Halftoning & Screening: Turning Tone into Dots

Look at a photo on your phone and you see a smooth, continuous slide from light to dark — millions of shades. Now look at that same photo printed in a newspaper through a magnifying glass, and the smoothness vanishes: it is just a field of tiny black dots, big in the dark areas, small in the light areas, sitting on white paper. This section explains the trick that bridges those two worlds. It is one of the oldest and cleverest ideas in all of printing.

7.1 Why a press cannot print "continuous tone"

Before we define anything, let's name two terms.

Continuous tone (contone)

An image where tone changes smoothly with no visible steps — like a photographic gradient or what you see on a glowing screen.

Halftoning (also called screening)

The process of converting a continuous-tone image into a pattern of dots that, viewed from a normal distance, the eye reads back as smooth tone.

Here is the core problem. A printing press is fundamentally binary: at any single spot on the paper, ink is either there or it isn't. A cyan ink film prints at one fixed density — it cannot lay down "50% cyan" as a thinner, paler film the way a screen simply glows dimmer. The ink is full strength or absent.

So how do you show a pale sky or a soft shadow with an ink that only knows "on" or "off"? You break the image into a grid of tiny dots and vary how much of the white paper those dots cover — the area coverage, or tint percentage. Tiny dots floating on white paper read as a light tone; dots so fat they nearly touch read as a dark tone. 0% coverage = paper white; 100% = solid ink.

The magic is in your eyeball. At normal viewing distance your eye cannot resolve the individual dots, so it averages — it blends the black dots and the white gaps into a single gray. This is called optical mixing or spatial integration. (The technique was first patented by William Henry Fox Talbot in 1852 and refined into the glass cross-line screen later in the 1800s.)

7.2 Two ways to vary the dots: AM vs FM screening

There are two fundamentally different strategies for arranging the dots.

AM — Amplitude-Modulated (conventional) screening

The dots sit on a fixed, regular grid at fixed spacing; only the dot size varies to change tone. ("Amplitude" means size.) Highlights get tiny dots, shadows get big dots, but every dot lives on the same evenly-spaced lattice. This is the traditional screen and still the default for most offset work — it gives very smooth, predictable, stable midtones. Its one weakness: that regular grid is exactly what causes moiré when several colors overlap (more on that below), which is why the plates must be angled.

FM — Frequency-Modulated (stochastic) screening

Here the dot size is fixed and tiny; only the spacing and number of dots varies (the "frequency"). Highlights get a few scattered microdots; shadows get a dense crowd of them. Placement is pseudo-random. The microdots are roughly 10–30 µm (microns) across — often cited around 20–30 µm. Because the pattern is random, there is no regular grid, so no LPI and no screen angle — and therefore moiré and rosettes essentially disappear. That makes FM excellent for tricky subjects (fabric weaves, fine detail, sharp diagonal lines) and for 6- or 7-color extended-gamut printing. The cost: the tiny dots are fragile, harder to hold on press, more prone to dot gain, and can look slightly grainy.

AM (size varies, grid fixed)        FM (spacing varies, size fixed)
  highlight    shadow                 highlight       shadow
  . . . .      @ @ @ @                 .   .  .        ..::..::
  . . . .      @ @ @ @                  .  .            .:..::.:
  . . . .      @ @ @ @                 .    . .         :.::..::
  (same lattice, dots grow)           (random scatter, count grows)

Hybrid / XM (cross-modulated) screening

Modern CTP (computer-to-plate) RIPs often blend the two: AM in the midtones for stability, switching to FM in the extreme highlights and shadows where AM dots would be too small to hold or too merged to separate. Best of both worlds.

7.3 Screen frequency — LPI (AM only)

LPI (lines per inch) is the number of rows of halftone dots packed into one inch — also called the ruling, line screen, or screen frequency. Higher LPI = finer, smaller, more numerous dots = more detail, but only if the paper and press can hold those tiny dots without smearing them together.

Coarser, more absorbent paper forces a lower LPI (otherwise the dots "plug" and merge into mud); smoother coated stock permits higher LPI.

The relationships everyone confuses: LPI vs PPI vs DPI

These three look alike and trip up beginners constantly. Keep them straight:

PPI (pixels per inch)

The resolution of your image file at its final printed size.

LPI (lines per inch)

The fineness of the halftone screen the press uses.

DPI (dots per inch)

The resolution of the output device (the platesetter/imagesetter) — how many tiny laser spots per inch it can place.

Image resolution rule (the "quality factor"): supply images at 1.5× to 2× the LPI in PPI at final size. So for a 150-LPI job, use 300 PPI images (2× — this is the origin of the famous "300 dpi for print" rule of thumb). 1.5× (~225 ppi) is the practical minimum; 2× gives a safety margin.

Why platesetters run at 2400+ DPI while the screen is only 150 LPI: each halftone dot is itself built from a little grid of device spots, and you need many spots per dot to vary the dot's size in fine steps. The number of reproducible gray levels is:

gray levels = (device DPI / LPI)^2 + 1

  2400 dpi / 150 lpi = 16
  16^2 + 1 = 257 gray levels   (>= 256 wanted for smooth tone)

That formula is why a 2400-dpi platesetter is needed to drive a mere 150-lpi screen smoothly. Never confuse the device's DPI with the screen's LPI — they are different things.

7.4 Dot shape and the "50% problem"

Halftone dots come in several shapes: round, square, elliptical (oval / "chain dot"), and diamond. A dot keeps its chosen shape until it grows big enough to touch its neighbors — and when it touches is where the drama happens.

As dots grow toward dark tones, at around 50% coverage adjacent dots can suddenly connect. The moment they join changes the geometry abruptly, producing a visible jump or step in midtone gradients (banding) — the "dot-gain cliff."

• Round dots: all four neighbors connect almost simultaneously near 50% → the most abrupt, most violent midtone jump (and a visible "chain" look in midtones).
• Square dots: corners touch first; very sharp, snappy detail, but still a hard transition through 50%. Past 50% the shape inverts to negative white dots.
• Elliptical / chain dots: the oval connects along its short axis first, then later along its long axis — two gentle transitions instead of one harsh one → the smoothest midtones. This is the most popular choice for general commercial work and for skin tones.

Tone growth (highlight -> shadow):
  o   o   o        round dots, separate (highlight)
  oo  oo  oo       growing
  ((  ((  ((       ellipse joins SHORT axis  (~40%)
  ##  ##  ##       midtone
  joins LONG axis later (~60%)  -> inverts -> solid

7.5 Screen angles — why every plate is rotated

Now the heart of color halftoning. If two screens with the same LPI overlap at the same angle, their grids beat against each other and create a large, ugly, wavy interference pattern called moiré (think of a faint plaid shimmer). Even a hair of misregistration makes it worse.

The fix is to rotate each color's screen to a different angle so the grids interleave instead of clashing. The standard US CMYK set is:

The key idea: the three strong, dark inks (C, K, M) are spaced 30° apart (15 / 45 / 75). A 30° separation is the "magic number" that shrinks any leftover moiré to its smallest, least-visible scale. Black gets 45° because it's the most visible ink and the diagonal orientation is where our eyes least notice a regular pattern. Yellow gets 0° because it's so pale that even though it sits only 15° from cyan and magenta (breaking the ideal 30° rule), the resulting moiré is essentially invisible.

Rosettes — the "good" pattern we are aiming for

When all four correctly-angled screens print over each other, the dots arrange into tiny, repeating, flower-like clusters called rosettes. This is the intended, healthy result — it's what your eye blends into smooth color. There are two kinds: open-centered (a small white hole in the middle — generally preferred and more stable) and closed / dot-centered. Moiré is bad (large and distracting); a rosette is good (small and designed). The entire angle system exists to guarantee a clean rosette instead of moiré.

Bad: same-angle grids        Good: angled -> rosette
  beat into MOIRE              tidy repeating flowers
  ~~~~  ~~~~  ~~~~              o o o o   (open-centered:
   ~~~~  ~~~~  ~~~~             o(.)o(.)   small white hole
  ~~~~  ~~~~  ~~~~              o o o o    at each center)
  (wavy plaid shimmer)

7.6 Trapping — overlaps that hide misregistration

Two more terms first:

Registration

The perfect alignment of all four color plates on top of each other.

Misregistration

When plates print slightly off (from paper stretch, web tension, plate/blanket movement, or press mechanics). Without protection, this leaves slivers of bare white paper peeking between two colored areas.

Trapping is the prepress fix: deliberately overlapping adjacent colors by a tiny amount so that even when registration drifts, no white gap (or wrong-color fringe) shows. There are two moves:

• Spread ("fattie"): make the lighter foreground object slightly bigger so it bleeds into the darker background.
• Choke ("skinny"): make the lighter background opening slightly smaller so the darker object overlaps it.

Golden rule: always spread the lighter color into the darker color. The light ink contaminates the dark ink the least, so the overlap is the least noticeable. Trap width is tiny — typically about 0.25 pt (~0.003–0.004 in, or 0.08–0.1 mm).

Two related terms: a knockout removes the background where the top object sits (no ink mixing, but it requires a trap); an overprint lets the top object print right over the background (e.g. black text overprints colored areas so no trap is needed at all). Modern RIPs perform trapping automatically.

7.7 Dot gain (TVI) — and how it ties everything together

Dot gain means the printed halftone dot ends up larger than it was in your file, so the print looks darker and muddier than intended. ISO calls it TVI (Tone Value Increase). It has two causes: mechanical gain (ink physically squashed wider under press and blanket pressure, plus absorption into the sheet) and optical gain (light scatters sideways under the ink edge inside the paper, making each dot look bigger than it physically is).

Dot gain is greatest in the midtones (~50%), because that's where dots have the most total perimeter — the most edge available to grow from. It's classically measured at the 40% and 80% tints, with 50% as the headline figure.

Notice how this section's earlier ideas all connect to dot gain:

• Higher LPI = more dot gain (smaller dots have more perimeter relative to area, so they fatten proportionally more) — another reason coarse papers are forced to lower LPI.
• FM/stochastic = more dot gain (microdots are almost all perimeter) — so FM must be characterized carefully.
• Dot shape spikes gain at 50% (the touch-point is the tonal-jump cliff from §7.4).

The fix — compensation: the RIP applies a dot-gain compensation curve that deliberately shrinks the dots in the file so that, after the press fattens them, they land at the right size. The correct curve is determined by printing to a standard — ISO 12647-2, G7, or GRACoL — and measuring a control strip with a densitometer or spectrophotometer.