T3D Re: Why Scheimpflug?

[john bercovitz <bercov@xxxxxxxxxxx>]

Joel writes:
> If I understand correctly, Scheimpflug changes the plane of focus,
> while front shift changes the 'composition'.

Correct.  So what they did here was toe in the two CCD cameras and 
then use tilt of the lensboards and CCD planes to achieve a common
imaged area on both CCDs.  A common rectangle in object space, in
oterh words

> In a view camera, IMHO, front tilt to adjust plane of focus is the
> most important (and most common) use of camera movements. 

Hear hear!  My opinion too.  I've shot using rise and it works fine
as long as you mount the print high on the wall so your eyes are 
over the sub-lens point on the print.

> I've always thought that the Heidescop would have been better 
> designed with front tilt than front rise...

Absolutely.  Use of rise without moving the viewing lenses to match
woud result in a distortion.

BTW, I shot off a note to the author of the paper, we'll see if he
has time for a reply.  See below.  I hope I wrote nicer than I'm often 
accused of writing.  8-(  

John B

> I downloaded your paper from www.nasatech.com.
> Looks like a very interesting paper but it pretty
> quickly got to where I was slogging in the math.
> 8-)  Fortunately, I was mostly interested in the
> optical arrangement and I understood that OK, I
> think.
>
> The depth of field using Scheimpflug isn't constant
> over the field, as I'm sure you saw in Rudy
> Kingslake's book.  Is that a problem in your setup?
>
> You said, "The use of large field angles in the
> displacement technique increases the sensitivity
> to out-of-plane displacement, but will also produce
> perspective distortion in the image.", and then
> cited R. Kingslake.  I'm not sure how the lens
> displacement technique increases sensitivity to
> out-of-plane displacement.  Can you expand on that?
> On the perspective distortion, I think that is
> explained by viewing from the wrong distance.  I
> assume you are talking about the picture where the
> balls are made into ellipses on the photograph.
> If that photo were viewed from its center of
> perspective, the balls wouldn't look ellipsoidal.
> Rudy hints at this in the text.
>
> I can see how insufficient field overlap would be
> a problem for the displacement technique if you
> are viewing a plane which is close to the cameras
> relative to the cameras' separation.  However, CCDs
> generally don't have too much absolute resolution so
> a lens could be designed with enough coverage to use
> displacement, I would guess.  If the CCD were larger,
> you could use a commercial camera lens.


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