LEDs flicker on camera because almost all LED fixtures and pixel products dim by pulse-width modulation (PWM): the LED is switched fully on and fully off hundreds or thousands of times per second, and the average sets the brightness. Your eye integrates that switching into steady light, but a camera sensor samples it, and when the camera's timing does not line up with the PWM rate you get rolling bands or strobing in the footage. It is a sampling interaction between two clocks, not a fault in the LEDs or the camera, and it can usually be fixed from either side: match the shutter to the PWM rate, or specify LEDs whose PWM rate the camera can no longer resolve.
Why your eye does not see it but the camera does
An LED has no meaningful persistence: when the drive current stops, the light stops within microseconds. And LEDs do not dim gracefully on reduced current (colour shifts, and behaviour varies across a batch), so nearly every LED product dims by leaving the LED at full drive and switching it on and off very quickly, varying the proportion of time it spends on. That is pulse-width modulation. At 50% brightness the LED is literally off half the time; at 10% it is off 90% of the time. The number of on-off cycles per second is the PWM rate (or PWM frequency), in hertz (Hz).
Human vision stops perceiving flicker somewhere around 60 to 90 Hz (higher in peripheral vision and during eye movement), so a source switching at 400 Hz or more looks perfectly steady. A camera is different: each frame is a discrete exposure of a precise duration (the shutter speed), often much shorter than the frame interval. A 1/1000 s shutter pointed at a 400 Hz PWM source captures less than half a PWM cycle per exposure, so the frame's brightness depends entirely on where the shutter landed. Land on the pulse: bright frame. Land between pulses: dark frame.
The beat-frequency intuition
Two clocks are running: the LED's PWM rate and the camera's frame rate. Each frame samples the PWM waveform at an instant set by the frame clock, so when the two rates are not an exact multiple, the PWM phase drifts slightly from frame to frame and the recorded brightness pulses slowly: a beat frequency. Take 400 Hz pixel tape filmed at 24 fps: that is 16.67 PWM cycles per frame interval, so the sampled phase advances two thirds of a cycle every frame and the footage pulses at roughly 8 Hz. The shutter does not set that rate; it sets how sensitive each frame is to the drift. An exposure spanning a whole number of complete PWM cycles (1/400, 1/200, or 1/100 s on a 400 Hz source) integrates the same total energy wherever in the cycle it starts, so the drift is invisible. An exposure spanning a fractional number of cycles (1/390 s, or anything shorter than one cycle) leaves each frame's brightness phase-dependent, and the slow beat shows.
The exact artefact depends on the shutter type:
- Rolling shutter (nearly all phone, mirrorless, and cinema CMOS cameras): rows of the sensor are exposed at slightly different times, so different rows catch different phases of the PWM cycle: horizontal light and dark bands crawl through the frame. Band count is roughly PWM frequency times sensor readout time: a 400 Hz source on a 10 ms readout gives about 4 band pairs.
- Global shutter: the whole frame is exposed at once, so instead of bands whole frames vary in brightness, strobing from frame to frame.
The game, from either side, is to get many PWM cycles inside every exposure: the more cycles averaged, the less misalignment matters.
Typical PWM rates in pixel LEDs, and why you cannot change them
In addressable pixel products (SPI-data strip and dots driven by controllers such as an ENTTEC OCTO, PLINK or Pixelator), the PWM generator lives inside the driver IC in each pixel. The rate is fixed in the chip's silicon: it is not sent over the data line and cannot be changed by the controller, mapping software, or any setting after purchase. The controller only sets how often new colour values are sent (typically up to 60 fps); the internal PWM runs independently and much faster. If the PWM rate is too low for your camera, the only lighting-side fix is different LEDs. See Pixel: Can the PWM Rate be adjusted for the full explanation.
Representative figures (many datasheets do not print the PWM frequency, and clone silicon varies, so treat these as commonly measured values, not guaranteed specifications):
- WS2812B: around 400 Hz. Fine for the eye; may be visible on camera at short shutter times or low dim levels.
- SK6812: around 1.2 kHz.
- APA102 (clocked SPI type): commonly reported around 19 kHz, high enough for most normal-speed camera work. Published figures vary by source and measurement method: the chip also has a separate, much slower global-brightness dimmer, which is why some sources quote a far lower rate. The high rate is a property of that particular IC design, not of clocked SPI chips in general.
For per-chip figures across the other common ICs (WS2813, WS2815, and the rest), see ENTTEC's pixel protocol pages, for example www.enttec.com/pixel-protocols/WS2812B.
Rule of thumb: at 2 kHz and above, artefacts at normal frame rates (24 to 60 fps) with sensible shutter settings are rare. Below roughly 1 kHz, flickering may be visible on camera depending on shutter speed, sensor readout, and dim level. Slow-motion work (120 fps and up with very short shutters) is the harshest test: even 2 kHz can band there, and multi-kilohertz sources are the safe specification.
Fixing it from the camera side
With lighting already installed, the camera is the adjustable side.
- Set the exposure to a whole number of PWM cycles. If each exposure spans exactly n complete cycles, every frame and every sensor row integrates the same energy and the banding disappears. For a 400 Hz source: 1/400, 1/200, 1/100, or 1/50 s. In practice you rarely know the rate, so scrub through shutter values until the banding freezes or vanishes. Many cinema cameras offer shutter angle or clear scan / synchro-scan in fine steps for exactly this purpose.
- Longer exposures are more forgiving. A 1/50 s exposure spans 8 full cycles of a 400 Hz source, so a fractional cycle at the edges changes total energy by only a few percent. A 1/2000 s exposure spans under one cycle and swings between full brightness and black. This is why footage at a 180 degree shutter (1/48 to 1/60 s at normal frame rates) often looks fine while the same scene at 1/1000 s bands badly.
- Control exposure with ND filters or aperture, not shutter speed. If the scene is too bright, do not shorten the shutter; add ND and keep the flicker-safe shutter time.
- Reconsider high frame rates. Slow motion forces short exposures: if a 120 fps shot bands and the 25 fps shot does not, the PWM rate is marginal for that shot. Small frame-rate changes (25 vs 24 fps) can also move a whole-frame beat out of the visible range.
- Always shoot a test clip. Flicker invisible on a small monitor can be obvious on a large screen; record 10 seconds and check the dim scenes especially.
Fixing it from the lighting side
- Specify high-PWM ICs for camera work. If a wall, set piece, or stage will ever be filmed (broadcast studios, wedding venues, livestreaming churches), make PWM rate a line item in the purchase specification, because it cannot be corrected afterwards. Ask the supplier for the PWM frequency in writing; if they cannot state it, assume the low end. When choosing form factors for a filmed space, see Pixel: Pixel Strip vs Pixel Dots; the IC family, and with it the PWM rate, is chosen per product.
- Run the LEDs brighter where the shot allows. Flicker visibility interacts with dimming: at low brightness the on-pulses are short and sparse, so an exposure is more likely to land in dead time and banding contrast is higher. Content dwelling at 5 to 20% brightness is the worst case; at or near 100%, many ICs are effectively continuously on and there is nothing to beat against. Raise the fixture level and bring camera exposure down with aperture or ND instead. Very low levels on 8-bit chips are doubly bad: few discrete PWM levels remain, so flicker and visible stepping worsen together.
- Do not confuse PWM flicker with data faults. PWM beat effects appear only through a camera. If the flicker is visible to the naked eye, or pixels glitch to random colours, it is a signal, grounding, or power problem instead: work through the Pixel Flickering Guide: Diagnose and Resolve Common SPI LED Issues.
Symptom, likely cause, fix
| Symptom in footage | Likely cause | Fix |
|---|---|---|
| Horizontal bands crawling through the frame, invisible to the eye | PWM rate beating against a rolling shutter | Lengthen the shutter toward a whole number of PWM cycles (try 1/100 or 1/50 s); compensate exposure with ND or aperture |
| Whole frames pulsing brighter and darker | PWM (or mains ripple on non-PWM fixtures) beating against the frame rate on a global-shutter or long-readout sensor | Nudge frame rate or shutter; on cinema cameras use clear scan / synchro-scan |
| Flicker only when the LEDs are dimmed low | Short duty cycle, so each exposure hits or misses the narrow on-pulse | Run the LEDs at a higher level and reduce camera exposure with ND or aperture; avoid content dwelling at 5 to 20% brightness |
| Flicker only in slow motion or at fast shutters | Very short exposures resolving a PWM rate that is fine at 24 to 60 fps | Shoot those shots at lower frame rates, or specify multi-kilohertz PWM ICs; no camera setting fully fixes a low-rate source here |
| Different strip batches band differently in the same shot | Mixed IC types or revisions with different PWM rates | Match IC type and batch across surfaces that share a frame; state PWM rate in the purchase spec |
| Flicker visible to the naked eye, or random colour glitches | Not a camera effect: data integrity, voltage drop, grounding, or termination fault | Follow the SPI LED flickering diagnostic guide |
The honest summary
On-camera LED flicker is not a defect. A WS2812B running at roughly 400 Hz is operating exactly as designed; the design target was human vision. Whether that chip suits your job depends on whether a camera will be pointed at it, because any PWM-dimmed LED can band under a sufficiently short exposure. What you control is the margin: slower shutters and higher PWM rates buy averaging, higher dim levels buy pulse density, and specifying the IC before purchase buys you out of the problem entirely. Test with the actual camera and settings you will shoot with. If the flicker survives a 1/50 s shutter test, or people can see it without a camera, stop adjusting the camera and diagnose the signal chain.