Lighting for glossy packaging inspection: stop the glare

A varnished, multicolor carton glints from every angle: tilt the box a few degrees and the hot spot jumps to another panel, so no single ring, spot or coaxial position clears the image. The mechanism is one mirror law, and the fix is one family: replace small directional emission with a large-area diffuse source, then control what the lens collects. This guide gives you the glint geometry behind the glare, sizing arithmetic for dome and flat diffuse lights, a polarizer light budget, and a strobe calculation for a moving belt. The fixes, in the order we reach for them on the bench: a dome light for curved or multi-panel cartons, a flat shadowless light for flat faces, crossed polarizers for foil and mirror varnish, then strobe and ambient control.

Why a varnished carton glints from every angle

A glossy varnish is a mirror. The law of reflection that separates bright field from dark field sends the camera's own direction straight back from every facet, so each tilted panel paints one bright lobe, and the lobe direction sits at twice the panel tilt measured from the camera axis. A folded carton carries dozens of different panel normals, which is why some panel glints at every setting.

Each source you tried failed for its own geometric reason:

• A ring light emits from a small area at one elevation. Only panels whose normals bisect the camera-to-ring angle mirror it into the lens, so you get a glare stripe that moves as the box rotates.
• A coaxial light sits on the camera axis, so flat facets of varnish mirror it straight back into the lens: a full-field burn.
• A small dome mounted close to a tall carton loses part of its emission cone to the curved wall, and what reaches the part behaves like a narrow ring. The dome concept was right; the size and mounting height were not.

If a blue ring looked best in early trials, that was ink contrast, not glare reduction: blue inks stay bright under blue light while red and yellow inks darken, so the artwork separates even though the lobe is unchanged. Rank the causes in this order when you troubleshoot:

1. A directional source (ring, spot, coaxial) on a multi-panel mirror target.
2. A diffuse source that is undersized or mounted too close, so its angle span is truncated.
3. Metalized foil or a polarization-keeping varnish that survives diffuse light alone.
4. Room light and motion smear misread as glare, collected by the long exposure a weak light forces.

Size the diffuse source before you order it

Two numbers size the light: the glint angle span and the part outline. Coverage comes first. A face tilted 20° from flat mirrors the vertical camera direction to 2 × 20° = 40° off axis, so the emitting surface must wrap that span or those facets see black and print contrast collapses at the folds. Dome lights wrap roughly 0-90° of elevation. A flat panel covers about ±45° when its width is twice the inspection window at the same mounting height, which is the sizing rule for flat sources.

Emitting area comes second, and a starting rule of twice the carton's largest dimension covers the three things that eat margin: the camera hole cut into the middle of the emitting surface, the clearance that keeps folded panels off the walls, and the grazing facets near the field edge. Work the rule on a 60 mm carton: 2 × 60 mm = 120 mm of emitting area. A Ø116 mm emitting surface sits just under that target; the next class up, a 194 mm outer diameter, clears it with room to spare. Height matters as much as diameter on a tall carton, because the same 2θ rule pushes upper-panel glint directions toward the top of the cavity.

Fix 1: wrap the carton in a dome light

A dome is a diffuse cavity that emits inward across roughly a half-hemisphere. Every glint direction on a folded carton lands inside that span, so the mirror image dissolves into a faint even background and the inks carry the contrast. Mount the carton near the vertical center of the cavity and keep every panel clear of the wall; a panel leaning against the glass loses its grazing light and goes dark.

One direction stays dark: the lens cone cut into the dome. Oversizing thins the shadow, offset mounting moves it off the print, and a second diffuse channel fills it; the fixes for the dark spot from a dome's camera hole cover all three. For curved, wrinkled or multi-panel parts beyond cartons, the guide to dome lights for shiny, curved and wrinkled parts goes deeper.

For the 60 mm carton from the sizing section, the rule asks for a 120 mm emitting area. The LS-DOM-0085 provides a Ø116 mm emitting area inside a Ø166 mm body. The LS-DOM-0062 provides a 194 mm outer diameter for cartons with deeper folds or taller walls. Both run from 24 V.

Keep white light for multicolor print on a monochrome camera. The category also lists 850 nm and 940 nm infrared variants for the day one ink must vanish or the line runs under bright visible ambient.

Fix 2: a flat shadowless light for flat faces

When one flat face carries the print and headroom is short, flat diffuse beats a dome: the same spreading effect at a fraction of the height. The shadowless lights family covers flat domes with a central camera window, square panels and arches for long packs. Keep the panel about twice the window width and mount it as close as the folding hardware allows; that ratio is what buys the ±45° span.

The LS-FDL-0002-L8 flat dome lights a 251.5 x 251.5 mm area. The LS-SDL-0006 square shadowless light measures 120 x 120 x 30 mm.

That 30 mm profile slides under a folding head where a dome cannot. Elongated packs want an arch instead: the LS-SDL-0026 arch shadowless light opens a 204 x 110 mm window.

Fix 3: crossed polarizers for foil and mirror varnish

Some cartons carry metalized foil or a varnish so mirror-flat that even a large diffuse source leaves a soft lobe on the critical panel. Polarization finishes the job. Put a linear polarizing film over the light and a crossed analyzer on the lens: mirror reflection keeps its polarization and is blocked, while paper and ink scattering depolarizes the light and passes. The cross-polarizer arrangement on lens and light shows the mounting and tuning steps.

Budget the light before you commit. A typical polarizing film passes about 35-45% of unpolarized light, and depolarized target light splits evenly between the axes, so the crossed analyzer passes roughly half of what survives: 0.40 × 0.50 = 0.20, about a 5x cut in usable signal. Specular energy falls by far more than that; the asymmetry is the entire point. Buy the print signal back with a brighter source, a shorter working distance or a strobe pulse, and match the film to the wavelength you run, because polarizing film performance shifts with wavelength.

Strobe the light and cap the exposure

Diffuse cavities waste most of their light: only a narrow cone from each emitting point reaches the lens. On a moving belt that inefficiency pushes exposure times up until the print smears. Strobe instead: a short, high-current pulse with the camera exposure clipped to it.

Worked example: a 120 mm window over 2448 pixels gives 120 mm / 2448 = 0.049 mm per pixel, so the blur budget is 0.5 × 0.049 mm = 0.025 mm. At a belt speed of 300 mm/s, t_max = 0.025 mm ÷ 300 mm/s = 0.000083 s, about 83 µs. Fire an 80 µs pulse and the smear is 300 mm/s × 0.000080 s = 0.024 mm, inside budget. Switch the camera to manual exposure before you trust the number; the maximum auto exposure cap article explains how auto exposure quietly re-extends itself on a belt.

A strobe cleans up ambient light too, because the sensor collects room light only during the pulse: 5 ms ÷ 83 µs = 60, so a strobed frame gathers about 60x less ambient than a 5 ms continuous exposure. The same discipline carries to robot cells, as the ambient light control for pick-and-place vision article shows.

Drive the pulse from a strobe-capable controller: the LS-CTL-0070 is a 4-channel, 200 W digital control box with RS232. The light controllers category and the controller finder filter strobe models by channel count and output type.

Keep the pulse within the light's rated duty cycle. Peak current buys brightness; average current heats the LEDs.

When diffuse is the wrong answer

Diffuse light inspects print. If the target is the coating itself, such as scratches in the varnish, dents in the foil or glue beads at the seam, diffuse lighting evens the surface out and hides exactly what you hunt. Low-angle dark field inverts the result: grazing light turns every topology break into a bright line on a black field. The multi-angle sequential method for scratches on glossy parts and the grazing-light versus dome comparison for engraved text on metal show the technique on comparable surfaces.

Decide by defect type first and source second: print, OCV and code work go diffuse; topology work goes grazing.

Bench order and the decision

Run the bench in this order and stop when the image holds:

1. Tilt test: rock the carton ±20° under the current light and mark where lobes cross the lens. Sweeping lobes confirm a directional source.
2. Fit the largest diffuse source the station allows: a dome for folded or curved cartons, a flat shadowless light for one printed face.
3. If a soft lobe survives on foil or varnish, add crossed polarizers and re-budget for the 5x light cut.
4. Strobe at the pulse width the belt speed allows and set manual exposure to match.
5. Only now judge wavelength: white for multicolor print, a single color or infrared channel to drop one ink.

The decisions by case:

• Folded or curved carton, print inspection: a dome sized to at least twice the largest dimension, strobed, white light.
• Flat face, wide window, headroom under about 100 mm: a flat shadowless light at twice the window width.
• Metalized foil or mirror varnish: flat diffuse plus crossed polarizers plus a strobe.
• Scratches, dents or glue in the coating: low-angle dark field with the diffuse source off.

The light finder filters both light categories by emitting size and wavelength, and the configurator at the light step pairs the light with camera and lens; for a belted line start the configurator at the controller step. If the geometry is unusual, send the carton's largest dimension, folding depth and line speed through the contact page and we will size the source with you.