Re: Close-up 3d of plants & flowers

[P3D John Bercovitz <bercov@xxxxxxxxxxx>]

Hans Gelton writes:

> Does anyone in the group have experience in taking close-ups 
> say at 12 to 18 inches (300-450mm) of flowers and plants.

I don't have experience with flowers quite that close but I do 
have with other objects.

> My questions about this are as follows:
> (a) What is the most economical method of doing this.

I would vote for a slide bar and an SLR because it's cheap and 
you have the most control.  This method means you can't shoot 
flowers out of doors in the wind unless you can set up a tent.  
(Hey, a tent of thin white sheet might give interesting 
lighting... but then you might need some fill from the strobe 
for better modeling... which is also useful in 3D.)

> (b) Most stereo pairs have a separation of 2.5 inches (6cm), 
> how much more or less should this distance be changed.

How are you at maths?  8-)  The answer depends on the separation 
you use, the focal length of the camera lens, and the depth in the 
scene, that is, the distance from the front of your lens to the 
nearest part of the scene and to the farthest part of the scene.  
If you can restrict the depth sufficiently, then you can use a 
full separation of 2.5" or thereabouts.  This will be desirable 
if you wish the flower to look to be full size.  Increasing 
separation above the nominal results in the reconstructed image 
appearing smaller than normal and vice versa.  So if you use a 
very small stereobase, say 10 mm, the flower will appear roughly 
six times normal size.  I'm speaking here of the geometry that's 
presented to the eye.  Most folks' brains will override the 
geometry presented to them under certain conditions and come up 
with something closer to actual size.

In solving this problem, usually one will start with the focal 
length of the camera lens.  Here I would heed Ferwerda's advice 
to keep the lens fairly long, say 70 mm, if the slides are for 
projection.  If they are for a hand-held viewer, the camera lens 
focal length should match the viewer lens focal length which will 
be about 50 mm, more or less.  

When we speak of the camera lens, we speak of it as it is used.  
So if you use a 100 mm lens and focus on something that is only 
300 mm (12") away, then the lens has to racked out a couple of 
inches and its actual in-use focal length in this case would be 
150 mm.  A 50 mm lens could therefore easily be a 60 mm lens in a 
close-up situation.

As a professional stereographer name of Bob Mannle told this list 
a long time ago, the criterion you want to use is maximum on-film 
deviation.  The two transparencies which make up a stereo pair are 
not exactly the same (or we wouldn't get any stereo effect).  If 
you lay one transparency on a light box and lay the other on top 
of it, you can see the discrepancies.  Maximum on-film deviation 
is what it sounds like - if you shift the transparencies so that 
the closest object in the scene coincides in the two 
transparencies, then the distance from one representation of the 
farthest object in the scene to the other is the maximum on-film 
deviation.  You want to keep maximum on-film deviation to 1/30th 
of the focal length of the lens or less.  So if a 2.5" separation 
doesn't cause an on-film deviation in excess of 1/30th of the 
focal length of the lens, you'll have a viewable result.  

You can always take a stereo pair and measure the on-film 
deviation on the light table and see if the result is going to be 
easily viewed.  If you want to predict the on-film deviation in 
advance, you'll need to do some calculations, and just how that is 
done is the subject of an upcoming article in the ISU journal so 
if you haven't joined ISU yet, _now_ is the time.  8-)

Rather than make this overlong article even longer by telling you 
how to do the calculations, I'll do some sample calculations and 
give you tables for an output.  If you want to do the calculations 
yourself, get a spreadsheet from:
ftp://bobcat.etsu.edu/pub/photo/photo-3d/technical/maofd/
I'll assume you're using a 50 mm lens on your SLR.  "max 
stereobase" is the maximum stereobase you can use and still stay 
under the 1/30 focal length limit for viewability.  This doesn't 
mean you can't use a smaller stereobase; it just means that if you 
exceed this stereobase, the viewing might be uncomfortable for 
some people under some viewing conditions.

Near point     Far point       best focus     max stereobase
 300 mm          450 mm          360 mm          26 mm
 350             500             412             34
 400             550             463             44
 450             600             514             54
 500             650             565             66

Now if you have a total scene depth of 150 mm (6"), you can see 
that if the near point is greater than 500 mm (20"), you can use 
the full 65 mm stereobase.

Let's suppose the depth in the scene is a bit larger (600 mm).  As 
you'd expect, this means you have to back off more or reduce the 
stereobase more, either one:

Near point     Far point       best focus    max stereobase
 300             900            450             13
 350             950            512             17
 400            1000            571             20
 450            1050            630             24
 550            1150            744             33
 650            1250            855             43
 750            1350            964             53
 850            1450           1072             66

Now you can use the full stereobase of 65 mm if you stay back at 
least 850 mm (33").

If you continue like this until the flowers stretch clear back to 
infinity, then the maximum stereobase becomes approximately 1/30 
of the distance to the nearest object in the scene, a _severe_ 
reduction in the maximum allowable stereobase:

Near point     Far point       best focus    max stereobase
 300           infinity         600              9
 350             950            700             11
 400            1000            800             12
 450            1050            900             14
 550            1150           1100             17
 650            1250           1300             21
 750            1350           1500             24
 850            1450           1700             28

> (c) If two cameras were used at a separation of 2.5 inches, 
> would both be angled inwards towards the subject to keep the 
> object centred in both viewfinders?

No.  This causes a particular type of warpage in the reconstructed 
image which you really don't want, though many people won't notice 
it unless it's really bad.  The preferred method is to crop the 
transparencies to center the images.

For details on warpage, see Andrew Woods' paper on image 
distortions which you can read at:
http://info.curtin.edu.au/~iwoodsa/spie93pa.html

or download in Adobe Acrobat from:
ftp://bobcat.etsu.edu/pub/photo/photo-3d/technical/ 
where the title is woods93.pdf

For an easier read, purchase the bible of 3D, called "The World 
of 3-D", by Jacobus G. Ferwerda, from:
http://www.stereoscopy.com/3d-books/photo-books.html#world

> (d) How successful would a close-up lens with a Pentax beam 
> splitter or similar work?

No solid thoughts on this but my opinion is that you'd do better 
with the slide bar.

John B


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