Re: Converging fields of view

[P3D Larry Berlin <lberlin@xxxxxxxxx>]

Hi All,

I haven't had the pleasure of using a "stereo" camera so perhaps some of the
stereo cameras do have non-parallel fields of view. If you are using a
single camera with still objects you have to slide the camera sideways to
obtain the second shot. If you do this very carefully with a slider bar, the
"plane of coincidence", as Gregg calls it, is at infinity. Masking of an
image will establish the "window" at some other depth.

This topic is very device dependent and has very little meaning unless you
are very specific about which device you are discussing and all the exact
specifics of an individual circumstance. I checked a bunch of my stereo
images and sure enough you appeared at first to be right, except for the
fact that where the discrepancy you mention is visible in the full image
frame, it only proves that I didn't maintain my camera at a parallel angle
for each shot in the pair. If such is true of images taken with a stereo
camera, then it is not designed with parallel fields of view, or mounting
has influenced where the "window" is located.

In a two camera shot, the center line of the lens axis matches the center
line of the field of view. If the two cameras are maintained parallel to
each other, and each lens has the same angle of total view, the left side
camera will see more of the left field than the right side camera, all the
way out to infinity. Your system may actually have some amount of
"convergence" in it's fields of view, ie: non-parallel. If the aperture is
offset from the center of the lens, it will result in the "converging"
fields of view that you describe, and could still have a parallel lens axis. 

Full frame stereo images shot by two cameras, maintained exactly parallel,
will not display the differences you describe. Images placed in mounts will
display a "window" wherever the masking has set it. 

I'm so used to working in the computer and not having a fixed "window" to
worry about that I had to get out the camera and a bunch of stereo pictures
to check this out. It helps deepen the overall sense of stereo to observe
these kinds of details in the images. 

Thanks,



>Neil Harrington  <nharrington@xxxxxxxxxxxxx> Writes:

>>Larry Berlin writes:
>>It is true that the fields overlap but it is NOT true that they converge.
>>The fields spread out fan like and OVERLAP as they proceed in a generally
>>parallel arrangement. The center of each field is OFFSET from the other and
>>these centers are supposed to be parallel for the least amount of
>>distortion. 
>
>Not so, Larry.
>
>Several here have observed that the LEFT frame includes more of the RIGHT
>side of the subject, and vice versa.  As I have mentioned, this is true, but
>only beyond the window.  Put a subject close in front of the camera and you
>can easily see that at closer distances the reverse is true: the left frame
>will contain more of the left side of the subject, and the right the right.
>Move the subject as close to the lenses as is necessary to convince yourself
>of this.    ;-) 
>
>So the field of view as experienced (so to speak) by each side of the camera
>starts out entirely on its own side and ends up more on the other side.
>There is no way this can happen without convergence and crossing over.  The
>point at which crossover occurs establishes the window distance.
Larry Berlin

Email: lberlin@xxxxxxxxx
http://www.sonic.net/~lberlin/


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