Varifocal

A varifocal display is a near-eye display for virtual reality (VR) or augmented reality (AR) whose focal distance can be changed at runtime, so that virtual content can be presented at the optical depth a user is actually looking at. The focal distance is adjusted dynamically, usually under the control of an eye tracker, by mechanically moving the display panel, by using a focus-tunable lens, or by switching between states of an electronically controlled lens stack. The purpose is to supply correct focus cues and reduce the vergence-accommodation conflict (VAC), a known source of visual discomfort in conventional headsets.

Almost all consumer VR and AR headsets through the mid-2020s use a fixed focal plane: every pixel appears to come from a single optical distance regardless of where it is drawn in the virtual scene. Varifocal displays change that single distance over time to track the user's gaze. As of mid-2026 the technique remains a research and demonstration technology in VR; it has been shown in laboratory prototypes by Meta Platforms (the Half Dome and Butterscotch Varifocal series) and others, but no mass-market VR headset ships with a gaze-driven varifocal display.

The problem it addresses

When a person looks at a real object, two ocular responses act together: vergence, the inward or outward rotation of the two eyes so their lines of sight meet on the object, and accommodation, the change in shape of each eye's crystalline lens that brings the object into sharp focus on the retina. In natural vision these two responses are neurally coupled and point to the same distance.

A stereoscopic headset drives vergence correctly, because each eye sees a slightly different image and the brain converges the eyes on the intended depth. Accommodation, however, is fixed: all of the light reaches the eye from the display's single focal plane, so to keep the image sharp the eye must always focus at that one distance even when it is converging somewhere else. This mismatch is the vergence-accommodation conflict, and prolonged exposure to it is associated with eye strain, fatigue, and difficulty fusing close objects. A varifocal display attacks the conflict directly by changing the display's focal distance to match the depth the eyes are verging on, restoring the link between the two cues.

How it works

A varifocal display presents a single image plane at any instant, but moves that plane to the depth the user is currently fixating. Two components are therefore required: a way to measure where the user is looking, and a way to change the system's focal distance.

Eye tracking supplies the gaze direction, and the convergence of the two gaze vectors gives an estimate of the fixation depth. The display then sets its focal distance to that depth, so the part of the scene the user is inspecting is rendered at the correct focus while the eye accommodates naturally. Because this depends on the gaze estimate, varifocal designs are sometimes called gaze-contingent displays, and they demand eye tracking that is both accurate and low-latency so the focus can keep up with rapid eye movements.

The focus change itself has been implemented in several ways:

• Mechanical actuation: the display panel is physically moved toward or away from the lens, changing the image distance. Actuators such as voice coils with flexure hinges have been used to make this motion fast, quiet, and durable.
• Focus-tunable lenses: a deformable or fluid-filled lens changes its optical power on command, shifting the focal plane without moving the panel. The Stanford prototype of Padmanaban et al. combined a focus-tunable lens with an integrated stereo eye tracker.
• Electronically switched lens stacks: a stack of polarization-dependent liquid crystal lenses and switchable wave plates is driven between discrete states, each combination producing a different focal distance, with no moving parts. Meta's Half Dome 3 used six such lenses to sweep through 64 focal planes.

Software can supplement the optics. Because a varifocal display brings only one depth into focus at a time, content at other depths should appear blurred, as it would in the real world. Meta's DeepFocus system used a neural network to render this gaze-dependent defocus blur in real time for the Half Dome prototypes.

Relation to other accommodation-supporting displays

Varifocal is one of several display architectures proposed to provide focus cues and address the vergence-accommodation conflict. In a 2017 review, Matsuda, Fix, and Lanman group these into varifocal, multifocal, accommodation-invariant, monovision, retinal scanning, light field, and holographic displays.

The closest relative is the multifocal display. A varifocal display tracks the eye and presents one focal plane per frame at the gaze depth, so it relies on accurate eye tracking; a multifocal display instead presents several focal planes at fixed depths each frame and does not need to know where the eye is pointed. The trade-off is that varifocal designs cover a continuous depth range with simpler optics but are sensitive to eye-tracking error, whereas multifocal designs are robust to that error but reproduce only a few discrete depths. Hybrid designs that combine a tunable focus with a small number of planes have been proposed to get both benefits.

A related Meta research line, the Focal Surface Display, uses a spatial light modulator as a dynamic freeform lens to bend the single focal plane into a curved focal surface that approximates scene geometry, rather than keeping it flat as a basic varifocal display does. Light field and holographic displays take a different route again, reconstructing the light leaving a scene so that focus cues emerge without an explicit eye-tracked focus mechanism.

History

Work on supplying correct focus cues in stereo displays predates head-mounted varifocal systems. In 2004, Kurt Akeley, Simon Watt, Ahna Girshick, and Martin Banks built a benchtop stereo display with three image planes at different physical distances, a fixed-viewpoint volumetric design that produced near-correct focus cues without eye tracking. Their work helped establish that incorrect focus cues in conventional stereo displays cause measurable problems for the viewer.

Gaze-contingent varifocal designs for VR were demonstrated in academic work in 2017. Padmanaban, Konrad, Stramer, Cooper, and Wetzstein of Stanford University presented a headset that paired focus-tunable lenses and a mechanically actuated display with eye tracking, and reported in PNAS that the adaptive-focus modes improved accommodative response and that users could also correct for their own refractive errors without glasses.

The most visible engineering effort has been at Reality Labs, the hardware research division of Meta Platforms. Meta dates its varifocal program to about 2015, and first showed the work publicly as the Half Dome prototype at the F8 conference in May 2018. Half Dome 1 used mechanically moving displays driven by eye tracking and offered a 140-degree field of view through Fresnel lenses. Half Dome 2 kept the mechanical approach but folded the optics into a smaller, roughly 200-gram-lighter package aimed at comfort. Half Dome 3, shown at Oculus Connect 6 in 2019, removed the moving parts entirely, using the stacked liquid crystal lenses described above to switch electronically among 64 focal planes. At SPIE in early 2020, Reality Labs research lead Douglas Lanman described the electronic varifocal approach as "almost ready for primetime," while noting that no consumer headset yet supported vergence and accommodation together.

In 2023 Meta showed Butterscotch Varifocal at SIGGRAPH, described as the first prototype to combine varifocal focus with a retinal-resolution display of about 60 pixels per degree, enough for 20/20 visual acuity. To prioritize image quality it returned to a mechanical varifocal mechanism (moving the display) rather than the electronic lens stack, and traded field of view down to about 50 degrees. It was shown alongside a computational-camera prototype called Flamera; Butterscotch Varifocal won the Audience Choice award at the conference, while Flamera won Best in Show in the Emerging Technologies program.

Status and applications

As of mid-2026 gaze-driven varifocal remains a research technology in head-mounted VR. Meta has described a compact concept headset called Mirror Lake that would combine varifocal optics with its other display research in a single goggles-like device, but Reality Labs chief scientist Michael Abrash has said it remains a concept that has not been built into a fully functional headset. The company has meanwhile continued to ship consumer Quest headsets without eye tracking or varifocal optics. Reaching consumer products is constrained by the need for fast, reliable eye tracking, by the added size, cost, and power of the focus mechanism, and by the rendering needed to produce realistic defocus blur.

A related idea has reached a commercial enterprise product on the mixed-reality passthrough path. The Varjo XR-4 Focal Edition applies gaze-directed variable-focus to its passthrough cameras rather than to the display panel: fast focus actuators change the cameras' focus distance to where the user is looking so that near-field real objects appear sharp. The device is sold for professional training and simulation rather than the consumer market.

The following table summarizes the principal varifocal VR research prototypes.

Prototype	Year shown	Developer	Focus mechanism	Notes
Padmanaban et al. (Stanford)	2017	Stanford University	Focus-tunable lens + actuated display	PNAS study; also corrected user refractive error
Half Dome 1	2018	Meta Platforms (Reality Labs)	Mechanical, moving displays	140-degree FOV via Fresnel lenses
Half Dome 2	2019	Meta Platforms (Reality Labs)	Mechanical, folded optics	About 200 g lighter; comfort focus
Half Dome 3	2019	Meta Platforms (Reality Labs)	Electronic liquid crystal lens stack	Six lenses, 64 focal planes, no moving parts
Butterscotch Varifocal	2023	Meta Platforms (Reality Labs)	Mechanical, moving display	About 60 PPD retinal resolution; about 50-degree FOV