a short course in image scanning.. V2
this was mostly culled from emails and other communications with clients over
the last few years, folks new to imaging, and particularly image scanning will
find some useful information here..
okay, you've got a photo in your hand, but what you really want is to have that
that image "on my mate's computer", or "on my web page", or *in my ftp folder*.
The purpose of this short course is to guide you through the steps required to
achieve that perfect scan.
First we need a scanner. This is a very simple device which simply passes a
photosensitive head over the image and converts what it sees into a stream of
..the computer reads this data using the "TWAIN" interface (usually) and
converts it into a viewable image. If it looks good, we can then save it to
disk, copy it to the clipboard, whatever we want now that it's "in the
As a rule I always save my scans immediately. I use a wintel box for these kinds
of jobs and crashing is always a possibility. USB scanners tend to be less
problematic than their parallel counterparts, and SCSI devices (and now
firewire) are the best of all. As a general rule, the faster the better,
considering that some images can be very large indeed..
it's feckin HUGE!.. many people get a shock when they scan their first image.
Knowing nothing of DPI, screen reselutions and other minor details, they hit
"SCAN" and BANG! something the size of 16 monitors appears, or at least a small
part of it appears. This isn't such a bad thing. As Ansel Adams said, "first you
get the perfect negative".
So long as your computer can take the pace (has sufficient memory and CPU power)
this is a good strategy. Scan big, then reduce on the computer. That way you can
keep the image quality high. Maybe you need to match the input resolution with
the output resolution of your printer. For some uses, a quick low-resolution
scan is all that may be needed.
At some point we're going to have to know about all this DPI stuff so that we
can make these kinds of informed decisions.
DPI = Dots Per Inch, a measurement of "resolution"
The bigger the DPI, the more dots (pixels) per inch and therefore, the bigger
the picture. A good standard setting (and usually the default setting of an
out-of-the-box scanner) is 300dpi. This is also the default for most printers,
so if you scan an image and print it directly (without altering it in the
middle) it should be the same size as the original.
to intelligently decide what resolution to use, all we need to know is one
WHAT IS THE IMAGE FOR? That's all that matters. We need to know the "target
resolution", if you like. If the image is going on a web page, then a 72 dpi
scan would probably be fine (though personally I'd use at least 150 dpi and
reduce it later with Photoshop). If we are scanning artwork that is to be
printed in a booklet, for instance, we would probably need to scan at 600 dpi,
1200, or even higher.
Here are some typical resolutions.
Computer screen: 72 (or 96) dpi standard home printer: 300 - 600
dpi high quality printer: 800 - 1600 dpi
This explains why a 300 dpi scan looks so huge on the screen. If your screen is
800x600, that's 800 pixels (dots) wide. At 800 dpi a one inch scan would fill
your entire screen!
The exception to this might be where you are scanning for screen resolution (72
dpi), but have text in the original image that you want to be legible. In this
case I would definitely scan bigger and reduce afterwards on the computer.
The big "interpolated resolution" con.. NEVER scan an image at a dpi higher than
your scanner's TRUE optical resolution. Simple. Many scanner manufacturers make
wild claims about "interpolated resolutions" like 9400 dpi or 24000 dpi, but in
reality the scanner is just rescaling the image itself. It might be argued that
the scanner, being closer to the original, should be able to make a better job
of this than some image editing software, but I've yet to see this. Most
scanners are around 600dpi X 300dpi, check your manual.
The big HOW-TO..
1. Place the image face down on the scanner bed. Close the lid.
2. Fire up your favourite image software. For quick scans I use Irfan, or PSP7
if I have a batch to to. Any software that supports the TWAIN interface will be
able to capture scan and make sense of it. Often TWAIN capability is built into
word- processors, OCR packages, DTP software, so check that first, then you can
miss out the middle-man, so to speak.
3. Chose "acquire" from the menu.. it's usually under the file menu somewhere,
maybe in a twain sub-menu. in Irfan it's CTRL-A. The scanner TWAIN application
will appear. If this is your first scan there will likely be a short delay while
the scanner warms up. If these seems archaic, the he reason is that the scan
head needs to reach a certain "colour temperature" before it makes the scan.
If the scanner app didn't appear, you probably haven't told the computer what
your preferred device is. Just chose "select Twain source" or something along
those lines from the menu, and chose your scanner from the list.
4. Do a Preview scan. this allows you to see what the scanner will see. It's
just a quick, very low resolution scan. Now you can set the cropping, colour,
brightness, etc. usually leaving everything "in the middle" is the best
strategy, altering such things with your imaging software.
5. Click "SCAN" if all goes well, the scanner will burst into life and begin
the scan operation. after a wee while (depending on how fast your scanner, and
it's interface is) In a few moments your image will appear in you imaging
software's window as a new file.
6. Chose "save" or "save as" from the file menu. Give your image a meaningful
name, and save it. You may want to work some more on the image, rotating,
scaling and altering it, etc, but now you can always go back to the saved image
if you make any mistakes.
** I'm going to butt-in here with a quick note about image formats..
It's still REALLY feckin HUGE!! You may have noticed that your saved image is
still enormous. If you intend to send this over any networks (like the internet
for instance) obviously this size thing is going to be a problem.
We need to somehow compress the file. I guess we could zip it (or RAR it) but
that probably wouldn't make much of a dent in the file-size; these types of
compression schemes just weren't designed for pictures (although the multimedia
scheme of WinRAR does a commendable job) Fortunately there is a much better way.
bitmaps and un-bitmaps.. When you save the image in your imaging software
(Irfan, Photoshop, PSP, etc) you will be given a choice of "image formats".
Entire volumes have been written about image storage schemes, but with the topic
firmly in hand I'll say there are basically two types. The first and most basic
is a bitmap.
Bitmap images are just that, a map of all the bits, or dots (pixels) of the
image. A bitmap is an exact record of the exact value of each and every pixel of
the entire image. Bitmaps can get very large indeed. Some common bitmap formats
are: BMP, TIFF, PICT and XBM. Most bitmap formats have introduced compression
schemes of some sort to try and save space, and some do a fair job too. if you
need an "exact" reproduction than you must use a bitmap format.
"smart" image formats.. Most people don't *need* bitmaps. The kind of person who
needs to use bitmaps probably wouldn't need to read this (unless it was just for
fun). The ubiquitous JPEG is just fine for most images, because most images are
*photographic*. Jpeg compression is specifically designed for photographic
images and the compression algorhythm is designed around what the human eye will
see. It's very clever indeed, and if you're interested in the nuts and bolts of
the format, the jpeg group produce an excellent FAQ on the subject.
Jpeg compression can be ridiculously high. I've seen 10MB files reduced down to
50k!! But, as always, there is a trade-off. Here it's a question of quality.
Simply put, the more you compress an image, the more you degrade it. A JPEG with
a 50% compression will be *tiny* compared to the original, but will be
Small text will be probably be illegible, a "blocky" effect will be in evidence.
But even at 10% compression (90% quality setting) the file will be significantly
smaller and an expert would probably find it hard to tell the difference.
Usually a setting of 75-85% will do just fine, sliding up and down depending on
the target audience.
There are many other compression image file-types too, the most famous being the
GIF, which is a proprietary format, owned by CompuServe.
JPEG is a lossy format. This is important. You must remember that every time an
image is saved in JPEG format, some quality is lost. If you save a *perfect*
image as an 80% jpeg, it will still look fine. But if the resultant image is
opened, and then saved again (maybe for a crop or something) then *more* quality
is lost. Do this ten times and the image will be tatties! A bitmap would, of
course, be perfect every time. Ideally Jpegs are saved only once, for transit,
from the original bitmaps.
*very* smart image formats.. There are some newer formats that seem to achieve
the best of both worlds, like the Lura-Wave format with extremely high
"lossless" compression, but these are mostly proprietary formats, and licensing
issues will always cause distribution problems for some time to come. So, to the
THE MIGHTY PNG! This is a modern, FREE format, which achieves superb compression
ratios with lossless quality, handles greyscale, full colour, transparencies and
much more. It works better on some images than others and a little
experimentation is required to find the best settings for the kinds of images
that you work with. It works extremely well with non-photo data, like desktop
screenshots, logos, graphics, etc. this is now my most common used image format.
As a rule of the thumb.. for photographic scans, save as a JPEG at about 85-90%
quality (10-15% compression) unless you have a good reason not to.
One final note about jpegs: HUGE images don't compress as well as smaller ones.
(it's illusion really, or rather "perception", so therefore, quite real!) Also
highly detailed images can foil the JPEG algorhythms. Allow me to illustrate..
I have an image 2048 X 1024 pixels, 24bit colour. It's a desktop screenshot of
my system, pasted 6 times over. It's huge. As an uncompressed bitmap image it
takes up 8MB of space.
As an 80% jpeg it takes up 445k. The jpeg is a little fuzzy around the text and
borders though still legible. Messy, but readable.
Saved as a mighty png it takes up 265k!! that's a mere 1/30th of the size! and,
the png is identical to the original! With desktop screenshots and similar
detail the PNG is obviously the format of choice being small and also perfect.
A photograph, on the other hand, would tell a very different story. Here the
jpeg would almost certainly be smaller. A quick 1024x768 experiment produced a
jpeg of 198k, compared to a PNG of over 1MB. They look the same. For
photographs, the jpeg is king.
Quite often the imaging software will have a jpeg optimiser which allows you to
set the SIZE you want and let the software work out what compression you will
need to hit your target. For peecee users, PSP (PaintShop Pro) has one built in,
and Photoshop (most platforms) has a number of third-party plug-ins that do a
The built in PSP optimiser very usable, gives a readout of the projected file
size and also a preview of the resulting image live, as you move the compression
slider up and down. It will even calculate the download times for you! Superb.
ImageReady (which comes with photoshop) does all these things too; and as psp
isn't available for mac, my weapon of choice for working with images for web.
well, that's it! hope this help you take better scans.