LED Wall Washer Camera Banding: PWM, Shutter Speed and On-Site Verification

LikeLight Tom · Operations · Technical reviewer: pending confirmation

Summary

An LED wall washer may appear steady to an observer and still produce dark bands, brightness pumping or color stripes on camera. The camera does not integrate light the same way as the eye: a rolling or focal-plane readout samples different image rows at different times, while PWM repeatedly switches or modulates LED current. When those time bases interact, one frame can contain several output phases.

Do not approve a “flicker-free” claim from a phone preview or one full-output test. Freeze the exact luminaire, driver, controller, color, dimming level, effect, camera, frame rate and shutter setting. Human temporal light artefacts and camera banding overlap in cause but are not interchangeable acceptance criteria.

1. Separate four symptoms

Symptom	Primary mechanism to investigate	First evidence
Horizontal bands in stills/video	sensor readout versus light modulation	original file and camera metadata
Whole-frame brightness beats	unsynchronised modulation/frames or multiple sources	frame-by-frame waveform comparison
Color bands only in RGB scenes	independent channel duty cycles or timing	single-channel R/G/B recordings
Visible stroboscopic or phantom-array effect	human temporal light artefact	application-specific TLM assessment
Random blocks or loss of control	data/control fault rather than periodic TLM	DMX status and fault log

CIE uses temporal light modulation (TLM) for measurable variation and temporal light artefact (TLA) for human perception [S1][S2]. Sony explains that sequential sensor exposure can render bright and dark moments at different positions [S4]. A camera band is evidence of an interaction, not by itself a complete human-flicker metric.

2. PWM frequency is only one variable

PWM controls average output through pulse timing. Frequency, duty cycle, waveform shape, channel timing, driver response and minimum dimming level all matter. DOE notes that no single metric covers every frequency and application, camera interaction included [S3].

Hypothetical cycle calculation

If—not a LIKELIGHT specification—a luminaire uses 2,000 Hz PWM, one cycle is 1/2,000 s = 0.5 ms. A 1/1,000 s exposure lasts 1 ms and spans two nominal cycles; a 1/250 s exposure lasts 4 ms and spans eight. This does not guarantee a clean image: rolling readout phase, duty cycle, channel phase, frame rate and exposure position still matter.

For another teaching case, 10% duty cycle at 2,000 Hz gives an idealized on-time of 0.5 ms×10%=0.05 ms per cycle. Short row exposure or phase drift may sample unequal light. These numbers explain the mechanism only; the actual waveform must be measured at the requested color and dimming level.

3. Why “looks fine at 100%” is insufficient

Operating point	Why it belongs in the test
100% output	some drivers may behave differently from dimmed states
50% and 10%	duty cycle/depth and camera bands may change
minimum usable level	often the hardest practical operating point
pure R, G, B and mixed white	channels may not share identical timing
static scene and dynamic fade	control updates can create non-periodic modulation
one luminaire and full array	multiple unsynchronised sources can produce beats

DOE stresses that waveform, modulation depth, duty cycle, frequency, load and dimming condition all affect the result [S3]. Test the delivered system, not only the driver datasheet or a single sample at full output.

4. Camera and lighting form one test system

Sony documents model-dependent remedies such as changing shutter speed, using mechanical shutter, anti-flicker or variable shutter [S4]. These can reduce bands for a particular capture; they do not change the luminaire waveform or prove compatibility with every camera.

Create an approved camera matrix:

List every production camera and whether it uses rolling, focal-plane, electronic or global capture in the selected mode.
Record frame rate, shutter speed/angle, sensor scan mode, ISO and anti-flicker/variable-shutter status.
Lock exposure and white balance; disable automatic corrections that hide repeatability.
Record full output, mid, low and minimum usable dimming for R, G, B and the approved mixed colors.
Repeat static, fade and fast content; include the actual number of simultaneously operating luminaires.
Inspect original frames, not only a compressed stream or preview.
Archive camera firmware, lens, luminaire/driver revision, controller export and DMX scene.

5. Measurement and acceptance

CIE TN 012:2021 recommends consistent, reproducible waveform measurement and notes that facade applications may need different input optics [S1]. Define detector position, field of view, sampling method, calibration, duration and operating state. A smartphone slow-motion clip is useful for screening but is not a substitute for a documented optical waveform.

Acceptance layer	Example deliverable	Boundary
optical waveform	calibrated time-series at defined settings	instrument and method required
human visual assessment	project-specific TLA criteria	does not predict every camera
camera acceptance	matrix of approved cameras/modes/scenes	applies only to tested combinations
control stability	DMX logs and repeated transitions	does not replace optical measurement

Set pass/fail rules before the test: allowed banding, brightness variation, color shift, cameras and modes, dimming range, repeat count and environmental range. “Looks acceptable” without those conditions cannot be reproduced.

6. LIKELIGHT evidence boundary

LIKELIGHT's public page identifies a 36 W, DC24 V, RGB, DMX wall washer and says it uses standard PWM dimming [S5]. The page does not publish an order-specific PWM frequency, phase relationship, modulation-depth curve, TLM report, camera matrix or broadcast certification. Therefore this article makes no flicker-free claim for that product.

For an inquiry, provide camera models, recording modes, target frame/shutter ranges, minimum dimming level, RGB scenes, quantity, controller, sync expectations and acceptance method. Request the exact driver/controller configuration and a sample test before project approval.

7. When not to solve the problem by changing shutter

Do not change production shutter if it breaks motion rendering or exposure requirements.
Do not declare the luminaire compliant because one camera setting hides bands.
Do not raise PWM frequency in firmware unless the driver, thermal behavior, dimming resolution and EMC implications are approved.
Do not confuse random DMX faults with periodic camera banding.
Do not use “flicker-free” without a defined metric, operating range and test report.
Do not extrapolate from one color, one dimming level or one luminaire to a full facade.

FAQ

Why is the wall washer steady to my eyes but banded on camera?

The eye and sensor integrate time differently. Sequential sensor readout can sample different PWM phases across the frame [S4].

Does higher PWM frequency always eliminate banding?

No. It can reduce risk, but duty cycle, waveform, scan time, phase, channel timing and camera mode still matter.

Can 1/50 or 1/60 second always fix LED banding?

No. Sony's guidance is source- and region-dependent, and variable shutter may be needed for some LED sources [S4]. Test the exact camera and light.

Is percent flicker enough for video approval?

No single metric predicts every application or camera interaction [S3]. Retain waveform data plus a camera acceptance matrix.

Why do bands appear only at low brightness?

PWM duty cycle or driver behavior can change with dimming. Measure the exact low-level waveform rather than inferring it from full output.

Can DMX synchronization guarantee a clean frame?

Not by itself. DMX updates, PWM generation, driver phase and sensor timing are different layers; all relevant timing must be verified.

Sources checked

2026-09-08

LIKELIGHT 36 W DC24 V RGB DMX wall washer

Submit project requirements

A wide COB line can improve luminous coverage, but its 30W/m load, 20mm profile fit, CCT choice and IP20 boundary must be resolved together. This guide turns the published values into a buyer decision.

A facade can pass its animation test yet fail at scale when DMX channels and power branches are treated as the same limit. This guide calculates both for the LIKELIGHT 8 pixels/m RGB linear light.

Module count alone cannot predict an even sign face. This guide uses the published 70×12mm body, DC12V, 0.96W and 20-module string to plan optical mock-ups, power zones and cascade order.