Motion blur

See also: Persistence and Judder

Motion blur is the smearing of a moving object or scene along its direction of motion in a captured, rendered or displayed image. It appears whenever light is integrated over a finite time while the image moves relative to whatever is collecting it: the film or sensor of a camera, or the retina of a viewer whose eye is following an object across a screen.

In virtual reality and augmented reality the term covers three related problems. The first is display blur: a head-mounted display that keeps each frame lit for the whole refresh period smears the image across the retina whenever the user turns their head, which is why modern headsets use low-persistence displays (see Persistence). The second is rendered motion blur, a visual effect that game engines add to simulate a film camera's exposure; Unity's Post Processing Stack v2, for example, does not support it in VR or AR. The third is camera blur in video passthrough, where the headset's outward cameras record blur during head movement, especially in low light.

Reviewed 15 October 2026. Claims, figures and quotations checked against each cited paper, archived blog post, specification page, documentation page and review. About review dates.

Causes

Exposure blur in cameras

In a film or electronic camera, the shutter lets light reach the image plane for a finite exposure time. If an object or the camera moves during that time, the object's image is spread along its path and shows up as a visible streak. Navarro, Seron and Gutierrez describe the effect in their 2011 survey as "the result of the combination of relative motion and light integration taking place in film and electronic cameras". They also note that image sequences with a moderate amount of motion blur are perceived as natural, while sequences with none at all look jerky and strobe.

Hold-type display blur

Displays produce a second kind of blur that exists only in the viewer's eye. LCD and OLED panels are "hold-type" (also called sample-and-hold) displays: the thin-film transistor driving each pixel holds the image steady for the whole frame time. A cathode-ray tube, by contrast, is an "impulse-type" display. When the eye smoothly tracks an object moving across a hold-type screen, the object's image stays fixed on the panel for a full frame while the eye keeps moving, so the image slides across the retina and is perceived as blurred. Peng and colleagues point out that for this reason an OLED panel can still show motion blur even if its pixel response time is zero. Slow pixel response adds further blur on top of this hold effect; the same authors note that nematic liquid crystals respond about 100 times more slowly than OLEDs, whose response time is about 0.1 ms.

Display motion blur in head-mounted displays

Head-mounted displays make hold-type blur much more visible than monitors or televisions do. Michael Abrash, then at Valve, gave the reasons in a June 2013 blog post: a headset's field of view is much wider, so objects can be tracked for longer, and the head can turn much faster than the eyes can normally follow a moving object while still seeing clearly, because the vestibulo-ocular reflex counter-rotates the eye. He added that images in a headset appear to be in the world rather than on a screen, so the visual system expects more of them. When the eye moves relative to the display in this way, each lit pixel sweeps a constant color across the retina for as long as it is lit (its persistence time). The result is a smear, followed by a jump back to the correct position at the next frame that can be seen as strobing.

Abrash gave a worked example. A leisurely head turn is in the neighborhood of 100 degrees per second; a 120 degrees per second turn on a 60 Hz headset moves the image two degrees per displayed frame, which he put at about 14 pixels on an Oculus Rift development kit. He also noted that the slow, ramped pixel transitions of the LCD panels in those development kits made the smear "longer and smoother", which hid the strobing but lost even more detail, and that the problem grows with pixel density because the smear covers more pixels. He called the combined effect of smearing and strobing "much like a choppy motion blur"; the wiki covers it under Judder.

Steven M. LaValle's textbook Virtual Reality describes the same retinal image slip and the engineering response. At 60 frames per second each image is held for 16.67 ms. In low-persistence mode the screen is lit for only about one to two milliseconds per frame and kept black for the rest, which greatly reduces the slip. LaValle adds that without the blank interval the frame rate would need to reach at least 500 frames per second, and that a 60 Hz low-persistence display flickers visibly, an effect he says is easy to see in the periphery of a bright scene in the Samsung Gear VR; at 90 Hz or above the flicker subsides for nearly everyone. He contrasts OLED pixels, which reach their target intensity in under 0.1 ms, with LCD pixels that may take as long as 20 ms, making a fixed virtual object smear in the direction of motion, which he says was easy to observe in the LCD-based Oculus Rift DK1.

Adoption of low persistence

Oculus VR announced the Oculus Rift DK2 development kit on 19 March 2014. The company stated that, like the Crystal Cove prototype, DK2 "uses a low persistence OLED display to eliminate motion blur and judder", which it called "two of the biggest contributors to simulator sickness", and said that low persistence also makes the scene appear more visually stable. In 2019 UploadVR reported that John Carmack had confirmed the original Oculus Rift's pixel persistence as 2 milliseconds and had stated that the Oculus Rift S has lower persistence than both the Rift and the Oculus Quest, even though the Rift S runs at 80 Hz instead of 90 Hz. The same article said that every VR headset then on the market, apart from very cheap ones, used low persistence. Valve lists "ultra-low persistence global backlight illumination" of 0.330 ms at 144 Hz in the technical specifications of the Valve Index headset, whose displays are LCDs.

Display researchers have also worked on LCDs aimed at the same problem. A 2018 Society for Information Display paper by Gou, Chen, Li, Lee and Wu calculated that, because of the hold effect, a 90 Hz panel would still have a motion picture response time of 8.89 ms even with an instantaneous liquid crystal response, and noted that the same applies to OLED. For their proposed fast liquid crystal design, they calculated that driving it at 90 Hz with a backlight lit for 17% of each frame would give a motion picture response time of about 1.5 ms, which they describe as similar to a cathode-ray tube.

Measurement

The standard figure of merit for display motion blur is the motion picture response time (MPRT). It is derived from the blurred edge width (BEW) that a viewer perceives when tracking a moving edge: MPRT (ms) = BEW (pixels) / v (pixels per frame) x T_f (ms per frame), where v is the speed of motion and T_f is the frame time. In practice MPRT is taken as the time between 10% and 90% of the perceived luminance change. Blurred edges can be captured with a pursuit camera that follows the moving pattern, or simulated by integrating the display's output over time.

Peng and colleagues derived a simple approximation, MPRT ≈ √(τ² + (0.8 T_f)²), where τ is the liquid crystal (or OLED) response time. With an ideal instantaneous panel the limit is 0.8 T_f, so blur on a hold-type display falls in proportion to frame time. When the backlight is lit for only a fraction of each frame (the duty ratio, DR) and the liquid crystal is fast enough to settle before the backlight turns on, MPRT is approximately 0.8 x T_f x DR. They also found that a liquid crystal response time of 2 ms or less gives almost the same MPRT as an OLED, and cited a requirement of under 4 ms MPRT to show an object moving at 480 pixels per second clearly. The two papers give the following values:

Display and drive condition	MPRT	Source
Cathode-ray tube (impulse-type)	About 1.5 ms
Typical 120 Hz LCD or OLED television (hold-type)	About 6.66 ms
OLED, 120 Hz	6.67 ms
Ultra-low-viscosity vertical alignment LCD (0.93 ms average gray-to-gray response), 120 Hz	6.88 ms
Hold-type panel with instantaneous response, 90 Hz	8.89 ms
Proposed fast LCD for VR, 90 Hz, 17% backlight duty ratio (calculated)	About 1.5 ms

The display testing site Blur Busters, in an article by Mark Rejhon, uses a simplified rule of thumb it calls "Blur Busters Law": on a display with instant pixel transitions, 1 ms of persistence produces 1 pixel of motion blur for every 1,000 pixels per second of motion. It presents this as a lower bound, since slower pixel transitions only add blur, and says it uses a full-transition "MPRT 100%" measure instead of the "MPRT 90%" of the scientific paper it simplifies. For monitors, televisions and laptops, the Video Electronics Standards Association (VESA) runs the ClearMR certification, which grades motion blur with a Clear Motion Ratio (CMR) based on the ratio of clear to blurry pixels. VESA says that current methods such as MPRT fail to reflect the true nature of blur, since a solely time-based metric cannot account for a number of image enhancement and blur-mitigation techniques; it also cites excessive overshoot and undershoot as sources of artifacts and distortions.

Rendered motion blur

Because a rendered frame is normally computed for a single instant, renderers must simulate motion blur explicitly, and the Navarro survey calls accurate temporal integration "one of the most expensive processes in the production of high-quality renders". The survey traces research back to the early 1980s. Korein and Badler's 1983 analytic method found the time intervals during which an object covers a pixel; Potmesil and Chakravarty's 1983 method blurred a rendered snapshot with point spread functions; and Cook and colleagues' 1984 distributed ray tracing sampled the time domain stochastically along with lens, light and reflection directions. Later approaches include accumulation buffers, which average many sub-frames across the shutter interval, and image-space post-processing. Post-processing blurs an already rendered frame using motion information from the images or the animation data. The survey found "a clear preference" for real-time post-processing in game architectures of the time, citing Sousa's 2008 Game Developers Conference presentation on Crysis effects, whose method computed a velocity field from the current and previous camera transforms, and Vlachos's 2008 presentation on post-processing in The Orange Box.

Unity's Post Processing Stack v2 describes its motion blur effect as one that "blurs the image in the direction of the Camera's movement" to simulate a real camera with its lens aperture open. The effect needs motion vectors and a depth texture, and the documentation states that "Motion blur doesn't support AR/VR."

Blur as a comfort technique

Some research has tested deliberate blur as a way to reduce cybersickness. Budhiraja, Miller, Modi and Forsyth (2017) proposed "Rotation Blurring": a uniform Gaussian blur over the whole screen whenever the player rotated the character with the mouse in a VR first-person shooter, with the blur strength proportional to the magnitude of the mouse's acceleration. The blur was triggered by mouse-driven rotation only; participants could still look around independently by moving their heads. In a within-subjects study with 15 participants on an Oculus Rift DK2, 8 participants had lower total sickness scores with blurring, 3 had no change and 4 had higher scores. The median total sickness score fell from 48.62 to 33.66, but the overall decrease was not statistically significant (p = 0.19); the authors reported that participants with acute cybersickness benefited significantly and that blurring delayed its onset. A 2020 study by Nie, Duh, Liu and Wang tested dynamic blurring based on saliency detection; with 40 participants split between a control condition and a dynamic-blurring condition, it reported a statistically significant reduction in average motion sickness severity.

Motion blur in video passthrough

Mixed reality headsets that show the outside world through cameras inherit ordinary camera exposure blur. In his review of the Apple Vision Pro, The Verge's Nilay Patel wrote that motion blur was easy to see when moving his head, that it increased in low light, where cameras must lengthen exposure at the cost of sharpness or raise ISO at the cost of noise, and that it led to some warping of straight lines. MIXED's review of the same headset found the passthrough image "severely blurred" when the head moved left or right and judged the Meta Quest 3 better on this point.

See also

• Persistence
• Judder
• Refresh rate
• Motion interpolation
• Motion Smoothing
• Asynchronous Spacewarp
• Timewarp
• Simulator sickness
• Depth of field
• Rolling shutter