Some General interest
questions
What is a digital image?
A digital image is an
image that is at some
point converted into
digital form for
convenient manipulation
by a computer or other
digital device. The
images discussed in this
newsgroup are most
commonly of the type
known as "raster"
images, where each pixel
is represented
individually (although
they may be compressed
for storage, depending
on the file type).
There are a number of
different formats, but
the most common for
photographic purposes is
the 24-bit RGB image,
where each pixel
consists of a single
byte of information for
each of red, blue, and
green. This means that
each of the three colors
can have 256 individual
intensity shades, giving
16.7 million shades in
total. Other formats
include greyscale, CMYK
and LAB. Greyscale is
useful for storing black
and white images, but
the others are useful
only in certain
circumstances which are
beyond the scope of this
document.
What do people mean
when they talk about
resolution?
There is a common error
made regarding the term
"resolution"–indeed.
Properly speaking,
"resolution" should only
be used to discuss the
amount of detail present
or resolvable within a
certain distance. Thus,
valid measures of
resolution would include
dots per inch (dpi),
pixels per inch (ppi),
and lines per inch (lpi).
Notice the length term
present in each of
these. Without this
length term, it cannot
be denoted as a measure
of "resolution".
What is the
difference between dpi,
ppi, and lpi?
Dots per inch is
generally used to
quantify the output of
printers. In this usage,
it means that a given
number of dots of a
single color can be
printed within a certain
distance. Thus, the
rated resolution of a
printer is valid only in
line art mode. As soon
as you try to print
shades of grey or colour,
the effective resolution
of the output goes down
since you need to blend
dots of different
colors.
Pixels per inch is the
proper term to use when
specifying the
resolution of a scanner.
The ppi rating of a
scanner is the number of
pixels (each composed of
red, green, and blue
values) that can be
sampled within the space
of an inch. Thus, a film
scanner rated at 4000ppi
would be able to sample
6000 pixels along a
segment of film
measuring 1.5" long.
Lines per inch is the
normal way to test the
resolution of optics and
film in the analog
domain. A printed target
of black lines is used
to determine at which
point you can no longer
distinguish between the
lines. Since there is
white space between the
black lines, it actually
would take a line of
white and a line of
black to give a single
"line" of resolution.
Thus, a scanner capable
of 1200ppi resolution,
would be theoretically
capable of a maximum of
600lpi.
How do I get a
digital image into my
computer?
There are two ways to
get an image of a
real-world object into
the computer–a digital
camera or a scanner.
The digital camera uses
lenses to project an
image of the actual
object onto a
light-sensitive
charge-coupled device (CCD).
The CCD then converts
the image directly into
digital form, at which
point it may be stored
or sent to the computer.
Digital cameras may be
used anywhere that film
cameras are used, and
generally look similar
to film cameras.
The scanner takes an
analog image and
converts it to digital.
The original item is
placed in the scanner,
at which point a digital
image is sent to the
computer. Flatbed
scanners are useful for
scanning prints, while
film scanners are
generally higher
resolution and are
optimized for negatives
and slides. Scanners
require at least two
steps
(nature->film->digital)
as opposed to the single
step (nature->digital)
of the digital camera.
What is better for
digital images, a
scanner or a camera?
It really depends what
you want to do. At
present, digital cameras
in the price range even
remotely affordable by
the amateur are quite
limited in their pixel
size. Currently they are
in the range of
1600x1200, with some
going somewhat larger.
This limits your print
size to about 4"x5" at
near photographic
quality. It is possible
to do a pretty good
5"x7" and even a decent
8"x10", but it is
noticeably softer than
the 4"x5". If this range
of output size is
satisfactory, or if you
are only going to be
displaying the images on
a computer screen or LCD
projector, then a
digital camera should
suit you just fine.
If, on the other hand,
you want a larger print,
or want to be able to
crop and still print the
image at the above
sizes, then you'll need
to spring for a scanner.
In this case, dedicated
film scanners give the
best results, but
flatbed scans of prints
can give excellent
quality as well.
[ Top of the page ]
Scanners
How do scanners work?
Briefly, scanners shine
a light on (or through)
a subject, and then an
array of sensors picks
up the image and
converts it to digital
form.
What different kinds
of scanners are there?
There are two kinds of
scanners which we will
be considering, flatbeds
and film scanners.
Flatbed scanners are
designed for reflective
material, which means
that the light source
(generally a fluorescent
bulb) and the sensor
array are on the same
side of the original.
Flatbeds are generally
able to accept inputs of
up to letter or legal
size (8.5"x11" or
8.5"x14"). Larger
scanners do exist, at a
much greater price.
Consumer flatbed
scanners can have
resolutions up to
1200ppi, although some
professional (and
extremely expensive)
scanners have
resolutions that are
quite a bit higher.
Flatbed scanners are
capable of doing an
excellent job of
scanning prints,
magazines, artwork, and
even small objects.
Film scanners are
optimized for film
(including both
negatives and slides).
This means that they are
designed to handle
smaller originals (from
APS to medium format in
size), and they are
designed for non-opaque
originals. This means
that the light source
(either a fluorescent
tube or an array of LEDs)
is on one side of the
original, while the
sensor array is on the
other side.
Film scanners generally
have much higher
resolutions than
flatbeds, which is
possible mainly because
the originals are so
much smaller. Film
scanner resolutions in
the range 2400-2700dpi
are common, with the
newer scanners coming in
at 4000dpi. Most film
scanners are optimized
for 35mm negatives and
slides, but it is
possible to find ones
that support APS and
medium format as well.
How many pixels can I
get from a scanner?
Well, a high-end
consumer film scanner
has a resolution of
4000ppi. With a standard
35mm negative, this
would give a pixel size
of 3780x5670 pixels, or
a total of 21 millions
pixels.
Scanning from an 8x10
print using a flatbed at
600ppi, you would end up
with an image size of
4800x6000 pixels, for a
total of 28.8 million
pixels. In comparison, a
normal digital camera
will give up to about
2000x1312 (for the Nikon
D1) pixels, or a total
of 2,624 million pixels.
[ Top of the page ]
Digicams
How do digicams work?
In a digital camera, a
series of one or more
lenses is used to
project an image onto a
light-sensitive
charge-coupled device,
or CCD. This CCD detects
the image, and converts
it to electrical
signals, which are then
converted into a
standard image format
and either stored on the
camera or transferred to
a computer.
How many pixels can I
get from a digicam?
This varies with the
camera. Low-end ones
will give 640x480
pixels, or sometimes
even less. Current
state-of-the-art
consumer digital cameras
will give up to about
2000x1312 pixels, or a
total of 2,624 million
pixels.
In comparison, a
high-end consumer film
scanner has a resolution
of 4000ppi. With a
standard 35mm negative,
this would give a pixel
size of 3780x5670
pixels, or a total of 21
millions pixels.
How are the pictures
stored?
The camera can be
tethered directly to the
computer, in which case
the image is sent
directly to the printer.
More commonly, however,
the image is written to
some form of temporary
storage while in the
field.
The two most common
forms of temporary
storage are the floppy
disk or flash memory
capable of holding up to
128MB, with capacity
still increasing. Flash
memory comes in two main
forms, CompactFlash and
SmartMedia. Sony has
recently released a
third form, called the
MemoryStick.
At the current rate of
digital camera
development, we should
reach the 8.64 MP level
of consumer level
cameras in a year to a
year and a half. The
34.56 MP level should be
reached within three
years, assuming we've
got flash cards big
enough to hold these
images and the patience
to download them to our
PCs.
[ Top of the page ]
Compression
What is compression and
why would I want it?
Compression is when,
instead of saving the
actual information
contained in the image
file, you save
information about that
image information.
Imagine if you took a
picture of 2000x3000
pixels that was
completely white. You
could save the
uncompressed image (at a
file size of 18MB), or
you could save a file
that just said "there
are 6 million pixels
that are all white" (at
a file size of under
1KB). This is a crude
example of what image
compression does.
The reason for
compression is rather
obvious–which would you
rather download, the
18MB file or the 1KB
file? Now of course in
real life the difference
isn't quite that
striking, but it is
still quite large.
Is there anything bad
about compression?
When you receive the
image, you have to
decompress it. This
takes CPU time. If you
have a slower CPU but a
very fast internet
connection it can
actually be faster to
transfer the
uncompressed image
rather than compress the
image, transfer the
smaller file, and
decompress the image.
For storing files on
your own hard drive, it
is generally faster to
read in a larger
uncompressed image than
it is to read in a
smaller image and
decompress it. Aside
from this, most
compression schemes will
throw away some
information to compress
the image to a smaller
size. They are pretty
smart about what they
throw out (most of it
wouldn't be visible to a
person anyway), but they
are still throwing out
information. This is
known as lossy
compression, of which
the commonly used JPEG
is a form.
What is the
difference between lossy
and lossless
compression?
Lossy compression has
already been mentioned.
It tries to compress to
a smaller size by
throwing out information
that it feels you
probably wouldn't have
noticed anyways. An
example of this is JPEG.
As you compress the file
smaller and smaller, the
artifacts (the
differences between the
compressed image and the
original) caused by
throwing out this
information become more
and more noticeable.
Lossless compression
uses similar methods as
the popular .zip file to
compress the image
without throwing
information out. These
give better image
quality at the cost of a
larger file size.
Because they don't throw
out any information,
they can't compress as
small as the lossy
compression. Examples of
lossless compression are
the GIF, LZW-compressed
TIFF, and PNG image
formats.
[ Top of the page ]
What resolution is
required for
photographic quality?
The general consensus is
that 300dpi of effective
resolution is sufficient
for "photographic
quality". Of course, if
you peer closely at the
output you will be able
to determine that it is
not a true photograph
even at 300dpi, but that
close an inspection is
very rare. 300dpi gives
very good images that
the average person will
not be able to tell from
a chemical print.
How do image pixels
correspond to ink dots?
Early printers had a
direct correlation
between the input and
the output ink dots.
Current printers use
sophisticated algorithms
(error diffusion is one)
to convert the input
image pixels to output
ink dots. Because of
this, there is no longer
a direct relation from
one to the other.
Rather, it depends on
image content.
[ Top of the page ]
Colour Management
Why doesn't the image
on the screen match the
real object?
There are three reasons;
your scanner/camera,
your monitor, and your
software.
The scanner/camera will
not be perfect in
turning the image into
digital form. The human
eye has a lot of
adaptability that
electronic devices do
not have. This leads to
bad white balance,
incorrect exposure, and
other such difficulties.
High contrast scenes may
simply be impossible to
accurately sample with
consumer equipment.
If the digital image is
correct, your monitor
may be adjusted
incorrectly. Brightness
and contrast controls,
as well as the color
temperature setting of
your monitor, can make a
huge difference in what
the image on the screen
looks like.
Finally, if the digital
image is correct and
your monitor is set
properly, your software
may be messing things
up. Make sure that if
your software has color
management settings, you
are using them properly.
Improper settings will
cause problems.
Why doesn't the image
from the printer match
the image on the screen?
There are again three
possibilities; the
monitor, the software,
and the printer.
If the image doesn't
match the image on the
screen, but it does
match the original, then
you have a problem with
either your monitor or
your software.
If the screen image
matches the original,
then it is a problem
with either the software
or the printer. If the
software has color
management settings for
the printer, you need to
use them to adjust the
output until it matches
what you see on the
screen and the original
item.
Picture CD/Photo CD
What are the
differences between
Picture CD and Photo CD?
Although both Picture CD
and PHOTO CD Discs are
ways your service
provider can deliver
your images to you on a
Compact Disc, requesting
one CD or the other will
give you completely
different results.
In general, images
stored on a PHOTO CD
Disc are intended for
professional or
commercial use,
available in six levels
of resolution, ranging
from 128x192 to
2048x3072 pixels.
Images stored on a
Picture CD are intended
for the average
picture-taker; they have
one resolution,
excellent quality at
1024x1536 pixels. You
can store approximately
100 images on a PHOTO CD
Disc, and you can add to
the images on the CD
many times. Images are
written to Picture CD at
the time of the original
processing from a single
roll of film (number of
exposures will vary),
and you cannot add more
images to it later.
PHOTO CD Discs require
the use of enabled
software to view and use
the images. Picture CD
comes with software
included on the CD. |
