VR multifocal display minimizes visual discomfort
Driving a focus-tunable liquid lens in an innovative way and synchronizing the display with it, the researchers propose a multifocal display that goes well beyond the typical limitations of focus-tunable lenses. In a paper titled “Towards Multifocal Displays with Dense Focal Stacks” presented at the SIGGRAPH Asia conference last December, the researchers describe an optical system able to produce thousands of focal planes per second, which if the display is fast enough, could be used to project just the images relevant to each focal plane.
focal plane using an eye tracker (left), the new displays
outputs dense focal plane stacks synchronized with the
accurate tracking of the focal-length of an oscillating
focus-tunable lens.
In that case, an observer could look anywhere at the display and would naturally get a coordinated vergence and focus thanks to the images’ depth maps sequentially assigned to a dense stack of focal planes. Multifocal displays are not a new concept per se, but previous demonstrations often found their limitation due to the focus-tunable liquid lens’ relatively slow settling time.
Based on an input driving voltage, such lenses can take about 5ms to settle onto a particular focal length, which would limit their output at 200 focal planes per second at most. Taking in consideration a display operating in the range of 30 to 60 frames per second (fps), this would limit the number of available focal planes to between three and six focal planes per frame, creating accommodation artefacts for the viewer.
Rather than trying to stabilize their focus-tunable lens at particular focal lengths (which limited previous attempts), the researchers used the device’s frequency response to its maximum, driving the lens with a high frequency triangular voltage so as to sweep the lens through its entire range of focal lengths at high speed.
By measuring the actual focus length in real time (monitoring the deflection path of an IR laser beam going through the lens), they were then able to synchronize a high frame-rate display to output just the right rendered 3D image at the right focal plane.
By tracking the focal length with a microsecond time resolution, the lab prototype described in the paper was capable of generating 1600 focal planes per second, fast enough to assign 40 focal planes per frame at 40 frames per second, sequentially. The authors are confident that the optical module necessary to track the lens’ focal length could be miniaturized to be integrated in wearable VR systems.
Of course, at such frame rates, display luminosity would need to be improved too as each focal plane is illuminated for a smaller fraction of time. Now for any given video, that’s many more views to render, which will call for more processing power too and possibly new ASICs to synchronize the new image rendering stack with the focal length detection and focal plane distribution set.
Interestingly, here the viewer can accommodate freely on arbitrary depths from 25cm to infinity (the authors established that the maximum possible depth range of 7cm to infinity would require 147 focal planes to be achieved effectively).
Carnegie Mellon University – www.cmu.edu
Related articles:
Anticipating AR’s future
Smartphone protective case delivers 3D content
Glass-free 3D display relies on eye-tracking to orient pixels
LCD-based holographic displays in the making
17.0-inch light field display could reach market by mid-2019
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
