Visual stereoscopy comes in many flavours. Depending on how the video feed is constructed, and whether or not it was originally designed with stereoscopy in mind, the technology used to deliver a slightly different positional viewpoint to the eye, differs greatly.

Stereoscopy at its heart is about delivering in pairs. Most of us possess two eyes, and for those who do not, work is continually underway to create artificial replacements. These two eyes sit in general, about two inches apart, measuring from mid point to mid point. This means that particularly for close objects, each sees a different angle, and thus a different view. When the two images are combined we gain a 3D, spatial perspective of the object. If this spatial location sense is missing, i.e. by delivering the exact same image to both eyes, then our sense of spatial location ceases to function. Realism and believability drop through the floor, and interaction becomes a whole lot harder.

Virtual Spatial Display

In VR, from a developer's point of view, the easiest way of producing stereoscopy is also the most computationally expensive. Normally, each user has a viewpoint into the virtual environment, a cone of sight that opens up towards the horizon or back clipping plane, and which contains a rendered view of the world.

To enable stereoscopy, a second view frustrum must be opened up alongside the first, with a second viewpoint feeding it. Instantly, you are performing twice the work of rendering the frame, as you must render each frame from scratch twice. Predictably, this halves the frame rate immediately. Where you might have been hitting 72 fps before, you are now only getting 36. Older computer hardware in particular does struggle, as where they may have been seeing a choppy but still usable 10fps before, now they see only 5.

Because of this, most publicly accessible virtual environments do not utilise stereoscopy by default. However, many do have the option, or capability to add it. The frame rate hit can itself be largely negated by doubling the GPU capabilities on a computer system, and many PCs set up for stereoscopy run two graphics accelerator cards in parallel.

The problem then, is transferring the two separate viewpoints from the graphic accelerator cards to the user. This requires specialised hardware, as a standard monitor display cannot handle more than one viewpoint, at a time, in most situations. It can handle two viewpoints together in some situations, but again specialised viewing hardware is necessary in addition, to decode the images produced.

Hardware

Stereoscopic HMD

The head mounted display is the oldest of the stereoscopic displays. Originally not even head mounted, but held on a ceiling or floor mounted boom due to their immense weight, they consist of a unit which is worn on the head. The originals were like crash helmets with ovens strapped on top. Modern variants are not much bigger or heavier than a pair of spectacles.

A stereoscopic HMD has a separate monitor display for each eye. Typically these are LCD or plasma based nowadays, and recreate the world for that eye's point of view only, blacking out the glare of the image as it passes over the nose, and blacking out outside light. Sometimes, as with the HMD pictured, the light suppression system is removable so that the HMD can be used without restricting access to the outside physical world.

Shutter Glasses

Shutter glasses work with shutters that switch on and off many times a second, alternating between the eyes. This happens too quickly for your brain to process it, but forces a situation whereby whenever one eye is looking at the display, the other is looking at a black, closed shutter. The shutter glasses and the display system ? monitor, holoprojector, retinal laser etc ? work in tandem, either through wires, or more commonly now, Wi-Fi. The shutter glasses sync their shuttering speed precisely to the refresh rate of the display system, so that the display can show images appropriate to the left and right eyes, at appropriate times.

They are a counter to the problems with doubling the view frustrum. Rather, the graphics hardware alternates between the two views like toggling a switch. The result is you still perceive the same fps as a single display, but you are seeing it through alternate eyes.

Thankfully, this trick works because the brain is intelligent enough to piece the images together seamlessly.

Anaglyph Glasses

Anaglyph glasses are basically those 'cheesy' glasses the public generally thinks of when they picture 3D.Anaglyphs are based on a visual display method that uses colour differentiation to take a single video stream and split it into two channels before it reaches the eye. Typically it utilises filter glasses that mask one eye with a red tint, and the other with a blue or green tint. When a video source is played that is half green, half red, slightly offset, each filter allows one half of the image to pass. The two images are the same scene, slightly offset from one another, which again the brain reassembles into a single stereoscopic image.

An advantage of anaglyph glasses is that from one displayed image source, an entire roomful of people with glasses can see the 3D effect.

Polarized Glasses

Polarised glasses are not really so different from shutter glasses. The key difference is that both sides are half-shuttered all the time; they do not flick off and on. Instead they are polarised such that each half can only accept light beams whose wavelength is in a precise alignment. Typically, the left eye looks through vertical polarizing film, right eye looks through horizontal polarizing film.

Two projectors are used with polarised glasses, bringing us back to the two view frustrums issue and a halving of fps again. One projector projects light which has a vertical waveform, the other projector, a horizontal waveform. The two images are out of sync with one another, with one image destined for the left eye, the other for the right eye. If a user looks at the projected images without the filter, they see a confusing jumble of two not-quite-aligned images superimposed and offset, for every item on the display, in the sam manner as occurs with anaglyphs.

An advantage of polarised glasses is that like with analgyphs, an entire room full of users can witness the same image, passively in 3D from a single projector source. Of course, this analogy fails if the view is to be interactive.