PDA

View Full Version : Why FOV cannot be made much larger than 90 deg.



Lurch1962
03-20-2007, 10:13 PM
I've occasionally seen comments in other threads whereby someone will pine for the ability to be able to zoom out to FOVs wider than the current 90 degrees. A view angle much wider than 90 degrees would result in a most highly distorted view.

Like any flight simulator I've ever seen, IL-2 uses the Gnomonic projection. To visualize this kind of projection, imagine a transparent globe of the Earth, with a tiny light bulb located in its exact centre. Place a very large, perfectly flat sheet of paper against the globe, say, at the location of your home town. Now turn on the light, and observe the lines of latitude/longitude and coastlines, etc. as projected onto the paper.

Within relatively close distances of your home, all will look pretty normal. But more distant regions will become progressively enlarged and stretched. In fact, it's impossible to map a full hemisphere (180 degrees) or more of a globe. Any place that's located 90 degrees from the centre of projection (your home town in this example) will appear to be projected infinitely far away on the sheet of paper.

So as you can see, with the Gnomonic projection there is an absolute limit of 180 degrees (a 90 degree radius from the centre of projection) for any projected view. In practice a more realistic limit, after which radial distortion becomes utterly extreme, would be about 120 degrees. But even that looks pretty weird. As I recall, the MS Flight Sim line allows such radically wide FOVs.

A little-appreciated aspect of this is the variation of image scale across a wide angle view generated with the Gnomonic projection. If one were to generate a near-180 degree wide view, the central portions would be in effect infinitesimally small. A plane crossing the field would be a wee point in the central parts of the view, but would expand to a huge, near arrow-like streak as it approached the edge of the view.

I'm sure you've all seen the effects of Gnomonic distortion while playing IL-2. The most readily observable symptoms are the enlarged oval versions of circular objects which appear toward the edges/corners of the screen when using the wider FOV settings, especially the widest.

To illustrate this even more forcefully, I've prepared a 3-panel graphic from cropped screen captures, available here:


http://picasaweb.google.com/Lurch1962/Il2/photo#5043820482345398850


So imagine just how much worse it would be with even wider FOVs available!

Now, if the Gnomonic has such limitations, why use it? Because it's the most efficient, hence rapid, projection to generate. That's because no matter where in the field they're drawn, straight lines are always straight. All other projections will make straight lines into curves if they do not exactly pass through the center of projection.

The maintenance of straight lines is especially important when objects are drawn as polygons. Think of the extra computation required to render a triangular polygon with smoothly curving sides, and to which a texture must be mapped!

------------------------------------------------

Another advantage of the Gnomonic projection... It allows for the seamless matching together of separately rendered views, as long as the FOV/scale are the same for each.

For example, the TripleHead2Go system is successful because of this. If the three monitors are of the same size and each is displaying its portion of the view at the same FOV setting (which necessarily is the case), the mating edges of adjacent screens match perfectly with no gaps or overlap. If the viewable areas of the screens could actually be made to come into contact, the combined 3-view set-up would be truly contiguous (seamless).

And finally a last tid-bit to toss out to you. Imagine building a sim-pit and placing it on a slim pedestal within a perfectly cubical room so that your head is located at the very centre of the room. Now imagine a 6-projector set-up, with each displaying a square (e.g., 1200x1200 pixel) image on each interior face of the room. That is, one view for each of the front, back, left, right, up and down view directions. The Gnomonic projection would result in a perfectly distortion-free, seamless view of the world as seen by the player, by virtue of the fact that he himself is located at the centre of projection. Wouldn't that be neat?


--Lurch--

Lurch1962
03-20-2007, 10:13 PM
I've occasionally seen comments in other threads whereby someone will pine for the ability to be able to zoom out to FOVs wider than the current 90 degrees. A view angle much wider than 90 degrees would result in a most highly distorted view.

Like any flight simulator I've ever seen, IL-2 uses the Gnomonic projection. To visualize this kind of projection, imagine a transparent globe of the Earth, with a tiny light bulb located in its exact centre. Place a very large, perfectly flat sheet of paper against the globe, say, at the location of your home town. Now turn on the light, and observe the lines of latitude/longitude and coastlines, etc. as projected onto the paper.

Within relatively close distances of your home, all will look pretty normal. But more distant regions will become progressively enlarged and stretched. In fact, it's impossible to map a full hemisphere (180 degrees) or more of a globe. Any place that's located 90 degrees from the centre of projection (your home town in this example) will appear to be projected infinitely far away on the sheet of paper.

So as you can see, with the Gnomonic projection there is an absolute limit of 180 degrees (a 90 degree radius from the centre of projection) for any projected view. In practice a more realistic limit, after which radial distortion becomes utterly extreme, would be about 120 degrees. But even that looks pretty weird. As I recall, the MS Flight Sim line allows such radically wide FOVs.

A little-appreciated aspect of this is the variation of image scale across a wide angle view generated with the Gnomonic projection. If one were to generate a near-180 degree wide view, the central portions would be in effect infinitesimally small. A plane crossing the field would be a wee point in the central parts of the view, but would expand to a huge, near arrow-like streak as it approached the edge of the view.

I'm sure you've all seen the effects of Gnomonic distortion while playing IL-2. The most readily observable symptoms are the enlarged oval versions of circular objects which appear toward the edges/corners of the screen when using the wider FOV settings, especially the widest.

To illustrate this even more forcefully, I've prepared a 3-panel graphic from cropped screen captures, available here:


http://picasaweb.google.com/Lurch1962/Il2/photo#5043820482345398850


So imagine just how much worse it would be with even wider FOVs available!

Now, if the Gnomonic has such limitations, why use it? Because it's the most efficient, hence rapid, projection to generate. That's because no matter where in the field they're drawn, straight lines are always straight. All other projections will make straight lines into curves if they do not exactly pass through the center of projection.

The maintenance of straight lines is especially important when objects are drawn as polygons. Think of the extra computation required to render a triangular polygon with smoothly curving sides, and to which a texture must be mapped!

------------------------------------------------

Another advantage of the Gnomonic projection... It allows for the seamless matching together of separately rendered views, as long as the FOV/scale are the same for each.

For example, the TripleHead2Go system is successful because of this. If the three monitors are of the same size and each is displaying its portion of the view at the same FOV setting (which necessarily is the case), the mating edges of adjacent screens match perfectly with no gaps or overlap. If the viewable areas of the screens could actually be made to come into contact, the combined 3-view set-up would be truly contiguous (seamless).

And finally a last tid-bit to toss out to you. Imagine building a sim-pit and placing it on a slim pedestal within a perfectly cubical room so that your head is located at the very centre of the room. Now imagine a 6-projector set-up, with each displaying a square (e.g., 1200x1200 pixel) image on each interior face of the room. That is, one view for each of the front, back, left, right, up and down view directions. The Gnomonic projection would result in a perfectly distortion-free, seamless view of the world as seen by the player, by virtue of the fact that he himself is located at the centre of projection. Wouldn't that be neat?


--Lurch--

Philipscdrw
03-21-2007, 05:44 AM
Interesting post? Article? This should be placed in a Collection of Useful Il-2 Stuff somewhere I think.

grifter2u
03-21-2007, 07:28 AM
excellent post !

Manos1
03-21-2007, 10:53 AM
I hear what you say Lurch but, look at what I mean http://forums.ubi.com/images/smilies/16x16_smiley-wink.gif

http://www.e-335thgr.com/Athos/TripleHead2Go_01.JPG

If it is "Field of Forward View" degrees? you people are looking for, you are right.

If you are talking about Field of Vision, you may be interested to know that this game, since its very conception, was able to handle multiple view (left, front, right) at the same time...

more info on http://forums.ubi.com/eve/forums/a/tpc/f/49310655/m/4521077615

(picture taken with all monitors put into a line in order to make the picture).

~S~

Pollack2006
03-21-2007, 11:29 AM
What we need is the option to de-couple the cockpit FOV from the environment FOV.

Lurch1962
03-21-2007, 04:30 PM
Pollack,
Decouple those two views? That would be disorienting, in a way. I know that the MS line of flight sims (the combat and earlier civvie line, at any rate) do offer a 2-D cockpit view which is "detached" from the outside world, in that as you zoom in/out the 'pit remains always at the same size. I really hate that! Much more preferable, in my view, to have all elements of the view behave the same.

By the by, I forgot to specify an important detail in the last paragraph in the original post. The 6 square views (1200x1200 in the imaginary example) to be projected on all 6 sides of the cubical room must each have a FOV of 90 degrees in order that they all match up seamlessly.

--Lurch--

Pollack2006
03-22-2007, 01:52 AM
Well what i'm saying is why should the environment expand and shrink in size just because the player moves his seat position forward or back?

An option to keep the cockpit displayed in wide-view and the environment rendered in medium-view would solve the trouble of aircraft appearing too small on screen.

Hawgdog
03-22-2007, 04:52 AM
Pfft, if Oleg could make the Pe-8 out roll a P-47 he could make a wider view!

Philipscdrw
03-22-2007, 06:03 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Hawgdog:
Pfft, if Oleg could make the Pe-8 out roll a P-47 he could make a wider view! </div></BLOCKQUOTE>Well, if the Avro Vulcan can outmanouever the McDonnell-Douglas F-15 then are you sure that's inaccurate? :-P

Lurch1962
03-22-2007, 08:53 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Well what i'm saying is why should the environment expand and shrink in size just because the player moves his seat position forward or back? </div></BLOCKQUOTE>

Pollack,

Moving the seat fore/aft (so as to better see the reticle for aiming) only changes the perspective of the cockpit (e.g., the gun sight and panel *appear* to zoom a wee bit as your virtual face gets closer to them when leaning forward). Distant objects are not affected in this way at all, because the pilot's moving only about a foot, which is nothing compared to something hundreds of feet away.

What we're referring about here is *zooming*, which is exactly like the zoom lens on your camcorder or digital camera. In IL-2, the zoom ratio is close to 3:1, due to the 30 - 90 degree wide FOV range (it's not *exactly* 3:1 because of the scale distortion introduced by the Gnomonic projection used, but who's quibbling? http://forums.ubi.com/images/smilies/16x16_smiley-wink.gif)

--Lurch--

Pollack2006
03-22-2007, 12:33 PM
You're right, i'm getting myself mixed up between zoom and FOV. What i'm clumsily trying to convey is the desire to see external objects appearing on my monitor the same size as if I was really looking at them from a cockpit. The middle zoom setting seems closest currently but you're then stuck too close to the dashboard of the aircraft.

Dunno if you've played RACE, a touring car simulator....there's an option whereby you can adjust your position in the cockpit and then adjust the field of view for the view out. It's very effective

Lurch1962
03-22-2007, 05:12 PM
Pollack,
1) What's the width of the viewable area of your screen (measure this with a ruler, if you would--the stated diagonal dimension as listed in documentation won't do)?
2) And at what distance do you typically plant your face from the monitor while "flying"?

With these two values I can tell you just what the FOV should be set at in order to see the outside world at true-to-life scale.

I know just what it is that you desire regarding independently settable image scales for the cockpit and externals. In the RACE game you referred to, I suspect that the method used is more along the line of separate *zoom* settings for these two elements, and not simply a positioning of the virtual driver's eyes within the car's cabin. (If the latter is the case, then the driver's perspective, with respect to the cabin, must be unrealistic from some settable positions.)

In reality, be it a car or plane cockpit, there is a rather limited degree of freedom for the driver/pilot to comfortably place his head and still access the controls.

IL-2 is a simulator, and the emphasis is on realism to the best extent possible. The zoom function is of course not a real operation the eye can perform, but it's a very useful way to overcome the limitations imposed by a computer screen's limited resolution. If you had a bionic eye that could zoom like Steve Austin's (in that old chestnut of mid-70's TV, The Six Million Dollar Man), the game's representation mirrors just what you'd experience. And that includes the changing apparent size of the cockpit along with the outside world--they always change scale in step and to the same degree.

To reiterate, when one changes to a narrower (zoomed in) FOV, on'e face is not closer to the instrument panel, gun sight, etc. The virtual pilot's eye point has in no way changed position. It's just that his "bionic" eye has zoomed in a bit, thus enlarging the apparent size of *everything* in his field of vision, whether it be near or far.

I'll repeat here what I've said numerous times in the past... Unless you sit rather close to your monitor, with almost all FOV settings you're in effect looking at a "de-magnified" view. Only with the more zoomed-in views (e.g., 40 degrees and less, for typical viewing distances) is the image scale more like that in reality. But of course the tunnel vision-like effect is hard to fly and fight with, so we must compromise and use wider, de-magnified, FOV's in order to achieve sufficient situational awareness.

I get around this limitation by using reading glasses and placing my eyes about 12-14 inches from my screen. This makes for a more realistic scale at the wider FOV's. (And more peripheral vision is used, too, which further improves situational awareness.)

--Lurch--