular, screen-like pattern. For true size-changing dots there is a free set of halftone patterns: gradients, flat tints and radial fades as vector files. Vector halftone covers making your own, and comic book halftone the look with large dots.
Questions
Can a halftone be made of lines? Yes. A line screen uses parallel lines that swell and thin in place of dots. The glass screens Horgan describes were made from two single-line screens crossed, and GIMP’s Newsprint filter offers Line beside Circle and Diamond. The free set has line halftones too.
Is a halftone the same as pixels? No. Pixels are equal squares that each hold a colour. Halftone dots are all one colour and differ in size.
Why do scanned magazine photos show odd patterns? The scanner’s pixel grid and the printed screen are two regular grids laid over each other. ISO 12647-1 defines a moiré pattern as an “unwanted periodic structure produced by interference between two or more two-dimensional periodic structures”.