Varifocal display

A varifocal display is a near-eye display that dynamically changes its focal distance so the optical depth of the virtual image matches where the user is looking. The aim is to reduce the vergence-accommodation conflict (VAC), the mismatch between the distance at which the two eyes converge and the distance at which each eye focuses that arises in conventional head-mounted displays (HMDs) because their imagery sits at a single fixed focal plane. A varifocal system pairs eye tracking with a way of altering focal distance (a physically moving display, a focus-tunable lens, or a stack of switchable lenses) and shifts the focal plane to the depth the eyes are verging on.

As of 2026 the technique remains a research subject. It is associated chiefly with prototypes from Meta's Reality Labs (the Half Dome and Butterscotch Varifocal series) and with academic work at NVIDIA, the University of North Carolina, and Stanford. No mass-market consumer headset ships a varifocal display: nearly seven years after Meta first showed Half Dome in 2018, the company still uses fixed-focus optics in its shipping Quest products.

The vergence-accommodation conflict

When a person looks at a real object, two oculomotor responses act together. Vergence is the inward or outward rotation of the eyes so that both fixate on the object and it is seen as single. Accommodation is the change in the shape of each eye's crystalline lens that brings the object into sharp focus on the retina. In natural vision these responses are neurally coupled, so the vergence distance and the focus (accommodation) distance match.

A stereoscopic HMD presents a separate image to each eye and varies their horizontal offset (binocular disparity) to place virtual objects at different apparent depths, which drives vergence. The physical light, however, comes from a panel at a single optical distance, commonly fixed near 2 metres, so accommodation does not change with the depicted depth. The eyes therefore converge on one distance while focusing on another. This conflict is a recognised cause of eye strain, visual fatigue, and discomfort in VR and AR, and it is one reason close-up virtual objects can look blurry or uncomfortable in a fixed-focus headset. Douglas Lanman, who has led display research at Meta, has described the standard headset as locking focus at roughly two metres, which causes strain when a user attends to nearby content.

How varifocal displays work

A varifocal display has a single focal plane at any instant, but it can move that plane. It first measures gaze direction with eye tracking, estimates the depth the two eyes are verging on, and then adjusts the optics so the virtual image is rendered at that depth. Because the focal plane follows the gaze, accommodation can track vergence as it would in the real world. The approach mitigates the conflict but does not by itself reproduce correct monocular focus cues (the natural blur of objects at other depths); systems usually add rendered depth-of-field blur for that.

The focal plane can be shifted in several ways:

• Mechanical (moving display). The physical display panel, or a mirror in the optical path, is moved closer to or farther from the eye to change the image distance. This is the approach in Meta's original Half Dome and in the Butterscotch Varifocal prototype.
• Focus-tunable lens. A lens whose optical power changes electrically, such as a deformable membrane or a liquid lens, retargets the image depth without moving the panel.
• Switchable solid-state lenses. A stack of polarisation-dependent liquid-crystal lenses with switchable half-wave plates produces a set of discrete focal states by switching layers on and off, with no moving parts. This is the basis of Half Dome 3.

Varifocal displays are usually contrasted with two other VAC mitigation strategies. Multifocal (multiplane) displays present several focal planes at once, so the eye can accommodate naturally within their range without needing eye tracking, at the cost of optical complexity and a limited number of planes. Light-field displays reproduce many rays per pupil position to approximate continuous focus, but demand very high resolution and bandwidth and tend to give a narrow field of view.

Research history

Work on focus-correct near-eye displays predates the varifocal label. In 2004 Kurt Akeley, Simon Watt, Ahna Girshick, and Martin Banks published "A stereo display prototype with multiple focal distances" in ACM Transactions on Graphics, a fixed-viewpoint bench display that presented imagery at several focal distances, an early demonstration that focus cues could be supplied in a stereo display.

Two academic varifocal AR prototypes appeared in 2017. A University of North Carolina group with collaborators at the Max Planck Institute for Informatics and NVIDIA (David Dunn, Cary Tippets, Kent Torell, Petr Kellnhofer, Kaan Akşit, Piotr Didyk, Karol Myszkowski, David Luebke, and Henry Fuchs) built a wide-field varifocal see-through display using deformable membrane mirrors held in vacuum chambers. Eye tracking set the focal power, and the mirrors could place the accommodation depth anywhere from 20 cm to optical infinity, with a response of about 300 milliseconds between focal extremes and a roughly 100-degree field of view. The same year, an NVIDIA-led team (Kaan Akşit and colleagues) presented a near-eye varifocal AR display that projected onto a thin see-through holographic screen and relayed the image to a chosen optical depth through a curved beam combiner, reported at SIGGRAPH Asia 2017 and SIGGRAPH 2017 Emerging Technologies.

Meta Half Dome

Half Dome is the best-known varifocal HMD program, run by the team then called Oculus Research and later Facebook Reality Labs (now Meta Reality Labs). The effort began when Douglas Lanman joined the lab in 2014; he led the Computational Imaging team, with contributors including Ryan Ebert (mechanical engineering), Alex Fix (computer vision), Robert Cavin (eye tracking), Brian Wheelwright (optics), and the research scientist Marina Zannoli, about 40 people in total across Oculus and Facebook.

Facebook first teased Half Dome at its F8 conference in May 2018 and Lanman detailed it at SID Display Week the same month. The original prototype used mechanically moving varifocal screens behind Fresnel lenses to give a 140-degree field of view, with eye tracking keeping near content sharp. Wheelwright summarised the design by saying the core of Half Dome was optomechanics rather than optics, and Lanman noted that in a quiet room he could neither hear the screens nor feel them moving.

In September 2019, at Oculus Connect 6, Michael Abrash and the Display Systems Research team described two successors aimed at comfort rather than raw immersion. Half Dome 2 kept a mechanical varifocal design but reduced weight by about 200 grams and replaced its actuators with voice-coil motors and flexure-hinge arrays to cut friction and noise; its field of view was narrower than Half Dome 1 but still around 20 percent wider than the Quest. Half Dome 3 removed moving parts entirely, using a stack of six liquid-crystal lenses (polarisation-dependent lenses plus switchable half-wave plates) to produce 64 discrete focal states by switching layers on and off. Lanman later said Half Dome 3 was "almost ready for primetime," placing it at NASA technology-readiness level 5 or beyond on a scale where 9 is a product shipping at scale, well above the level 2 to 4 his team usually worked at; even so, the technology did not appear in a consumer headset.

Butterscotch Varifocal

At SIGGRAPH 2023 Meta's Display Systems Research team showed Butterscotch Varifocal, which combined the mechanical varifocal approach of Half Dome with the retinal-resolution panel of the earlier 2022 Butterscotch prototype. The headset moved its display nearer to or farther from the eyes according to gaze and reached roughly 60 pixels per degree, enough for 20/20 visual acuity, at a reduced field of view of about 50 degrees (compared with more than 90 degrees on Meta Quest 2). The prototype received the Audience Choice Award at the conference; the light-field passthrough prototype Flamera shown alongside it was named Best in Show. Meta described Butterscotch Varifocal as built solely to showcase the experience of retinal resolution and stated that the technologies it investigates may never reach a consumer product.

Current status

Varifocal displays remain experimental in 2026. Meta has shown working varifocal prototypes since 2018 but continues to ship fixed-focus optics, such as the pancake lenses in Meta Quest 3, in its consumer headsets, and its CTO Andrew Bosworth has argued publicly that some of the features bundled into these prototypes carry penalties in price, weight, and battery life that make them hard to justify for a consumer device.

A related but distinct capability has reached the professional market. The Varjo XR-4 Focal Edition, sold by Varjo for training and simulation, advertises gaze-driven autofocus that tracks the eyes at 200 Hz and adjusts the focus of its passthrough cameras so real-world objects at different distances stay sharp. This refocuses the camera feed of the surroundings rather than changing the optical depth of the virtual image, so it addresses passthrough sharpness rather than the display-side vergence-accommodation conflict that varifocal displays target.

See also

• Vergence-accommodation conflict
• Light field display
• Focal Surface Display
• Eye tracking
• Near-eye display
• Head-mounted display