Re: focal lengths

[P3D John Bercovitz <bercov@xxxxxxxxxx>]

>Ray Hannisian: If using a higher focal length simulates
>>being closer to an object, should 'zooming in' require an
>>increase in the distance between the cameras (to simulate
>>the 'line of sight' angles for a person actually standing
>>closer to the subject)?
 
>Larry Berlin: Yes, adjust the stereo base (distance between 
>cameras) to obtain the equivalent separation angle based on 
>how close the object appears on film and how much parallax is 
>wanted. Is there a *right* term for this angle?
 
I'm a little confused as to these angles you guys are talking 
about.  Could someone expand on this?  You're not talking about
using toe-in instead of parallel views, I think.  Something else.

I mean it sounds like you're talking about vergence but vergence 
takes place when you're looking at views, not when you're taking
pictures.  If you had vergence when you took pictures, you'd
have vergence^2 when you viewed them resulting in the distortion
of reconstructed space so ably described and calculated by the
researcher Andrew Woods in his seminal paper on the subject.

I really don't think you're saying that so what is it that I've
twisted here?

If I interpret it to mean increase camera separation (keeping
camera backs parallel) to simulate the vergence of a closer
view and hence give the parallax levels you would get in a 
closer view, then I understand but you have to recognize that
this causes the reconstructed scene to be smaller in all dimensions.
It's the famous "Lilliput" or "dollhouse" effect.

If you use increased camera focal length relative to viewer focal 
length to increase parallax, the result is disproportionate, causing 
compression of the reconstructed third dimension relative to the 
other two dimensions.

So I'm really confused here.   Would appreciate any help.

Thanks,
John B

Quote of the month:

Larry Berlin wrote:
> It [cardboarding] has to do with a situation where there is 
> sufficient parallax to determine the relative placement of  
> various subjects within the scene, but insufficient parallax  
> visible to see roundness or relative depth of [within] each 
> individual visible subject.

If that doesn't say it, I don't know what does!  I've tried to
say that so many times in 100 times as many words.  8-)



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