[MF3D.FORUM:1684] Re: Using front shift to simulate camera movements?

[Tom Hubin <thubin@xxxxxxxxx>]

Hello Don,

The field of view limits are determined by the camera itself. In
particular, by the film size and the lens focal length. The amount of
light is controlled by the lens fNumber. No reason to loose any of the
field of view or light just because you use a mirror.

If you use too small a mirror then it acts like a rectangular aperture
placed near your lens. No loss in field of view but some light will be
lost just like increasing your fNumber. This may show up as vignetting.
If extreme is may cut into your field of view.

Lets draw a top view picture. Use graph paper if possible.

Draw the two camera lenses as you would without mirrors. These are
cameras without bodies so we can put them as close as we want to each
other as long as the lenses do not overlap. Draw the lenses as thin
lenses with just simple lines. Side by side. Correct diameter. Correct
interocular.

Now draw the film, as a thin line, the correct horizontal size.

Now draw field of view lines from the edges of each film through the
center of the corresponding lens. Extend the lines beyond the lenses as
far as you like (too the edge of the paper is fine). These large Xs
define your angular field of view. If this were a pinhole camera we
would be finished.

To avoid vignetting you need to evaluate the sides of the lenses as well
as the center. So from the extremes of each lens draw lines away from
the camera and parallel to the pair passing through the centers of the
lenses.

You now have a sketch of all of the light that can travel from a subject
at infinity, through a lens, and onto the film. Very close to the camera
lenses there is no overlap of the light paths because the lenses
themselves do not overlap. Some distance away from the camera lenses the
right path and the left path do overlap. Any apertures or mirrors in the
area where they overlap will be a problem of some sort. This cross point
is as far away as you can place the farthest (common) edge of the
mirrors. Farther away from the camera lenses than this cross point will
cause losses in light (vignetting) and possibly field of view if way
beyond this point.  

However, between that crossing point and the lenses you can impose the
right size mirrors with impunity.

Now here is a trick of the trade. Cut out the outline of your drawing.
There will be a V formed by the cutout of the triangle between the
lenses and the crossing point. Now fold each path at 45 degrees so that
the lenses are at right angles to the way you started. In particular,
fold the paper about where you would like to place the mirrors.

Now the figure looks like what you had in mind. The creases where you
folded the paper ARE the mirrors. As long as the mirrors are long enough
to cover all of the rays you loose nothing.

You may sometimes here mirrors like this referred to as folding mirrors.
Now you know why.

You may be able to do a second fold in each path. This would then
produce normal images rather than mirrored images. 

Also, do not limit your folding to 2 dimensions. You can fold lots of
ways. Both up or one up and one down or whatever. Or at angles other
than 90 degrees with 45 degree mirrors.

Give me some realistic numbers for lens focal length, lowest fNumber for
that lens, film width and I will come up with mirror size and location
that should work with no losses.

I think I have Power Point and Word here, although I have seldom used
either. Would a sketch in Power Point help? Would copying that sketch
into Word be better? Would it be useful to have such a sketch on
somebody's website?

Tom Hubin
thubin@xxxxxxxxx   

******************************

Don Lopp wrote:
> 
> Tom . if I am wrong ,please tell me wh at angle of field I can expect t be
> able to utilize . i do not epect it to be very much by normal stereo
> standards by which I do not consioder View Master a decent standard. DON.
> ----- Original Message -----
> From: "Tom Hubin" <thubin@xxxxxxxxx>
> To: "Medium Format 3D Photography" <MF3D.Forum@xxxxxxxxxxxxx>
> Sent: Tuesday, April 10, 2001 8:40 AM
> Subject: [MF3D.FORUM:1664] Re: Using front shift to simulate camera
> movements?
> 
> > Hello Don,
> >
> > Not true. It depends entirely on how you do it. If you connect all
> > points on the subject to all points on the lens and the mirror
> > intersects all of those rays then you loose nothing.
> >
> > For stereo there is a practical limit on this single mirror approach.
> > The system is creating the illusion that the lenses are side by side
> > with the camera bodies overlapping. They cannot appear closer than
> > touching each other without clipping something. For example, if the
> > lenses are 50mm diameter then 50mm is the smallest interocular that you
> > can create without clipping.
> >
> > To avoid clipping at 45 degrees you need a mirror that is the square
> > root of 2 larger in the folded dimension. So a minimum of 50mm by 71mm
> > ellipse would be needed in the above example.
> >
> > Tom Hubin
> > thubin@xxxxxxxxx
> >
> > **************************************
> >
> > Don Lopp wrote:
> > >
> > > Plus the angle covered would be relatively narrow DON
> >
> > > ----- Original Message -----
> > > From: "Bill Glickman" <bglick@xxxxxxxx>
> > > To: "Medium Format 3D Photography" <MF3D.Forum@xxxxxxxxxxxxx>
> > > Sent: Wednesday, April 11, 2001 12:31 AM
> > > Subject: [MF3D.FORUM:1657] Re: Using front shift to simulate camera
> > > movements?
> > >
> > > > Tom
> > > >
> > > >         I'm not understanding this....  If you look forward you see an
> > > > image, if you turn 90 deg and use one mirror at 45 deg, you will see
> the
> > > > same image.  So why wouldn't the film see the same thing looking in
> the
> > > > mirror?  Are you sure you would have to view the slide with the
> emulsion
> > > > side away from our eyes to get the same effect as if the cameras shot
> > > > straight at the scene?  Hmmmm..if so, that does not sound good, but
> then
> > > > again we look straight through the slides anyway?
> > > >
> > > > Bill g
> > > >
> > > > > Clever idea but the images will be mirrored. So you will need to
> print
> > > > > with the emulsion side up or view slides with the emulsion side away
> > > > > from the eye or project with the emulsion side away from the screen.
> > > > > This will reduce the quality of the image.
> > > > >
> > > > > A pair (or any even number) of mirrors for each camera is more
> > > > > complicated but would avoid the mirror image problem. For that
> matter,
> > > > > two mirrors on one camera and no mirrors on the other camera would
> also
> > > > > work. But then the distance from camera lens to subject is not the
> same
> > > > > for both cameras.
> > > > >
> > > > > Tom Hubin
> > > > > thubin@xxxxxxxxx
> > > > >
> > > > > **********************************
> > > > >
> > > > > Matthew V. Ellsworth wrote:
> > > > > >
> > > > > > Bill
> > > > > > I'm not sure if I'm understanding what you want to do, but maybe
> this
> > > > idea will
> > > > > > help:
> > > > > > Two cameras can be mounted facing each other on an adjustable
> rail --
> > > > with a
> > > > > > small front-surface mirror mounted at a 45 degree angle in front
> of
> > > each
> > > > lens.
> > > > > > This allows you to get both cameras quite close (limited by the
> size
> > > of
> > > > the
> > > > > > mirrors), and allows easy access to the focusing and film advance
> > > > mechanisms.
> > > > > > Matt
> > > > > >
> > > > > > Bill Glickman wrote:
> > > > > >
> > > > > > >       I am trying to develop a MF camera system that allows one
> to
> > > set
> > > > > > > interocular bases at any seperation without physical
> > > > limitations.....As we
> > > > > > > all know, this is physcially impossible sometimes with two non-
> > > shift
> > > > > > > cameras.  My idea was to use two cameras side by side, but each
> one
> > > > would
> > > > > > > have the capability to shift the lens horizontally.   This has a
> > > very
> > > > > > > dramatic effect vs. spacing the cameras.  For example, I can
> > > simulate
> > > > 24"
> > > > > > > interocular distance with only 3mm of front shift on one camera.
> I
> > > > have
> > > > > > > tested this, it works.  So a small amount of lens shift would
> > > simulate
> > > > > > > interocular spacing fro 0 to 500 + ft.
> > > > > > >
> > > > > > >       I have found that myself and my audience all prefer very
> small
> > > > OFD's,
> > > > > > > say 1.3mm, hence the need for bases much tigher than any two
> cameras
> > > > can
> > > > > > > every physically acheive.   A fixed stereo camera (sputnicks)
> don't
> > > > appeal
> > > > > > > to me because I want the flexibility of adjusting the
> interocular
> > > > distance
> > > > > > > when required.
> > > > > > >
> > > > > > >         Although this sounds good in theory, I am curious if
> this
> > > will
> > > > > > > produce the same stereo effect as utilizing two cameras at the
> > > proper
> > > > > > > interocular distance.  I don't plan to use it for excessive
> bases,
> > > > only for
> > > > > > > 24" and less.  Has anyone every tried this before?  Any input?
> > > > > > >
> > > > > > > Bill G
> > > > > >
> > > > > > --
> > > > > > ______ Matthew V. Ellsworth ______________________________
> > > > > >       oakridge@xxxxxxxxxxxxx      http://www.oak-ridge.com
> > > > >
> > > >
> > > >
> >
> >