Digital Reformatting of
Photographic Formats
-OverviewAaron Choate
Digital Library Production Services
The University of Texas Libraries
Photographic Format: Examples
Photographic Processes



Prints and Positives
(Albumen prints - Tintypes)
Photographic Negatives
(paper, glass and film)
Photographic Transparencies
(glass and film)
3x4
House typical to the Latin-American section of
town. Corpus Christi, Texas. -- Russell Lee
The Center for American History
35 mm negative
Students in the auto repair and painting
class at Del Mar College. Corpus
Christi, Texas. -- Russell Lee
The Center for American History
35 mm slide
Marian Davis Slide Collection
School of Architecture Visual Resources Collection
Mounts



Glass
Paper
Plastic
Benchmarking for Conversion


(pages 27 – 60 of Moving Theory into
Practice)
Even after the issue has been identified
and studied, the problem of
benchmarking digital imaging systems
remains open.
Physical properties



What is it about the original that you are
trying to capture?
In many cases it is possible to capture
the grains of the emulsion on the film.
Is this necessary for the project to be
successful?
Visual Perception



Subjective comparison between a digital
print and an otherwise reformatted object
may lead you to decide that you only
need the digital version to be legible.
How much is enough… how much is too
much.
What is enough for the items that you
are working with?
Physical Type, size and
presentation



What equipment will you need to use to
fit the object?
How large a file will you be dealing with?
Can these items be bulk processed?
Physical condition





Fragile Glass
Brittle/yellowing Paper
Scratches on negatives
Dust on slides
Fading color
Document Classification

Photographs are continuous tone
material…
• Tone and color reproduction become just as
•
•
important or more so than detail
Always scan at multiple bits
Bit depths greater than or equal to 8 per
channel are suggested by most imaging
guidelines
Medium and support


How does the glass, paper or film affect
the representation of the item in digital
form.
Do you want to capture or minimize the
support’s physical properties.
Tonal Representation



Dynamic Range –
the range of tonal difference between the
lightest light and the darkest dark.
Range from 0 – 4 with the lightest
portion being potentially 10,000 brighter
than the darkest portion
The higher the dynamic range, the more
potential shades can be represented.
Typical Dynamic Ranges for
source documents







Newsprint -- .9
Printed material – 1.5
Coated stock – 1.5 – 1.9
Photographic prints – 1.4 – 2.0
Negative films – 2.8
Commercial color slides – 2.8 – 3.0
High grade transparencies – 3.0 – 4.0
Key type



High key
image’s detail is concentrated in the
highlights of the document
Low Key
detail is in the shadows
Balanced
detail is evenly distributed
Ah… but it can’t be that simple


Scanner data on dynamic range isn’t
always reliable
Bit-depth only loosely correlates with
tonal resolution because the upper bits
of most scanning devices carry a
declining percentage of reliable
information
Tonal reproduction ==



Scanning bit depth
Signal-to-noise ratio
Final bit depth output
Tonal accuracy


You have determined that your scanner
can reproduce the tonal range, but will it
map them properly?
Using a grayscale target will allow you to
anchor your “aimpoints” to a source with
known digital values
Color Appearance




Is color reproduction necessary to the
document’s meaning?
What is the nature of the color?
What purpose does the color serve?
How important is maintaining the color
appearance?
Color capture options
Omar Pasha
Helmolt, H.F., ed. History of the World. New York: Dodd, Mead and Company, 1902.
color capture options
Bitonal



One bit per pixel
Best suited for high-contrast documents
(printed text)
color capture options
Grayscale


Multiple bits per pixel (shades of gray)
Best suited for continuous tone (B&W)
documents such as photographs
color capture options
Color



Also multiple bits per pixel (in color…)
Best suited for continuous tone items
with color information
(RGB? CMYK?)
Bit Depth



1 bit
2 colors (usually black and white)
8 bit
256 colors/shades of gray
24 bit
16 million colors/shades of gray
Measuring color


GretagMacbeth – color evaluation
system that could be adapted to digital
imaging programs.
Most reports on digitization of cultural
resources confirms the difficulty and
frustration of capturing and presenting
color that approximates the original
So what can we do color-wise




Calibrate your scanning device to a color
standard
Calibrate the rest of the imaging system
to the scanner
Control your lighting environment (same
levels for production and evaluation)
Scan grayscale and color targets with
each image or production batch
But wait there’s more!





Stay close to the scanning device’s capabilities
and try to make only minor corrections to the
master images
Consider capturing at greater than 8
bits/channel
Save the image in RGB with International
Color Consortium profiles
Save the image as a TIFF and preserve the
color profile info
Maintain scanning-related metadata
Resolution


No perfect resolution for all types of
material
Considerations when choosing
resolution for your project
Detail


Measure the finest scale of line that
should be represented in the digital
image. (telephone line, individual hair)
Film formats should be measured using
magnification
Detail as stroke



width of the finest line, stroke of dot, or
marking that must be captured in the
digital surrogate
The finest feature should be covered by
at least 2 pixels for it to be properly
represented
A feature at 1/100 inch would require
200 dpi to be properly depicted
Detail as Scale



Dominant scale –
the distance covered by the most significant
structural feature
(ie: knots in a carpet)
How granular do you get? Can we capture the
film grain? Do we want to?
Some institutions base their decision on what it
would take to print an 8”x10” from the digital
file
Detail and Visual Perception


Adding dpi and bit-depth increases the
amount of accuracy to a point, but at
some point you reach a level of
diminishing returns.
400 – 600 dpi for most visible structures
in printed text or images should be fine.
(slides and transparencies will require
higher resolutions)
Testing the scanners performance

Modulation Transfer Function (MTF)
• Potentially correlates known details
•
(frequencies) and how well the system
detects the details.
There has been a great deal of work on this
and a product called Imatest has incorporated
it into the tests it will run on your camera
Equipment - examples
equipment
Flatbed Scanners
Epson 10000XL – Graphic Artist
 2400 x 4800 dpi
 48-bit color
 3.8 Dmax
~$2700
equipment
Flatbed Scanners

Creo – Eversmart
•
•
5600 dpi optical
16 bit color
$30,000 – $40,000
All formats
equipment
Film Scanners

Nikon – Coolscan 5000 ED
•
•
4000 dpi optical resolution
16-bit A/D converter
$1,100
35 mm slides and negatives
Film strip accessories
equipment
Film Scanners

Nikon Coolscan
9000 ED
•
•
4000 dpi true optical
resolution
16-bit A/D converter
$2000
medium format positive
and negative
transparencies
equipment
Overhead Scanner Systems

I2S Digibook
•
•
•
10000 pixel linear
array
RGB
200 – 800 dpi (and
higher)
$70,000
All formats (with optional
light box attachment)
equipment
Drum scanners

ICG 380 Drum Scanner
•
•
12,000 dpi
CMYK-on-the-fly at drum
speeds of up to 1800 rpm
$50,000
equipment
“Virtual” Drum Scanner

Imacon Flextight 949
•
•
•
3 8000 pixel CCDs
80 – 8000 dpi
(non-interpolated)
24, 36 & 48 bit color
$20,000
Imaging Guidelines

Image Files
• Archival Master
• Derivatives
• Access
• Thumbnail
• Publication Master
imaging guidelines
Master Image





As large as possible
Uncompressed
Unedited
Serves as long term source for
derivatives
Usually stored as a TIFF
imaging guidelines
What can we do to the master?
Generally accepted enhancements
 Reduction of greater than 8 bit/channel
linear data to 8-bit nonlinear data
 Contrast stretching
 Minimal adjustment for color and tone
 Descreening/rescreening of halftone
imaging guidelines
What can we do to the master?
Debatable image processing techniques
 Image sharpening
 Despeckling for bitonal images
 Deskewing
 Software controlled color/tonal enhancements
 Application of color management profiles
 Conversion to CMYK or sRGB color
imaging guidelines
Access Image




Generally fits within the viewing area of a
computer monitor
(1024 x 768)
Files sized for fast download
Compressed for access speed
Usually stored in JPEG format
imaging guidelines
Thumbnail Image


Very small image
(150 pixels on the longest dimension)
Usually stored in GIF or JPEG
imaging guidelines
Publication Master



Manipulated image that is suitable for
publication
High Resolution
Often stored in compressed TIFF or high
quality JPEG form
File formats




Uncompressed TIFF
JPEG
GIF
JPEG2000
(A note about LZW compression)
software
Photoshop


(and scanner manufacturer plug-ins)
or photoshop elements
software
Vuescan
software
Book Restorer
So then we scan…