Re: 3D stereo panoramic Cirkut photography dilemas (fwd)

[John Roberts <roberts@xxxxxxxxxxxxxxxxx>]

>Date: Thu, 16 Nov 1995 19:11:40 -0600
>From: s.spicer@xxxxxxxxx
>Subject: Re: 3D stereo panoramic Cirkut photography dilemas (fwd)

>Somebody wrote (I don't have the original reference):

Arvid Olson (Nov 14-15)

>>Working on Stereo Panoramic Cirkut photography using 2 #8 Cirkut
>>cameras to get two prints 8" wide by 5 feet long and laying the images
>>on top of each other and getting a 360 degree panoramic view in 3D by
>>utilizing a over/under prism mirrored viewer merging the 2 images.  Has
>>any body worked out the distance the cameras would have to be separated
>> in order to get a 3D panoramic effect.

>This made me think about an issue that I think was raised a long time ago.
>If you consider a stereo image from a fixed pair of cameras that can cover
>a 180 degree field, the effective camera separation drops to zero for
>images entering the camera lense from the far left or far right (ie: +/- 90
>degrees from on axis). So, the stereo effect diminishes for off-axis
>images.

>I am probably imagining a camera set-up that is too simple, or maybe there
>is some way around this problem. Are there any comments?

>Steve Spicer

What you say is true, but it also applies to human vision - when you shift
your eyes to the side, the effective interocular is reduced, and therefore
the degree of depth perception. The way around it is to mimic the human
ability to turn the head, in order to look at things that are off to the
side with the eyes still pointed straight forward from the face. The
360-degree cameras would do this - I presume this is the type that runs
the film past a narrow slit while the camera rotates. (The cameras should
be rotated together as one body, ideally about a vertical axis at or somewhat
behind the front nodal points of the two lenses, but using the rotating
mechanism of one camera to drive both cameras may produce acceptable results
if no subject is too close.) I'm somewhat concerned about trying to view the
results in a viewer that requires the eyes to shift to the side - seeing full
depth at an angle where depth would normally be reduced may or may not result
in a reduced sense of reality.

For viewing a 360-degree panoramic 3D photo in a manner that allows you to 
rotate your head an keep the eyes pointed straight, there are several possible
options. One is the commercial system called SPACESHOT (reference attached at
the end of this post). 

Several others involve variations of the technique Arvid described. The ideal
way to view a 360 degree panoramic strip is to roll it into a cylinder, and
stick your head into the cylinder. This eliminates all the distortions caused
by the panoramic mapping. To get a reasonable eye-photo distance, the cylinder
should be as large as possible - 20-foot radius may be ideal, but smaller may
be acceptable. For prints or transparencies, stereo could possibly be achieved
by the prism viewer, but anaglyph or Vectograph would probably work better
since they don't require precise orientation of the head (so there's no need
for a head restraint or pedestal).

Huge prints or transparencies are bound to be expensive - there may be a way
to set up a 360-degree projection system. (The Canada exhibit at EPCOT does
a pretty good job in 2D with multiple projectors and screens all the way
around the theater.) For a projection system, 3D could be provided using
polarization or perhaps liquid crystal shutters.

The methods I described would be for still pictures. I can't think of a way
to get completely smooth 360-degree 3D for video, but if multiple screens with
small gaps between them are acceptable, then a 3D variation of the technique
used in the Canada exhibit may work.

In answer to Arvid's original question, the "1 in 30" rule for camera
spacing would probably be a reasonable first approximation.

John Roberts
roberts@xxxxxxxxxxxxxxxxx

---- Attached reference ----

>Date: Tue, 23 May 1995 23:32:32 -0500
>From: John Roberts <roberts@xxxxxxxxxxxxxxxxx>
>Subject: A new 3D viewing system

>One of the exhibitors at SID '95 has a new 3D viewing system called SPACESHOT.
>It uses a rotating 2-lens camera to take a 360-degree panoramic stereo
>photo on 35mm film. The two pieces of film are formed into loops, and mounted
>in a cartridge. The cartridge can then be inserted in a hand-held viewer
>which contains drive motors and a solid-state gyroscopic sensor. When you
>turn the viewer on and look into it, you see a stereo scene. As you turn
>your head (and the viewer), the sensor detects the motion and rotates the
>film loops to pan the 3D scene in accordance with the motion of your head.
>The result is that you get the impression of being able to look all around
>you at a 3D scene. (There's also a pedestal format for exhibitions.)

>I spoke to the exhibitors, and asked if they were aware of the Photo-3D
>mailing list. They replied that they are aware of it, but their only network
>connection is Compuserve, and they have had trouble getting to the Internet.
>They added that they have had trouble explaining their 3D system to the readers
>in the Compuserve forum. They were enthusiastic about getting their product
>mentioned on Photo-3D.

>The address of the company is:
>  Pan3D Ltd
>  Premier House
>  11 Marlborough Place
>  Brighton, England BN1 1UB
>  Tel & Fax +44-1273-556809
>  CompuServe 100022, 3456

>John R

[According to Andrew Peters, the price is in the $15K range.]



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