Polarizer on light and lens: glare reduction and light loss

Specular glare off glossy film, foil and shiny plastic buries print, labels and defects under a white flare that no exposure setting removes. The crossed-polarizer pair — a polarizing film on the light, a filter rotated 90° on the lens — blocks that mirror reflection while keeping the scattered light that carries the image. The debt is brightness: the pair passes roughly a quarter of the light a diffuse part returns, so compensation has to be planned, not discovered on the line. This guide gives you the loss budget, a worked example at belt speed, the levers that pay the light back, and the cases where a dome beats polarizers outright.

What crossed polarizers block, and what they let through

An LED light emits unpolarized light: the electric field vibrates in every plane across the beam. A polarizing film on the light passes only the component aligned with its axis, so the part is illuminated in one known polarization direction.

What the part does with that light decides the whole method:

• A mirror-like reflection keeps the direction. Glare from glossy film, foil or polished plastic arrives at the lens still polarized, only redirected.
• A diffuse return, scattered by paper, matte print or filled plastic, comes back largely depolarized, spread across all directions again.

The lens filter sits crossed, at 90° to the light film. Malus's law says a polarizer passes cos²θ of polarized light, so the crossed filter extinguishes the polarization-preserving glare almost completely. The depolarized diffuse return loses only about half its energy, so print, texture and defects survive at roughly a quarter of the unfiltered brightness. That asymmetry, killing the mirror component while keeping the scattered one, is the entire technique.

Two consequences follow. One film alone is a dimmer, not a glare killer: without the crossed analyzer nothing separates glare from feature. And no real film reaches perfect extinction, which is why a killed highlight leaves a faint grey residue rather than true black; grades built for deeper extinction pass less light when aligned, so the film choice trades residue against brightness.

The filter pairing and mounting mechanics are covered step by step in the cross polarizer guide on lens and light and in the crossed polarizer suppression article; what follows here is the light you lose and how to pay it back.

The light budget: two stops on a diffuse part

Common polarizing sheet passes roughly 40-45% of unpolarized light; higher-transmission grades polarize less cleanly, deeper-extinction grades pass less. Planning on a factor of two per stage is close enough to size a light. The second factor is set by the part: matte and fibrous surfaces depolarize strongly and pay the full ×0.5; glossy paint and brushed metal sit partway, so budget between ×0.3 and ×0.5; bare polished metal preserves the polarization, and the analyzer blocks the useful return along with the glare.

Two stops is a large debt: the gap between a 1 ms and a 4 ms exposure at the same aperture, or between f/4 and f/2. The stop-by-stop accounting against lens f-numbers is laid out in the polarizer light loss article; this guide takes the budget to a moving line.

Worked example: glossy print inspection on a 300 mm/s belt

The case: date codes and labels printed on snack packs in glossy film, inspected from above while the belt runs. Inputs, with their source:

• Belt speed 300 mm/s — line specification.
• Field of view 160 mm across a 1,600-pixel sensor, which is 0.1 mm per pixel — camera and lens geometry.
• Blur rule: keep motion blur under one pixel, 0.1 mm — print-inspection practice for readable edges.
• Baseline pulse 0.30 ms from a diffused bar light exposes the unfiltered image correctly — a bench value measured on the actual part for this case.
• Polarizer debt ×0.25 — the budget above: 0.5 at the light film, 0.5 at the depolarized return.

Rule in words: whatever light paid for the unfiltered image must be multiplied by 1 ÷ 0.25 = 4, paid in exposure time or in lamp output.

Arithmetic:

• Blur-limited pulse: 0.1 mm ÷ 300 mm/s = 0.33 ms.
• Pulse if the debt is paid in time: 0.30 ms × 4 = 1.20 ms.
• Blur at that pulse: 300 mm/s × 1.20 ms = 0.36 mm, which is 3.6 pixels — over the one-pixel rule by 3.6×, so time alone is rejected.
• Exposure can therefore contribute only 0.33 ÷ 0.30 ≈ 1.1× of the factor of four: nothing usable.

Decision: pay in aperture and lamp, not in time. Open the lens one stop, f/4 → f/2.8, for ×2, and double the lamp with a high-luminance bar in the same footprint class or a second bar for the remaining ×2. The pulse stays at 0.30 ms, so blur stays at 300 mm/s × 0.30 ms = 0.09 mm, inside the 0.1 mm budget. If depth of field will not give up that stop, take the whole ×4 from the light; paying by pulse alone only works below 0.1 mm ÷ 1.2 ms ≈ 83 mm/s belt speed.

Four levers that pay the light back

Work them in that order. Aperture first where focus allows: two stops of aperture covers the whole debt, with focus as the only side effect. Lit area next, because a wider or doubled light also evens the field over wrinkled film, which polarization alone never fixes. A higher-luminance light then buys intensity without moving the geometry. LED current comes last: output rises sublinearly with current while heat rises faster, so a lamp driven hard ages quicker than a larger lamp driven gently. On a static station the shutter is a fifth lever; the video exposure and frame rate article covers why frame-rate-limited captures behave differently from stills. As a bonus, the lens-side filter also trims partially polarized ambient glare, such as daylight off glossy wrap under skylights.

The high-luminance route in a compact footprint looks like this: the LS-BAR-0400 is a 312 x 30 mm high-luminance bar light.

Fitting the film and the filter

Fit the light-side film as the last optical element, on the outside of the diffuser. A diffuser works by scattering, and scattering partially depolarizes, so film mounted under the window loses its extinction and the crossed pair leaks a grey halo. Cut the film to the lit window, keep it flat, because a wrinkle rotates the local axis and prints banding into the image, and fix it outside the lit area.

Lens side, use a threaded polarizing filter on the front thread and rotate it while watching the worst glare: extinction sits at 90° to the returning polarization. Mark both axes once set, so a lamp swap or a lens clean does not silently rotate the pair apart. Both the light-side film and the lens filter are lighting accessories, not camera parts.

Run the light in strobe mode where the line allows. A short pulse puts far less heat into the film and the LEDs, which slows film fading and holds output stable across a shift. Pulse timing is its own fault source: when the strobed image comes back black, the strobe timing article walks that fault down, and lights with integrated strobe drivers move the pulse timing off your controller.

Where the pair fails: curved shiny parts and bare metal

Curved shiny surfaces defeat a fixed analyzer. Oblique reflection shifts polarization, and on a curve the local reflection geometry changes across the field, so one filter angle cannot extinguish glare everywhere: extinction moves along the curve and bright bands survive on either side of the killed zone. When the part is curved in the image plane — bottles, caps, handles, can ends — change the illumination geometry instead. A dome surrounds the part with diffuse light from every direction, so no single reflection concentrates into flare; the dome light guide sizes one for shiny, curved and wrinkled parts. Code reading on curved reflective lids fails the same way, and the reflective and curved QR article compares the lighting routes for it.

Bare polished metal preserves polarization through the reflection, so the crossed analyzer blocks the feature along with the glare and the part images nearly black. Metal wants the same diffuse treatment with reduced or no polarization; the shiny metal pins article works through that choice.

One apparent failure is a feature. Stretched transparent film rotates polarization, so a crossed backlight lights up stress and orientation in clear plastic; that transmission method has its own article on crossed polarizer backlight stress inspection. In reflection setups on thin wrap, the effect is usually too small to matter.

The decision rule: flat glossy dielectrics — film, foil, glossy print, paint — take the crossed pair. Curved shiny parts or bare metal take diffuse geometry first, and polarizers only trim what remains.

Choosing the light the film goes on

The film needs a flat window and headroom in output. Diffused bars suit elongated parts and film-wrapped trays: the window takes a cut film sheet directly, and the diffusion evens the field across wrinkles. Red output earns its keep when ambient light leaks in, because a matching bandpass filter on the lens turns the scene into near-darkroom conditions; the red LED and bandpass article shows that pairing.

The LS-BAR-1168 is a 423 x 40 mm diffused red bar light.

Ring lights suit compact stations where the filter must hug the lens. The LS-RNG-0517 is a 120 mm dia., 0° diffused ring light; the LS-RNG-0143 is a 100 mm dia. ring light with diffuser and 30° angle.

For the wider geometry call, the ring light vs bar light comparison sorts the two by part shape and glare, and the bar lights and ring lights categories carry size, colour and angle filters to narrow the field. Size the lamp to the polarized budget, not the unfiltered one: the pair gives back only a quarter, so a light that barely worked before the film will not work after it.

Set-up checklist

• Confirm the glare is specular. Tilt the part: a hot spot that slides across the surface is a mirror reflection and polarizers will kill it; one that stays put is a diffuse bright region and no filter angle removes it.
• Film outside the diffuser, flat and wrinkle-free. Scattering under the film depolarizes the light and the pair leaks; wrinkles print banding.
• Set extinction on the worst glare, not the field centre. Rotate the lens filter while watching the flattest offender, then mark both axes.
• Measure the drop before choosing compensation. Capture with and without the pair and divide pixel levels: near ×0.25 confirms a normal diffuse part; much deeper warns that the feature itself is polarized.
• Recompute the pulse against blur before wiring. pulse ≤ pixel size ÷ belt speed; at 0.1 mm per pixel and 300 mm/s that is 0.33 ms.
• Strobe where the line allows. Less heat into film and LEDs, stable output, sharper freeze.
• Curved shiny or bare-metal part? Change the geometry instead. Domes and diffuse bars win where a fixed analyzer angle cannot.

Next step: size the light to the polarized budget. The shop finder filters bar and ring lights by size, colour and angle, the light configurator steps through camera, lens, light and controller together, and the rest of the lighting methods sit on the lighting topic page. If the part refuses polarizers, start from the dome guide above and fix the geometry first — that call, made before ordering, is what keeps a glare project from ending as a drawer of unused filters.