To properly use stereo viewing methods, this page requires that
JavaScript, cookies
and popup windows are enabled.
Home
Up
Search
Mail
NEW
Determining stereo base
Browser should support JavaScript:
Lets take the following example:
In the below my starting point is how we experience (target) things with our eyes (this is contrary to
normal
practice in stereo world, but I just find my eyes too important). So even the eye diameter is playing a role, but don't despair it is canceled out further in the process;-).
So bear with me, even thought it start with the human experience, I will come back to the technical stereo base.
I want to make get a stereo pair that is
experienced (targeted) like
a real scene that has an farthest object as infinite (~6 [km]) and a nearest object of a certain distance.
Parallax angle can vary considerable say between 1° and 4°:
a parallax angle smaller then 1° (or LAOFD), gives a flatish 3D image
a parallax angle larger then 4° (or HAOFD), gives a quite uncomfortable stereo view (more chance on eye strain).
A general accepted comfortable parallax (or OOFD) is around 1.88° (=2*atan (65.5/2000/2)) due to a nearest object of 2 meters and a farthest object at infinity.
From that assumption I want to determine how to make the stereo pair with a photo camera (in the below by
default a digital camera
).
I use parallel viewing method as the default in the below, in the below calculations
other stereo viewing methods
can be set.
I am aware of the
Stereo Calc of Takashi Sekitani
(this one assumes that Fc << focused distance), the
spreadsheets of Mike Davis
(this one assumes an MOFD= Fc/30) and the
JavaScript of Gerhard Herbig
.. These will gave comparable results as below calculations, when determining the
stereo base based on MOFD
.
Some guidelines and formulas can be seen here
.
I have the stereophographic guidelines also as functions (Add-in) for Excel available, so one can build one's own application around this Excel Add-in. If you want more information,
let me know (remove underscore in presented e-mail address)
!
Some
camera presets:
36*24 mm camera (F
c
=40 [mm], filmwidth=24 [mm])
1/2.7-inch
digital camera (F
c
=7.5 [mm], filmwidth=4 [mm])
1/2.3-inch
digital camera (F
c
=6.3 [mm],
effective filmwidth=6.16*0.88=5.42 [mm]
): e,g FinePix REAL 3D W1
Some stereo
viewing method presets:
Stereoscope (picturewidth=75, F
v
=135 [mm])
Slideviewer (picturewidth=23, F
v
=44 [mm])
Parallel when
mounted/displayed to infinite/far
(picturewidth=65.5, picturedistance=500 [mm])
Crossed (picturewidth=120, picturedistance=500 [mm])
Projected (picturewidth=1200, picturedistance=2000 [mm])
Anaglyph/hor. or ver. interleaved/page flipping (picturewidth=170, picturedistance=500 [mm])
Lenticular (picturewidth=203, picturedistance=325 [mm])
Other presets:
Orthostereo (make nearestobject
exp=
nearestobject
act
and farthestobject
exp=
farthestobject
act
)
Point acuity
=
[']
Photographic stereo acuity
=
[']
Allowed scale difference due to convergence
=
[%]
Here are my calculations to get from wanted parallax experience (p
exp
) towards the stereo base (b0).
Fields near an
italic and
red
name
can't be changed:
Parallax angle target is: p
exp
= 2*atan(eyeseparation/2/nearestobject
exp
)
(in this simplified formula: farthestobject
exp
= infinite)
Using:
eyeseparation =
[mm]
nearestobject
exp
=
[m] (This is for targeted experience, so no relation with the actual objects in the scene)
farthestobject
exp
=
[m] (This is for targeted experience, so no relation with the actual objects in the scene)
p
exp
=
°
MORD (Maximum On Retina Deviation
) becomes: MORD = tan(p
exp
)*eyediameterinair
Using: p
exp
=
°
eyediameterinair =
[mm] (
diameter eye in water is 22.22 [mm]
and in air this seems smaller due to reflection index: 1.333)
MORD
=
[mm] (see also the graph on
this page
).
MOPD (Maximum On Picture Deviation) is: MOPD = MORD*distancepicture/eyediameterinair
Using: MORD =
[mm]
distancepicture (or F
v
) =
[mm] (distance to picture or focal length viewer when using slideviewer/stereoscope)
eyediameterinair
=
[mm]
MOPD
=
[mm]
MOFD (Maximum On Film Deviation) becomes: MOFD = MOPD*filmwidth/picturewidth
Using: MOPD =
[mm]
picturewidth =
[mm] (width of a single picture,
not
the pair: max. width when using parallel viewing is
eyeseparation
:
[mm])
filmwidth =
[mm]
MOFD
=
[mm]
Stereo base
(b0) is: b0 = MOFD*(nearestobject
act
/F
c
-1/2)
(in this simplified formula: farthestobject
act
= infinite: case 2 of
Bercovitz
)
Using: MOFD =
[mm]
nearestobject
act
=
[m]
farthestobject
act
=
[m]
F
c
=
[mm]
b0
=
[mm]
diaphragm (f#)
>=
[-]
focus distance
=
[m]
non overlap on film (no convergence)
=
[%]
convergence distance
>=
[m]
If I make a stereo pair with a stereo base of 60 mm or 220 mm I get the following stereo pairs (nearestobject
act
at 2 m, farthestobject
act
at 6 m, F
c
= 6 mm):
60 mm stereo base
220 mm stereo base
The one with a stereo base of 220 mm is
certainly not
viewable comfortable with the parallel viewing method.
I made a another stereo pair (nearestobject
act
at 2 m, farthestobject
act
at 50 m, F
c
= 6 mm):
60 mm stereo base
240 mm stereo base
I want to get as much constructive feedback as possible on the above.
Let me know (remove underscore in presented e-mail address)
!
Change stereo viewing applet/plugin properties
You can change your screen properties, eye separation and applet/plugin
here
.
Disclaimer and Copyright
Home
Up
Search
Mail
Major content related changes: April 20, 2005