Information From the Periphery
Because the central visual field, especially the central few degrees, has so much higher acuity than the peripheral field, so much greater bandwidth per unit solid angle, it is tempting to assign a similarly greater importance to it. However if we recur to the concept of information in the Shannon sense (Claude Shannon in "The Mathematical Theory of Communication") a maybe arises. You are in your SCUBA outfit. You are concerned about sharks. You find that acknowledging a shark at the edge of your field of view requires more information than if the shark is center field. This is because you look where you expect to see something, so a shark at the edge of the field is less expected than one in the center. Indeed, to be certain that a shark is at the edge of the field, you look there quickly to add the necessary information to drive the message home, or to change it. The peripheral field thus is important as a gathering place for far more messages than the central field can cover. Because it covers a much larger area, the periphery will reveal most dangers first, no matter how cleverly you keep watch.
Why a Wide Visual Field is Critical to Immersive VR
by Eric Howlett — February 25, 2006
To (most of) us, the real world is more familiar than computer-generated worlds, and therefore it makes a good laboratory for getting back to basics. I want to start by asking you to examine your world in a way that you may not have done before.
The Human Fields of View
You can make a useful estimate of your own fields of view (FOV) without moving from your chair. Do the following — without turning your head:
- Swivel your eyes as far to the right as you can.
- Raise your right index finger at arm's length to your right.
- While wiggling the finger, move your hand slowly to the right.
- Stop when the wiggling finger begins to disappear out past the right edge of your visual field.
- Hold your left index finger up straight ahead in front of your nose.
You can easily see that the angle between the two fingers is large. In fact, measurements show that this angle is between 130° and 140° for the average face. The left side of your visual field is roughly the same size as the right, so doubling this number shows that your total lateral field of view is somewhere near 270°.
- Now close your right eye and look at the bridge of your nose with your left eye.
- Again, raise your wiggling index finger at arm's length and move it to the right.
- Stop when the wiggling finger begins to disappear.
- As before, hold your left index finger up straight ahead in front of your nose.
You'll notice that the angle between your two fingers is now considerably smaller. The actual angle measurement depends on the shape of your nose, but probably lies somewhere between 40° and 60° Repeating the experiment with the right eye will yield a similar result for a total lateral stereoscopic region of 80° to 120°. That's significantly less than 270°. By now you may have realized that the size of your nose is pretty important here.
- 1. Close one eye.
- 2. Look around at objects in the world around you.
Does the world seem any less three-dimensional? You might not be able to catch a fast-moving football or land an airplane, but the answer is no. The world around you still feels real and you still feel immersed within it.
You have just seen that the combined total lateral FOV of your two eyes is somewhere around 270°. You have also seen that the area of the lateral field that your two eyes share — that is to say, the sterescopic field of view — measures somewhere between 80° and 120°, depending strongly on the shape of your nose.
It should now be clear that the absence of stereo perception over most of your visual field is not something you complain about much. Why is that?
Already, you have reasonably confirmed that your real-life, bare-faced, lateral FOV is only about one third stereoscopic. Although it varies widely, the following sketch outlines approximately the entire pear-shaped region that both eyes can fix on simultaneously for a typical face.
The very top of the pear is formed by the central portions of the brow ridges. The very widest part of the pear is formed by the left and right views of the bridge of the nose, and the narrowest part by the tip of the nose. At the bottom, the pear angles sharply down to a point, following the contours of the cheeks.
Because we used the widest part of the stereoscopic field in our calculations, the estimate of the ratio of the solid angle of your stereoscopic field to your total visual field would be even less one third — probably ten to twenty percent of your total solid-angle visual field. These numbers, of course, depend strongly on the size of the nose that circumscribes the sides of your stereo field. In any event, assuming both eyes work, even people who can't see stereo at all have a similarly-sized total solid-angle visual field.
Relevant Observations at LEEP Systems
Cyberface1 had two co-planar LCDs, each centered on one of the two parallel optical axes of the LEEP stereophoto magnifiers, the same type used by VPL and NASA Ames. In viewing very wide-angle stereo photos, we had noticed that the stereo field was a bit wider than in real life because the magnifiers showed parts of the field that would be behind the nose in real life. We didn't make much of it.
Cyberface2 had optical axes that diverged 25° — much more than any normal eyes can diverge. We did this mainly to accommodate larger LCDs (more pixels) at angled focal planes centered on, and perpendicular to, the diverging optical axes of the two lenses. The added total FOV was a nifty bonus. Nobody ever, so far as I know, commented on the narrower stereo field.
Cyberface3 had only one, still larger, LCD viewed with both eyes simultaneously through prismatic lenses. First-time viewers almost always assumed that they were seeing a stereo image and often were only persuaded otherwise once it had been pointed out that there was only one LCD. The lateral field of view was barely 70°— about half that of the Cyberface2 — but it was still wide enough so that they didn't complain of the loss of stereo — a fact that was actually disputed by some. Only recently have I realized in retrospect that this was a clear indication that a substantial FOV, at least in a display that permits turning the head, was a much more powerful agent in creating a sense of immersion than stereopsis ever could be.
Implications for VR HMDs
The absence of stereo is not a significant loss to the illusion of presence — even when the FOV is only 70 degrees. The wider the field, the greater the illusion of presence — up to that 270° mark. To provide an immersive VR experience, HMDs should show as much of the 270 degree field of view as possible. A large FOV in an orthoscopic rendering provides a greater sense of immersion than stereopsis in a narrow field. So while Videowrap, as with the Cyberface1, supports full stereoscopic region beyond what is possible in the real world, the critical factor for Virtual Reality immersion is the FOV.