Technology Review: Head Mounted Displays

Article for VR News, November 1998

© copyright VRNews 1998, All Rights Reserved

Jerry Isdale

Head Mounted Displays (HMDs) are seen by many people as the defining technology of VR. I use a broader definition of Virtual Realty, but historically, VR did emerge with the development of the HMD. Head Coupled Displays (HCD) were actually one of the early computer graphic displays. Ivan Sutherlandís projects in the mid to late 1960s used CRT displays suspended from the ceiling. Military researchers have long worked with HMDs as wearable Heads-Up Displays (HUD) or night vision systems. They have developed a significant body of knowledge on HMD design. Virtual Environment research at NASA Ames Research Centerís during the mid to late 1980s gave us the first VR-HMDs. Commercialization of the technology in the early 1990s was one of the converging trends that brought VR to the general public.

Early HMDs for VR were often crude, simple devices that slapped a display and some lenses on some sort simple support. Unfortunately the poor fit, low display resolution, and bad optics of the early VR-HMDs caused many problems. Several studies were done in the early 1990s that indicated potential health and safety hazards with contemporary HMDs. The low and mid range HMDs vendors took a big hit and many are no longer with us. Most of the current HMDs give much more consideration to the user, alleviating much of the hazard concern. (Wow, human factors , what a concept. Regular readers will find I am very much in favor of usability design considerations.) An excellent reference on HMD design is the book "Head-Mounted Displays: Designing for the User" by James Melzer, Kirk Moffitt (McGraw Hill, ISBN 0070418195). The 12 chapters by a variety of authors cover a wide range of HMD issues, from basic requirements, through optics design, head and neck biometrics (and myths of anthropometrics), stereoscopic vision, and, yes, usability testing. There is also a chapter on Brain Actuated Controls for those seeking the exotic input devices.

HMDs basically consist of an image source (display), optics and a supporting device. Often a tracking device is added so head motion can be used to control the view. Three main flavors of HMDs might be monocular, biocular and binocular, with color or monochrome options in a variety of visual and display resolutions. Monocular displays have only one display source. Biocular HMD have two displays with separate displays and optics paths, but show only one image. Binocular HMDs provide stereoscopic viewing. This requires two image generators, which can greatly increase the overall cost of a VR System. There are many applications that benefit from stereoscopic vision, but there are also others that derive no or marginal utility. If your application does not require it, a biocular or monocular HMD could provide a more cost effective approach. A desktop VR approach can reduce the head tracking needs as well, further reducing capital investment requirements. The bottom line in selection of a HMD model or even another approach should be how effective it makes the experience for the user.

Image sources used in HMDs are typically either CRTs or flat panels such as an LCD. CRT displays offer generally higher resolution especially in grey scale. Color wheels or shutters can be used to produce field-sequential color. They are heavier and more bulky than LCDs but have been the display of choice when very high resolution (1280x1024 or better) is required. The high end (US$20-100,000+) HMD and most HCD use CRT displays. Flat panel display resolution has increased and size has been dropping dramatically over the last two years. Last year several vendors (Planar, Micron, Epson, Hitachi, etc) demonstrated color flat panel displays under 1 inch. The Phillips Scuba and Sony Glasstron are two HMDs using these devices to give NTSC video viewing. That is about 320x240 pixels for a computer image, same as many early HMDs. Phillips is currently selling reconditioned Scuba for US $99. Yes, a HMD under one hundred dollars, but low res and composite video. Sony announced the PC Glasstron (PLM-S700) in September. It will be an SVGA resolution (800x600) device available in Japan later this year. Kaiser Electro Optical (KEO) is introducing the ProView XL series this month. It uses the new small flat panels to deliver full XGA resolution (1024x768) stereoscopic color HMDs at a price competitive with professional VGA (640x480) displays. KEO is known for quality ergonomics in their ProView series, which saw a price cut prior to the introduction of the XL series. KEO also has the unique HiDef series that provides high resolution/narrow field of view (FOV) in one eye, and a lower resolution/wide FOV to the other. This combination rivals the performance of high end CRT systems.

An alternative display technology is the Virtual Retinal Display (VRD) developed at the University of Washington HITLab. It uses low power lasers to draw directly onto the eye. This produces a much brighter image than any of the screen based displays. Microvision has commercialized the technology and delivered several prototype HMDs for research this past year, including two to SAAB. The daylight viewable, see-through capabilities of a VRD was a major factor in securing a contract in October from the US Navy for sea trials of a HMD in December of this year. Real product is probably over a year away for the monochrome version. They are still looking for small green and blue lasers diodes or LEDs needed for a wearable color VRD.

Optics and their mountings are very important to a properly designed HMD. The optics is used to focus the image, spread it across a field of view and place it before the eye. Our eyes naturally turn inward (converge) to view close objects and outward (diverge) to view far objects. They also adjust focus (accommodate) based on view distance. These two physiological reactions are linked in binocular the perception of depth. Adjusting the horizontal position of the displays (aka interpupillary distance or IPD) allows more natural eye vergence. Adjusting the focal distance helps the eyes to properly accommodate. A large FOV is desirable but must be delivered within an acceptable eye motion box (exit pupil diameter) to avoid vignetting . Larger FOV also out spreads the pixels so each appears bigger. Designing an optical system to meet the sometimes conflicting HMD requirements is difficult. The digital lens technology from Retinal Displays may be about to change this aspect of HMDs.

The supporting device of a HMD is the third critical component. It must distribute the weight of the display and hold it snuggly but comfortably. Too many HMDs require the user keep one hand on the display to balance and keep it in place. Head Coupled Devices like the Fakespace Boom and Push systems use external mechanical supports. This gives the benefits of a head tracked display without the head mounting problems. The mechanical trackers can also greatly reduce lag time. Virtual Binoculars are another alternative approach that requires no head support, and vignetting is almost expected.

The emerging use of HMDs in non-immersive viewing applications like augmented reality and wearable computers is helping to reduce costs. There are now a number of monocular (one-eye) displays available (or nearly so) that cater to these markets. Xybernautís new wearable includes a bone conducting microphone for voice command. One very interesting monocular display is MicroOpticalís Eyeglass Display System (EDS). It relays the display image through reflectors within the lens of a pair of eyeglasses. Instead of a visor or Ďborg eyepiece, there is only a small prism visible on the glasses. I donít know how well this approach would work with a two eye HMD, given the greater requirements for proper optic positioning. I wonder if they could use a VRD instead of an LCD? That might make me really consider a wearable.

If you have any questions, comments or suggestions, please feel free to contact me by electronic mail at isdale@acm.org.