Screen flicker
More actions
Screen flicker (also called display flicker) is a rapid, repeated change in the light output of a display that a viewer can see, or that affects the viewer without being consciously seen. Oculus VR's developer guide describes display flicker as "a rapid 'pulsing' of lightness and darkness on all or parts of a screen", and notes that some people are extremely sensitive to it and experience eyestrain, fatigue or headaches, while others never notice it.[1] Lighting researchers call the physical stimulus temporal light modulation (TLM) and reserve "flicker" for one of the visual responses it can cause.[2]
Flicker comes from the way displays produce light: older cathode-ray tube (CRT) monitors refreshed at rates that some users could see, LED light sources and some OLED screens are dimmed by switching them fully on and off (pulse-width modulation), and low-persistence displays in VR headsets deliberately light each frame for only a short flash.[2][1] In head-mounted displays it matters more than on a desktop monitor because a wide field of view puts the image into peripheral vision, where flicker is seen most easily, and because low persistence trades motion blur for flicker risk. Oculus later stated that the 90 Hz panels of its consumer Rift are fast enough that most users will not perceive noticeable flicker.[1][3][4]
Definition and terminology
The International Commission on Illumination (CIE) technical note CIE TN 006:2016 groups the visible effects of temporal light modulation into three types, as summarised in a 2022 review by Naomi Miller and colleagues at the Pacific Northwest National Laboratory:[2]
| Effect | Viewing condition (CIE TN 006:2016) | Typical frequency range |
|---|---|---|
| Direct flicker | Static observer in a static environment; the unsteadiness is seen directly, on or off axis | Typically said to occur between 3 and 90 Hz[2] |
| Stroboscopic effect | Static observer in a non-static environment; requires a moving object in the field of view | Generally recognised as occurring between about 80 and 2000 Hz[2] |
| Phantom array effect (also called ghosting) | Non-static observer, one who moves the eyes in large saccades across a light source or a scene | Reported in experiments at frequencies far above the direct-flicker limit[5] |
Whether a modulation is noticed depends on its fundamental frequency, its modulation depth (how far the light falls between peaks, often expressed as percent flicker), the shape of the waveform, the duty cycle (the share of each cycle during which the light is at its high level) and the light intensity.[2] Early work by H. de Lange and D. H. Kelly found peak sensitivity to direct flicker at roughly 10 to 15 Hz, where modulations of less than 1% are visible to an average observer.[2] The multiple-image artifacts that appear on displays during eye or head movement are covered in the article Strobing; the related question of how long each pixel stays lit is covered in Persistence.
Perception
The frequency above which a flickering light looks steady is the critical flicker fusion frequency (CFF), also called the flicker fusion threshold.[6] A 2021 narrative review in Medicina summarises the common view that the eye cannot detect flicker above 50 to 90 Hz, depending on intensity and contrast, and lists the factors that move the threshold: luminance, modulation frequency and amplitude, retinal location, wavelength, ambient light, stimulus size and viewing distance, as well as age, fatigue and time of day.[6]
Two of these factors are central for headsets. Under the Ferry-Porter law, CFF rises linearly with the logarithm of retinal illuminance, so brighter images flicker more visibly.[7] Christopher Tyler and Russell Hamer found that the slope of this relationship more than doubled between the foveola and 35 degrees of eccentricity, which they interpreted as a shorter time constant of the visual response at greater eccentricity.[8] For computer screens, Bauer, Bonacker and Cavonius reported in 1983 that flicker was detected most readily not straight ahead but off axis, peaking at 30 degrees eccentricity, and recommended refresh rates above 100 Hz to reduce flicker perception among 95% of users; brighter screens (320 cd/m2) produced stronger responses than dimmer ones (80 cd/m2).[2]
Laboratory CFF values describe a steady gaze. When the eyes move, much faster modulation can be detected. Roberts and Wilkins had 11 observers make 20 to 40 degree saccades across a flickering light in a dark room; the light appeared as a spatial pattern (a phantom array) during the saccade, which let observers tell flicker from steady light at frequencies averaging 1.98 kHz.[5] James Davis, Yi-Hsuan Hsieh and Hung-Chi Lee reported in 2015 that people perceive flicker artifacts at rates over 500 Hz when a display shows high-frequency spatial edges, many times higher than the 50 to 90 Hz usually quoted. They hypothesised that unconscious saccades across those edges were responsible, and concluded that displays using complex spatio-temporal coding need to update much faster than conventional televisions.[9] Because the observers' gaze was not fixed, Miller and colleagues later described the study as effectively measuring detection of the phantom array during saccades rather than a conventional CFF.[2]
Sources of flicker in displays
Refresh and scanning
Visual display terminals built on CRTs spread from the 1980s, and their refresh rates, usually 60 to 75 Hz, drew complaints from some users; manufacturers responded by raising refresh rates and by interlacing the raster lines.[2] Davis and colleagues note that traditional cinemas and televisions presented images at 48 to 60 Hz. They add that the 50 to 90 Hz fusion figures from earlier research were incorporated into international standards for display ergonomics, and they quote an earlier claim that "a frame rate of 72 Hz for computer displays is sufficient to avoid flicker completely".[9] The Oculus developer guide states that OLED displays carry "some degree of flicker, similar to CRT displays".[1]
Pulse-width modulation
LED drivers can switch the current to an LED on and off very quickly, and pulse-width modulation (PWM) uses this to dim the light: the output is modulated by 100% and the duty cycle is shortened to lower the average brightness.[2] Some LED products modulate at 400 to 2000 Hz, a range in which Roberts and Wilkins found visible effects that the existing flicker and stroboscopic models did not predict.[2][5] PWM is also used to dim OLED smartphone displays. A 2023 study published in the SID Symposium Digest of Technical Papers compared OLED smartphone PWM frequencies from 360 Hz to 1920 Hz, plus a direct-current (non-pulsed) mode, and found no statistically significant difference in subjective or objective measures of visual fatigue after about 35 minutes of viewing in the dark.[10]
Low-persistence and strobed displays
A low-persistence display lights each frame for only a short illumination period instead of holding it for the whole refresh interval, which keeps the image sharp during head motion. Valve, for example, states that the reduced illumination period of the Valve Index displays allows "imagery to remain just as sharp when your head is in motion as when you're standing still".[11] The cost is a deep modulation of light at the refresh frequency. Reporting on John Carmack's 2014 Oculus Connect keynote, Road to VR summarised that at 60 Hz with low persistence the screen is black for most of each frame, and that more frames per second are needed to stop users noticing the blanks.[12] Michael Abrash, then at Valve, wrote in an October 2013 reply to a reader comment on his Valve blog that any significant variation in frame time on a low-persistence display "will cause variable problems with flicker and strobing".[13]
Field-sequential and high-speed displays raise the same issue in a different form. Davis and colleagues point out that DLP projection builds full-color images from dozens of very brief monochrome sub-frames, and that stereo 3D televisions show coded left-eye and right-eye frames at a total of 120 Hz; such coded displays depend on the eye integrating light over time.[9]
History in VR headsets
The 2014 Oculus VR Best Practices Guide (v0.008) told developers that flicker "can generally be noticed at refresh rates lower than 60 hz on CRT and OLED displays", that the 60 Hz LCD of the Oculus Rift DK1 should not contain perceivable flicker, and that flicker "plays a significant role in simulator sickness". It also noted that a wide display field of view makes users particularly susceptible to subtle flicker in peripheral regions, because motion perception is more sensitive in the periphery.[3] The Oculus Rift DK2, announced at GDC on 19 March 2014, moved to a 5-inch OLED panel (960 x 1080 per eye) with low persistence, listing refresh rates of 75, 72 and 60 Hz with persistence settings of 2 ms, 3 ms and full; Road to VR found the image "substantially sharper in the DK2 when moving your head".[14]
In his Oculus Connect keynote in September 2014, Carmack named flicker as one of the main limitations of the Samsung Gear VR, calling it "a 60 Hz low persistence display which has flicker problems for a lot of people".[4][12] According to the published transcript, he added that many people could still see flicker on the 75 Hz DK2, especially with white content at the outer edge of the screen in peripheral vision, and that "90 Hz is where probably 95-99% of the people really don't see it".[4] The 2017 edition of the Oculus Best Practices guide states that the 90 Hz panels of the Rift "are fast enough that the majority of users will not perceive any noticeable flicker", while warning that OLED panels still carry some flicker and that sensitivity differs from person to person.[1]
Valve lists the Valve Index as using dual 1440 x 1600 LCDs with "ultra-low persistence global backlight illumination (0.330ms at 144Hz)" and frame rates of 80, 90, 120 and 144 Hz (the 144 Hz mode is described as experimental). It states that the displays have an illumination period of 0.330 to 0.530 ms depending on frame rate, which it calls a five-fold improvement over first-generation PC VR headsets.[11]
Applications and design guidance in VR and AR
Content and rendering guidance
The 2017 Oculus guide lists the factors that decide how much flicker a user perceives: how fast the display cycles between on and off, how much light is emitted in the "on" phase, which parts of the retina are stimulated, and even the time of day and the user's fatigue. It tells developers that people are more sensitive to flicker in the periphery than in the centre of vision and that brighter images produce more flicker, so bright imagery in the periphery, such as standing in a bright white room, can make flicker noticeable; it recommends darker colors, particularly outside the centre of the view, and states that a higher refresh rate makes flicker less perceptible.[1] The guide also says flicker "plays a significant role in the oculomotor component of simulator sickness" and can lead to headaches and eyestrain even after it becomes less consciously noticeable.[1] A separate kind of flicker comes from the content: the guide warns that a rendering effect shown in only one eye can look like flickering or shimmering.[1]
Photosensitive seizures
Flashing content is a medical risk for a small part of the population. A 2005 review for an Epilepsy Foundation of America working group estimated that photosensitivity occurs in about 0.3 to 3% of the population and that seizures from light stimuli affect about 1 in 10,000 people (1 in 4,000 aged 5 to 24). It found frequencies of 15 to 25 Hz the most provocative, within a range of 1 to 65 Hz, and listed video games among known triggers.[15] The Oculus guide tells VR developers not to create purposely flickering content, warns that high-contrast flashing stimuli and high-spatial-frequency textures such as fine black-and-white stripes can trigger photosensitive seizures, and notes that the International Organization for Standardization was developing a standard for image content to reduce that risk, citing the draft ISO/DIS 9241-391.2.[1]
Passthrough and ambient lighting
Passthrough cameras on mixed reality headsets meet a second source of flicker: room lighting powered by alternating current. Meta's passthrough documentation states that in flicker-free lighting conditions (without pulsing fluorescent or LED sources at certain frequencies) passthrough synchronizes the cameras with the display, removing judder, and that this camera-display synchronization is available only on the Meta Quest 3 and Meta Quest 3S.[16] Apple's technical specifications for the original Apple Vision Pro (introduced in 2024) list a "Flicker sensor" among its sensors and supported refresh rates of 90, 96 and 100 Hz; the specifications for the 2025 model with the Apple M5 chip also list a flicker sensor and add 120 Hz.[17][18] MacRumors reported in October 2023 that code in a visionOS beta described the 100 Hz mode as "Adjusted to 100Hz to compensate for detected 50Hz flicker from artificial lighting".[19]
Standards and recommendations
IEEE 1789-2015, IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers, was the first published recommended practice aimed at reducing health risks from the temporal light modulation of solid-state lighting. It sets limits from a combination of percent flicker and fundamental frequency:[2]
| IEEE 1789-2015 level | Recommendation (as summarised by Miller et al.) |
|---|---|
| Low risk | Percent flicker below 0.08 times the fundamental frequency; based on minimising visual discomfort and annoyance and keeping the risk of headaches and photosensitive seizures low |
| No observable effect | A 2.5-fold reduction relative to the low-risk criterion |
| Seizure prevention | Modulation below 5% at all frequencies below 90 Hz |
The recommended practice was written for LED lighting rather than for displays. Miller and colleagues describe it as controversial: too strict in some areas (its low-risk limits would exclude the modulation of incandescent lamps) and not strict enough in others, especially above 400 Hz, where the phantom array remains visible.[2] The same review notes that the Epilepsy Foundation recommends a modulation depth of no more than 5% for light sources and screens modulating between 3 and 65 Hz, while observing that 5% modulation is clearly visible at 15 Hz.[2] Other measures in use include the IEC flickermeter metric (PstLM) for direct flicker and the stroboscopic visibility measure (SVM), which is being documented through the CIE.[2]
Research
Lighting studies found the most commonly reported health effects of modulated light are headaches and migraine. In an office study by Arnold Wilkins and colleagues (1989), replacing 100 Hz magnetic fluorescent ballasts with high-frequency electronic ballasts significantly reduced the frequency and severity of headaches among the workers most prone to them.[2]
Writing in 2019 that little was known about user sensitivity to these artifacts in head-mounted displays, T. Scott Murdison, Christopher McIntosh, James Hillis and Kevin MacKenzie reported psychophysical experiments measuring user sensitivity to flicker and eye-movement-induced ghosting at the refresh rates and persistence levels relevant to HMDs, and made general recommendations for HMD design.[20] A 2020 follow-up by Alexander Goettker, Kevin MacKenzie and Murdison used a high-speed, AR-like head-mounted display prototype to compare low duty cycles (which produce phantom arrays) with high duty cycles (which produce motion blur). Detection of phantom arrays varied widely between subjects, but low duty cycles consistently disturbed saccade targeting, while motion blur became more likely during head movements at high duty cycles and low refresh rates.[21]
At Stanford University, Brooke Krajancich, Petr Kellnhofer and Gordon Wetzstein built a custom high-speed VR display around a Texas Instruments DLP projector running in a 360 Hz 8-bit grayscale mode, and measured CFF jointly across eccentricity and spatial frequency. Their model shows an "anti-foveated" pattern, with the highest thresholds in the near to mid periphery rather than at the centre of gaze, and they estimated that exploiting it could give bandwidth savings seven times larger than spatial-only foveated models. The authors note that their model covers fixated viewing only and does not include the saccade effects reported by Davis and colleagues.[7]
See also
References
- ↑ 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 "Oculus Best Practices (Version 310-30000-02)". Oculus Developer Documentation. Oculus VR, LLC. 2017. https://static.oculus.com/documentation/pdfs/intro-vr/latest/bp.pdf. Retrieved 2026-10-06.
- ↑ 2.00 2.01 2.02 2.03 2.04 2.05 2.06 2.07 2.08 2.09 2.10 2.11 2.12 2.13 2.14 2.15 2.16 N. J. Miller, F. A. Leon, J. Tan, L. Irvin (2022-08-04). "Flicker: A review of temporal light modulation stimulus, responses, and measures". Lighting Research & Technology, vol. 55, no. 1. pp. 5-35. doi:10.1177/14771535211069482. https://doi.org/10.1177/14771535211069482. Retrieved 2026-10-06.
- ↑ 3.0 3.1 Richard Yao, Tom Heath, Aaron Davies, Tom Forsyth, Nate Mitchell, Perry Hoberman (2014-04-30). "Oculus VR Best Practices Guide v0.008". Oculus VR. https://s3.amazonaws.com/arena-attachments/238441/2330603062c2e502c5c2ca40443c2fa4.pdf. Retrieved 2026-10-06.
- ↑ 4.0 4.1 4.2 "John Carmack's Keynote at Oculus Connect 2014 (Transcript)". The Singju Post. https://singjupost.com/john-carmacks-keynote-oculus-connect-2014-transcript/. Retrieved 2026-10-06.
- ↑ 5.0 5.1 5.2 J. E. Roberts, A. J. Wilkins (2012). "Flicker can be perceived during saccades at frequencies in excess of 1 kHz". Lighting Research & Technology, vol. 45, no. 1. pp. 124-132. doi:10.1177/1477153512436367. https://doi.org/10.1177/1477153512436367. Retrieved 2026-10-06.
- ↑ 6.0 6.1 N. D. Mankowska, A. B. Marcinkowska, M. Waśkow, R. I. Sharma, J. J. Kot, P. J. Winklewski (2021-10-13). "Critical Flicker Fusion Frequency: A Narrative Review". Medicina (Kaunas), vol. 57, no. 10. pp. 1096. doi:10.3390/medicina57101096. https://pmc.ncbi.nlm.nih.gov/articles/PMC8537539/. Retrieved 2026-10-06.
- ↑ 7.0 7.1 Brooke Krajancich, Petr Kellnhofer, Gordon Wetzstein (2021). "A perceptual model for eccentricity-dependent spatio-temporal flicker fusion and its applications to foveated graphics". ACM Transactions on Graphics, vol. 40, no. 4, article 47. doi:10.1145/3450626.3459784. https://doi.org/10.1145/3450626.3459784. Retrieved 2026-10-06.
- ↑ Christopher W. Tyler, Russell D. Hamer (1990). "Analysis of visual modulation sensitivity. IV. Validity of the Ferry-Porter law". Journal of the Optical Society of America A, vol. 7, no. 4. pp. 743-758. doi:10.1364/JOSAA.7.000743. https://doi.org/10.1364/JOSAA.7.000743. Retrieved 2026-10-06.
- ↑ 9.0 9.1 9.2 James Davis, Yi-Hsuan Hsieh, Hung-Chi Lee (2015). "Humans perceive flicker artifacts at 500 Hz". Scientific Reports, vol. 5, article 7861. doi:10.1038/srep07861. https://doi.org/10.1038/srep07861. Retrieved 2026-10-06.
- ↑ Zexuan Wu, Lili Wang, Yan Tu, Yexi Sun, Chunsheng Jiang, Daniel Wang (2023). "51.1: Effect of PWM Dimming Frequency of OLED Smartphones on Visual Fatigue". SID Symposium Digest of Technical Papers, vol. 54, no. S1. pp. 379-382. doi:10.1002/sdtp.16308. https://doi.org/10.1002/sdtp.16308. Retrieved 2026-10-06.
- ↑ 11.0 11.1 "Valve Index Headset". Valve Corporation. https://www.valvesoftware.com/en/index/headset. Retrieved 2026-10-06.
- ↑ 12.0 12.1 Paul James (2014-09-25). "John Carmack's Brilliant Oculus Connect Keynote Goes 'off Message from the Standard PR Plan'". Road to VR. https://roadtovr.com/john-carmacks-brilliant-oculus-connect-keynote-goes-off-message-from-the-standard-pr-plan/. Retrieved 2026-10-06.
- ↑ Michael Abrash (2013-07-26). "Down the VR rabbit hole: Fixing judder". Ramblings in Valve Time. Valve. https://web.archive.org/web/20150104173414/http://blogs.valvesoftware.com/abrash/down-the-vr-rabbit-hole-fixing-judder/. Retrieved 2026-10-06.
- ↑ Ben Lang (2014-03-19). "GDC 2014: Oculus Rift Developer Kit 2 (DK2) Pre-orders Start Today for $350, Ships in July". Road to VR. https://roadtovr.com/oculus-rift-developer-kit-2-dk2-pre-order-release-date-specs-gdc-2014/. Retrieved 2026-10-06.
- ↑ Robert S. Fisher, Graham Harding, Giuseppe Erba, Gregory L. Barkley, Arnold Wilkins (2005). "Photic- and Pattern-induced Seizures: A Review for the Epilepsy Foundation of America Working Group". Epilepsia, vol. 46, no. 9. pp. 1426-1441. doi:10.1111/j.1528-1167.2005.31405.x. https://doi.org/10.1111/j.1528-1167.2005.31405.x. Retrieved 2026-10-06.
- ↑ "Passthrough best practices". Meta Horizon OS Developers. Meta Platforms. https://developers.meta.com/horizon/documentation/unreal/unreal-passthrough-bp/. Retrieved 2026-10-06.
- ↑ "Apple Vision Pro - Tech Specs". Apple Support. Apple. https://support.apple.com/en-us/117810. Retrieved 2026-10-06.
- ↑ "Apple Vision Pro (M5) - Tech Specs". Apple Support. Apple. https://support.apple.com/en-us/125436. Retrieved 2026-10-06.
- ↑ Juli Clover (2023-10-09). "Apple Vision Pro Supports Up to 100Hz Refresh Rate". MacRumors. https://www.macrumors.com/2023/10/09/vision-pro-100hz-refresh-rate/. Retrieved 2026-10-06.
- ↑ T. Scott Murdison, Christopher McIntosh, James Hillis, Kevin J. MacKenzie (2019). "3-1: Psychophysical Evaluation of Persistence- and Frequency-Limited Displays for Virtual and Augmented Reality". SID Symposium Digest of Technical Papers, vol. 50, no. 1. pp. 1-4. doi:10.1002/sdtp.12840. https://doi.org/10.1002/sdtp.12840. Retrieved 2026-10-06.
- ↑ Alexander Goettker, Kevin J. MacKenzie, T. Scott Murdison (2020). "Differences between oculomotor and perceptual artifacts for temporally limited head mounted displays". Journal of the Society for Information Display, vol. 28, no. 6. pp. 509-519. doi:10.1002/jsid.912. https://doi.org/10.1002/jsid.912. Retrieved 2026-10-06.