A uniform input should produce a predictable spatial output. For a camera, that means imaging a uniform radiance field; for a display, it means driving a uniform white or gray image. Smooth falloff, stripes, and localized patches can look similar in the result while arising from different physical mechanisms. A useful test separates those mechanisms before applying compensation.
Imaging: distinguish illumination from sensor response
Lens relative illumination compares image-plane irradiance across the field, commonly against the center:
Natural obliquity effects, pupil geometry, and mechanical vignetting can reduce peripheral illumination. A simple law is an idealized starting model, not a universal lens specification. Illumination nonuniformity and angle-dependent sensor response can further alter a captured flat field.
Photo-response nonuniformity (PRNU) describes spatial variation in light-response gain. Dark-signal nonuniformity (DSNU) describes spatial variation in the mean dark signal at a specified operating condition. Neither is simply another name for radial lens falloff. Row and column structure can contain offset and gain components; temporal noise varies between frames. Average repeated frames when estimating a stable spatial pattern.
Display: map emitted or reflected luminance
Drive a uniform field and measure luminance at defined locations and viewing angles. Backlight or emitter variation, optical films, projection geometry, and angular response can produce field variation. Localized nonuniformity is often called mura; its visibility depends on size, contrast, drive level, and surrounding content.
Measure more than one gray level when the application requires it. A map measured at full white does not characterize near-black mura or color-channel mismatch. Hold warm-up, drive settings, average picture level, and thermal conditions consistent. An imaging photometer captures dense spatial structure; a small set of spot ROIs can miss it.
State exactly what the uniformity number summarizes
For positive luminances measured over a declared ROI set, a common summary is:
Also retain the mean, extrema, and spatial map. Nine, thirteen, or another number of points may be appropriate to a specified protocol; the count alone does not establish standard compliance. Use a uniform test image, not a contrast checkerboard. For an acquisition test, apply the same logic to a linear, dark-corrected flat-field signal while accounting for the uniformity of the reference source.
Correct the stable components without hiding limitations
A simple camera calibration uses dark frame and flat frame acquired under compatible conditions:
The denominator must be valid, positive, and unsaturated. The correction can remove repeatable offset and gain variation in that calibrated configuration; it cannot recover clipped information or eliminate random noise. Strong amplification in dark corners also amplifies uncertainty.
For a display, spatial drive correction adjusts the emitted pattern toward a target luminance. Its calibration belongs to the selected gray level, channel, angle, and thermal state; limited headroom may require reducing the target brightness. In both branches, verify residual nonuniformity with independent captures and representative operating conditions.
Iris Optics can help diagnose camera shading and display mura, design defensible spatial measurements, and validate compensation without confusing a corrected map with recovered signal quality.