Secrets of direct digitizing

[Kenelm Philip]

	I have received no new information in response to my earlier
posting re digitizing, but I gather there is some interest in how this
may best be done. So here are some hints:

	Definition: Direct digitizing = making digital images without
going through the 'normal' procedure of making a photograph and then
scanning the slide or print (or making a PhotoCD).

	Methods: 1) flatbed scanners, 2) digital cameras. There is a
third method, using single frames from video cameras--I have nothing to
offer regarding that approach.

1)	Using flatbed scanners: These devices have sufficient depth of
field (which may vary among different makes and types) to achieve reason-
ably good focus on butterfly wings even though the body (of a spread
specimen) keeps the wings from contacting the platen. Before attempting
this on your scanner, you should calibrate the depth of field of field
by scanning a ruler with one end on the platen and the other end propped
up so the ruler is inclined to the platen. The image can then be examined,
and a simple calculation will show you what the limiting distance from the
platen is for acceptably sharp focus. When in doubt, make a print--the
resolution on the print may be higher than on the screen.

	In my case (H-P ScanJet 4C) the depth of field does not allow me
to scan pinned specimens--they must be depinned. A method for doing this
was described in the News of the Lepidopterists' Society (1 May 1975, page
1). I have constructed a more elaborate version as follows: Take a Variac
(autotransformer) and connect it to the 120V end of a 6.3V filament trans-
former capable of producing at least 5 amperes at 6.3V. Attach a piece of
zip cord to the low-voltage terminals of the filament transformer, and
to its free end attach two very small alligator clips. You may find that
having an ammeter in the low-voltage circuit can be a help. Attach the
alligator clips to the pin above and below the body of the specimen, and
slowly crank up the Variac. For black pins, you will have to scrape the
paint off to make better contact where the clips are placed.
	Depending on the size of the pin, you should find that something
around 2 to 4 amperes will heat the pin enough to soften the insect's
tissues enough to let you withdraw the pin. Warning: letting the pin get
red hot will produce a strong and unpleasant odor of roasting bug...
	
	You will then need to whomp up some kind of device to spear the
butterfly through the (somewhat enlarged) hole left by the pin, and hold
it just over the platen with the ability to adjust the height and angle
of the wings wrt the platen.

	Very fine images may be made this way. For a butterfly with a
two-inch wingspan, scanning at 600 dpi gives about 1200 pixels across the
image. Printing at 300 dpi on a color printer gives you an approximately
life-sized image of good quality.

	One problem, as Gary Anweiler reported today, is that the background
tends to run gray since it's usually farther from the platen than the
specimen. An ideal specimen-holder would thus have to let you have a back-
ground (of whatever color you wanted) immediately behind the specimen.
Gary was fortunate in that his scanner could handle a pinned specimen, which
mine does not have the depth of field to do--or else Gary will be pleasantly
surprised with the improvement if he tries to scan a depinned specimen.

2)	Using digital cameras: Here the important parameter is the macro
focusing ability of the digital camera. To date I have found three digital
cameras with true macro capability: the Ricoh RDC series, the Sony DKC-ID1
(about twice the cost of most consumer-grade units), and the new Sony
Mavica MVC-RD7. Warning: many makes of digital cameras _claim_ macro
capability, as the Olympus D200-L and D300L units, the Apple QuickTake
series, etc. Some (i.e. Olympus) even show a butterfly closeup in their
advertising brochure--which some experimentation with the camera, and
careful examination of the figure in the brochure, indicate was almost
certainly actually made with a conventional film camera. Most such units
can focus down to about 3.5 inches, which means, with their somewhat
wide angle lenses, that the horizontal extent of the field in the subject
plane at closest focus is around 5 inches--not at all sufficient for
butterfly shots.

	The three cameras listed above as having true macro capability
can all focus to within 1 cm. The Sony Mavica (which is the one I am
using) can cover a horizontal field of 1.25 inches at closest focus,
giving a reasonably large image of even a Lycaenid with a one-inch wing-
span. The main problem with these units is that their macro focusing
is only obtained when in wide-angle mode. The Mavica has a 10 to 1 zoom
lens (40mm to 400mm 35mm equivalent) but must be in its widest-angle
mode for close focusing. This means that providing proper illumination
of the subject is tricky, and will involve the construction of a special
lightbox to get diffuse lighting directed at the specimen into the narrow
space between the camera body and the specimen. I will know more about this
topic later this fall...

	The Mavica has a 640x480 pixel CCD, so you can get over 600 pixels
across the image of most specimens. Prints at 144dpi or 300dpi are ade-
quately sharp for publication purposes. The Mavica is unique (to date)
among digital cameras in storing its images on a 3.5" floppy disk (JPEG
files on an HD MS/DOS disk). These images may be opened in Photoshop on
either a PC or a Macintosh--no cables needed. You can get 15 to 20 full-
resolution images on a floppy.

	The Mavica has auto-exposure (which may be locked out just before
taking a picture) and either auto or manual focus, using a 2.5" LED color
viewscreen on the back of the camera to check the image. The auto-exposure
may be offset plus/minus 1.5EV, which allows you to compensate for the
effects of a light or dark background.

	Preliminary results with the Mavica look encouraging, but a final
evaluation will have to await construction of the lighting source, and
trial shots of a number of different species of leps.

	Of course, the ultimate digital camera would be one of the Nikon
(or other make) units which use a high pixel-count CCD inserted in a
standard 35mm body. With prices running from $6K to $16K or so, these are
not an option for 'the rest of us'! Color scanners, and consumer-grade
digital cameras, are at least items that many of us can hope to purchase.

	I hope these notes prove helpful to someone contemplating direct
digitizing of leps. I will be most interested to hear about results...

							Ken Philip
fnkwp at aurora.alaska.edu