Re: Large FOV

New Message Reply Date view Thread view Subject view Author view

Angus Dorbie (dorbie++at++sgi.com)
Mon, 14 Jun 1999 16:30:31 -0700

Angus Dorbie wrote:
>
> Devrim Erdem wrote:
> >
> > > Which distortion are you thinking of? Error as viewer moves away from
> > > the design eyepoint? Different packing of world-space information into
> > > displayed pixels as you move from the center of projection toward the
> > > edges? Errors of fog or lighting based on the common "Z-depth" rather
> > > than correct "radial range" as a function of FOV? Loss of Z-buffer
> > > resolution as a function of angle or distance from the center of a wide
> > > FOV display? Light fall-off from a flat screen based on angle of view?
> >
> > I can not think of the distortions you mentioned since this is my first VR app
> > using large FOV. A frog in a well says "The sky is as big as the mouth of my well".
> >
> > AFAIK, I shall expect a distortion when horizontal FOV exceed sth like 60. I was told
> > in a Vega training that this is because of the difference between a rectangle and
> > pie slice. The solution was to use 2 channels instead of one. Now I wonder if the
> > can be guessed correctly to prevent motion sickness or other problems.
> >
> > Secondly, which one of the distortions you mentioned is most likely to appear ?
>
> The one you are most likely to encounter is simple geometric inaccuracy.

I'm sure this is your problem now.

> It's listed above as "different packing" but is really a manifestation
> of your display geometry not matching your 'virtual' projection
> geometry.

Actually the pixel packing mentioned is a different problem due to tan
theta pixel distribution.

It's just ocurred to me that you might actually simply want to draw a
wide field of view and someone has misled you by suggesting that you
need to split the channel up because it looks funny. There are subtle
reasons you might want to do this (most covered by mtj's post) for
example resolution due to tan theta resolution distribution, but in a
simple projection there is no distortion. The point is that when you
display your 130 degree image, you must ensure that it subtends 130
degrees with the viewer when displayed. Just because you draw the image
and display it on your desktop and it looks distorted doesn't mean that
it is incorrect. The image probably only subtends say 40 degrees at most
w.r.t. your eye. If you can get closer to the screen and retain focus it
will begin to look correct. The biggest problem will be lack of image
resolution towards the center of the image, and z fog as already
covered.

Now, if you then want to wrap that image onto a curved screen, say a
Panoram display system then you have a different class of problem.

If you want to render an image in which image Y = constant azimuth then
conventional computer hardware doesn't render that kind of projection.
And the best way is to remap the pixels you do draw to an image where y
= azimuth.

Cheers,Angus.

-- 
"Microsoft's system was like a forest that hadn't had a controlled
 burn in decades, just waiting for one person with a match to turn
 it into a disaster. Melissa was Microsoft's fault. They left their
 system wide open to this sort of abuse, they knew it could happen
 and did nothing." -- Bruce Perens


For advanced 3D graphics Performer + OpenGL based examples and tutors:
http://www.dorbie.com/

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Jun 14 1999 - 16:30:37 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.