Skip to main content

05 / 05   Stray light & flare

Application note

Stray Light, Flare, and Contrast Loss

Trace unwanted light paths in imaging and display systems before choosing a measurement or mitigation.

Useful image formation directs light to the intended location. Stray light reaches a detector or viewer through an unintended path. A bright scene feature can wash out a camera image; a bright display region can contaminate a neighboring dark region. The mechanisms differ by system, but the shared engineering task is to identify the source, coupling path, and affected measurement.

Imaging: distinguish veiling flare from structured ghosts

Scattering from surfaces, contamination, and internal structures can add a broad veil that reduces scene contrast. Multiple reflections can form localized ghosts, arcs, or recognizable images of a bright source. Sources outside the nominal field may still couple through the entrance aperture, especially without effective baffling.

Measure a dark region while varying a controlled bright source’s angle, position, and level. Fix exposure, focus, aperture, and processing; check saturation and sensor blooming separately. The measurement lens, cover window, and fixtures can also contribute. A flare map is therefore more informative when accompanied by the source geometry and a reference exposure.

Display: distinguish optical spread from drive behavior

Bright-to-dark leakage can arise in projection optics, scattering layers, cover structures, or the screen’s angular response. A curved screen can redistribute reflected light between regions. These paths can raise simultaneous-pattern black levels even when the full-black state remains dark.

Use bright patches or complementary patterns to map the spatial and angular response. Keep average picture level and local-dimming behavior documented. A halo from zone control and stress-induced LCD leakage are different mechanisms from optical flare, even when their images look similar. Also measure instrument flare: a camera viewing a high-contrast display can create an apparent display defect of its own.

A bright scene source couples unwanted light into an imaging detector, while a bright display region couples unwanted light toward a dark region and the viewer. Both paths require separating device and instrument effects.

View diagram at full size

Quantify the contrast penalty

For an illustrative uniform additive veil LsL_s applied to high and low luminances:

CM,s=LH−LLLH+LL+2Ls,CRs=LH+LsLL+LsC_{M,s}=\frac{L_H-L_L}{L_H+L_L+2L_s},\qquad CR_s=\frac{L_H+L_s}{L_L+L_s}

The level difference stays fixed while the background rises. A nonuniform ghost requires a spatial model; the simple veil cannot represent its shape. Do not label every brightness gradient “flare”: relative illumination primarily describes how intended light is distributed, while flare adds an unwanted contribution.

The same additive veil model reduces Michelson contrast and white-to-black ratio in a captured image and an observed display, while a localized ghost requires a spatial model.

View diagram at full size

Describe directional scatter with BRDF and BTDF

The bidirectional reflectance distribution function relates reflected radiance in one direction to incident irradiance from another:

fr(θi,ϕi;θo,ϕo)=dLr(θo,ϕo)dEi(θi,ϕi)f_r(\theta_i,\phi_i;\theta_o,\phi_o)=\frac{dL_r(\theta_o,\phi_o)}{dE_i(\theta_i,\phi_i)}

BRDF has units sr⁻¹. It is not simply a reflectance percentage: it describes where reflected power goes. A Lambertian reflector has fr=ρ/πf_r=\rho/\pi; a glossy surface may concentrate light near the specular direction. Wavelength and polarization can also change the response.

The same surface-scatter description applies to imaging and display components. BRDF describes reflected directions, BTDF transmitted directions; internal lateral propagation requires additional geometry.

View diagram at full size

BTDF describes transmitted angular scatter. Neither a surface BRDF nor a single reflectance number alone describes lateral propagation inside a panel, bulk scattering, or repeated internal reflections. Those require the relevant material and geometric model.

From directional illumination to reflected luminance

For a fixed outgoing direction ωo\omega_o, add contributions from the incident hemisphere:

Lr(ωo)=∫Ωifr(ωi,ωo)Li(ωi)cos⁡θi dωiL_r(\omega_o)=\int_{\Omega_i} f_r(\omega_i,\omega_o)L_i(\omega_i)\cos\theta_i\,d\omega_i

LiL_i is incident radiance, dωid\omega_i is a solid-angle element, and θi\theta_i is measured from the surface normal. The cosine converts illumination arriving from an oblique direction into its contribution per unit surface area. Compatible photometric quantities may be used when the spectral weighting is treated consistently.

For a Lambertian surface, fr=ρ/πf_r=\rho/\pi is independent of direction. Surface illuminance is already the incident angular integral, so:

E=∫ΩiLi(ωi)cos⁡θi dωi,Lr=ρEπE=\int_{\Omega_i}L_i(\omega_i)\cos\theta_i\,d\omega_i, \qquad L_r=\frac{\rho E}{\pi}

If EE is measured on the display or target surface, do not multiply by another incidence cosine. If illuminance is instead specified on a plane perpendicular to a collimated beam, the surface receives E⊥cos⁡θiE_{\perp}\cos\theta_i. This distinction prevents counting the projected-area factor twice.

For a directional surface, equal illuminance under two different angular source distributions need not give equal luminance toward the viewer. The relevant BRDF and source geometry must remain in the calculation.

For both imaging and displays, mitigation follows the identified path: baffling, coatings, cleanliness, absorptive structures, optical redesign, or changes to drive control. Verify the result with the same source and collection conditions used to reveal the failure.

Iris Optics can help isolate optical and electronic contributions, design stray-light measurements, and evaluate corrective changes for camera, projection, and display systems.

Explore the model

Contrast loss from veiling light

Add a uniform pedestal to a camera scene or a display measurement and see how contrast falls.

Change the values to update the results and visualization.

Original contrast ratio
200:1
With veiling light
34.167:1
Original Michelson contrast
99.005 %
After-pedestal Michelson contrast
94.313 %
Modulation retention
95.261 %

The absolute bright–dark difference is unchanged, but its ratio to the raised baseline is smaller. Camera scene contrast and display contrast can both suffer this effect.

Contrast loss from veiling light: interactive plotLuminance / cd/m² against Location: dark (0) → bright (1). Original levels; With equal veiling pedestal.000.245.10.490.20.61350.81801226Location: dark (0) → bright (1)Luminance / cd/m²
Original levelsWith equal veiling pedestal
Formula & assumptions

CRafter = (Lhigh + Lv)/(Llow + Lv); CM,after = (Lhigh − Llow)/(Lhigh + Llow + 2Lv).

Equal additive luminance at the bright and dark locations, with no clipping or exposure adjustment. This isolates a contrast mechanism; real camera ghosts, display halo, scatter and instrument flare can be spatially varying. The entered pedestal is an assumed or independently measured contribution, not a prediction from a lens or panel design.

Explore the model

Diffuse reflection and incidence

Relate a surface’s Lambertian BRDF to reflected luminance, for a display cover or an illuminated object seen by a camera.

Change the values to update the results and visualization.

Illuminance on surface
707.11 lx
Lambertian BRDF
0.012732 sr⁻¹
Reflected luminance
9.0032 cd/m²

This incidence-angle falloff comes from projected illumination. It is not a cosine falloff in the viewing-direction luminance of a Lambertian surface.

Diffuse reflection and incidence: interactive plotReflected luminance / cd/m² against Incidence angle from normal / degrees. Fixed beam-normal illuminance.00182.8365.6548.47211.29014Incidence angle from normal / degreesReflected luminance / cd/m²
Fixed beam-normal illuminance
Formula & assumptions

Esurface = Ebeam cos θi; fr = ρ/π sr⁻¹; Lreflected = fr Esurface.

A collimated beam and ideal diffuse, nonfluorescent reflection. Ebeam is measured on a plane normal to the beam; do not apply the cosine again to illuminance already measured on the surface. The plot varies incidence, not viewing angle: ideal Lambertian luminance is independent of viewing direction. Real BRDF data depend on illumination/view angles and wavelength, and may include a specular lobe.

From estimate to application

Identify the light paths limiting performance.

Iris Optics can help isolate unwanted light in an imaging or display system and develop a reproducible test to evaluate mitigation.

Discuss your stray-light problem