In the past, 3D stereoscopic displays and 3D tracking systems have kept the prices of fully immersive VR environments very high. Recently, however, new commodity technology has become available, and it is now possible to build a fully immersive VR environment, i.e., a head-tracked stereoscopic display system with tracked input devices, for about $7000. This web page documents how to assemble one such system, based on a 3D-capable high-definition DLP projection TV, such as those offered by Samsung or Mitsubishi, and the NaturalPoint OptiTrack optical tracking system.
|
| Figure 1: A quartz crystal being manipulated using the Nanotech Construction Kit on a Samsung 3D TV and a tracked Wii controller. The "tracking antler" mounted on the Wii controller, consisting of three retro-reflective balls, is picked up by the optical tracking system to determine the controller's position and orientation in space. |
With the release of the Razer Hydra 6-DOF desktop input device, it is now possible to build even lower-cost entry-level VR environments. While not head-tracked, and therefore not immersive or "holographic," these are still very useful for a wide variety of 3D applications, such as the Nanotech Construction Kit as shown in this video and in Figure 2. The only required components are a graphics-capable PC, a 3D TV, and the Razer Hydra (for around $150). See the installation instructions.
 |
| Figure 2: A low-low-cost VR environment using a 3D TV and a Razer Hydra 6-DOF desktop input device. |
One of the more important factors when deciding what 3D TV to buy is stereo quality; unfortunately, this is also the most difficult to evaluate, because it is typically not mentioned in the specifications for a particular TV model, and it cannot easily be gauged by viewing a TV model in a showroom. This is because typical 3D content is low-contrast and has low eye separation, following the "standard" guidelines for 3D content. These guidelines originate from rules of thumb put up by people who are doing stereo wrong, and tried to minimize the ill effects of their fundamentally flawed approach by "flattening" the 3D images they display. It's a long and sad story. Anyway, using 3D TVs as we do exposes limitations of the technology that the manufacturers might not even be aware of.
By "stereo quality" we primarily refer to the amount of cross-talk between the displayed left and right eye images. Ideally, cross-talk should be 0%, in which case left and right eye images are completely separate. In non-zero crosstalk, images from one eye show up more or less faintly in the other eye, which causes problems when the viewer's brain tries to fuse the two images into a 3D view. The more cross-talk there is, the more difficult is it for the viewer to perceive the 3D effect, and at some viewer-dependent level of cross-talk, 3D breaks down completely.
Cross-talk is caused by both the TV, and whatever eyewear is required to view the TV. In active (time-interleaved) stereo, it is caused by the TV's ability to switch between images rapidly, and by the active shutter glasses' ability to switch between opaque and transparent rapidly. On the TV side, projection DLP TVs have 0% inherent cross-talk, because the main benefit of the DLP display technology is its ability to switch between images extremely rapidly. LCD-based TVs have inherent cross-talk because the per-pixel LCD elements cannot switch rapidly enough. Active shutter glasses use (large) LCD panels as well, and have the same problems. Polarization-based "passive" 3D TVs use polarization filters on the display, and polarized glasses, to create separate left/right images. Here cross-talk is caused by the less-than-perfect polarization filters on both ends.
The first generation of 3D TVs were all based on DLP projection technology, since their 3D capability was an incidental side effect of the then-current method of creating a 1920x1080 HD picture with a micromirror with only half as many pixels using a method called "wobulation." These days, there are several types of 3D TVs based on different combinations of technologies:
The bottom line is that, unless one is able to pay for an OLED 3D TV, there is no clearly superior technology; all stereo methods have their drawbacks. Picking the proper TV model is a matter of what exact setup one has in mind, what screen size and footprint is desired, and how much money one has to spend. The good news is that all technologies listed above are supported natively by our software, and work out-of-the-box.
With the advent, or renaissance, of consumer head-mounted VR over the last several years, real low-cost VR has finally arrived. It is now possible to pick up a fully-functional and full-featured VR headset, specifically, as of November 2016, an HTC Vive, at many online or brick&mortar retailers. The only other component needed is a high-performance gaming PC with a high-end graphics card such as, of November 2016, an Nvidia GeForce GTX 1070. The entire bundle can be had for less than $2000, easily.
Also as of November 2016, the Vrui VR toolkit, version 4.2 or higher, runs natively on HTC Vive headsets on Linux, and basically out-of-the-box, without shopping for individual components, custom engineering, or high-precision calibration requiring specialized tools. Therefore, the low- and low-low-cost VR systems described on this page are basically obsolete for most applications and use cases. This page will remain online for archival purposes, and for the occasions where big-screen projected VR environments still have a leg up on head-mounted VR, specifically, collaborative data analysis. For those cases, it might be ideal to combine a low-cost VR system as described here with a head-mounted VR system in a single shared environment.