Monster Media 1994 #2/Monster Media No. 2 (Monster Media)(1994).ISO/freq_qa/dtpfaq.zip/DTPFAQ.EXE/SCANTIPS.FAQ

Text File | 1993-01-28 | 39KB | 782 lines

-------------         PAGEMAKER/QUARK LISTSERV               -----------------
-------------   S C A N N I N G    F A Q    UPDATE 1/22/93   -----------------

 Special Thanks...

 To Geof Peters, Monte Olsen, Cindy Stone, Gary Hartling, Jeff Needleman,
 and the guy who ask for a explanation of the meaning of life that gave
 me a deluge of new material to use in updating this FAQ.  Many of you will
 note your remarks incorporated into this FAQ without the slightest regard
 for crediting the source.

 Also my thanks to Pete Fink, the president of Peter Fink Communications (the
 guy who helped Adobe create Adobe Accurate Screening technology), and
 publisher of "Desktop to Press" newsletter for allowing me to use excerpts
 from his resource known as the "How to Make Sure What You See Is What You
 Get" booklet.

 We like it so much that we give it to all of our Service Bureau clients.

 Much of this FAQ is drawn from excepts of this resource, and I would like
 to give him full credit for allowing me to use his booklet as a springboard
 for this FAQ, for reasons of consistency, readability, and because I am
 plugging his booklet so thoroughly his office deemed it unnecessary to foot-
 note all occurrences of information taken directly from his work.  I recom-
 mend you contact him for a complete copy:

     Peter Fink Communications, Inc.
     120 Q Street NE
     Washington, DC  200002
     [phone]

 Get the booklet ($10.95 US plus shipping and handling, less in quantity),
 it goes on to cover Halftones and Dot Gain, Optimum LPI, Fonts, Cropping
 and Scaling, Masking and Flatness, Blends and Banding, and more all in a
 24 page booklet you can keep by your computer.

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

 Scanning artwork can be broken down into four primary catagories:

    1) Line Art
    2) Halftones
    3) Greyscale
    4) Color

    Each of these catagories have their own unique problems when scanning,
 before I continue with some techniques for greyscale scanning, I'd like
 to define and discuss these catagories.



 Line Art Scanning
 -----------------

    Line Art scanning is best defined as simply transferring pen and ink
 style drawings, clip art, and some pencil sketches into the computer
 system.  Line Art has no shading nor color other than black.  Line Art
 scanning is considered 1-bit scanning, that is the computer sees the
 image as either black or white.

    Line Art scans, since they can only be black or white, do not have the
 luxury of anti-aliasing (a technique that makes smooth transitions between
 black and white by adding varying steps of grey), and therefore are prone
 to having jagged edges around curves when scanned at lower resolutions
 than the final output device.  Therefore, you should not use Line Art style
 scanning to scan photographs or textured art unless you are attempting to
 achieve some special (read - often weird) effects.

    Line Art should be scanned at as high a resolution as the final output
 device to eliminate the "jaggie" effect. In other words if you have original
 art that is 5x7 and you're going to output at 5x7 on a 1200 dpi imagesetter,
 ideally it should be scanned at 1200dpi to achieve the best possible image.
 Note that if your original is larger than the final size, you should adjust
 your original scan resolution by the same ratio so as to not have more
 resolution than necessary.

    On a personal note here, while the goal of matching original scan
 resolution with final output resolution is desirable, very few of
 your clients will be able to discern any difference between 600dpi scans
 and 1200dpi scans, especially when your final output will be printed on
 glossy paper stock.  However, the difference in imaging time to your printer
 or imagesetter will be quite remarkable since you are dealing with 4 times
 as much data in a 1200dpi scan.

    Geoff Peters adds, One last thing worth mentioning here is tracing.
 The majority of line art that he scans are simple cartoon drawings or doodles.
 He then uses Corel Trace to convert it to postscript and thus enables him to
 increase the image to any size with out the fear of the Jaggies.



 Halftone Scanning
 -----------------

    Halftone scanning evolved because someone discovered that they could
 scan photographs from magazines.  While I won't get into the legal issues
 here, suffice it to say that there will be a legal problem with using any
 pre-published images.  Assuming that you have the permission to use a
 pre-printed piece, don't -- ask for the original photograph instead.

    Halftone scanning is essentially treated like a greyscale scan by most
 software and the conventional wisdom says to scan the image at a
 resolution lower than the original in order to reduce the threat of
 moires.  Well, about all you can do is reduce the number of moire patterns
 that you will have, not reduce the threat.  If you scan a halftone, you
 will have moires.  Without  getting too technical, the reason for this is
 that all halftones generated by a press have "irrational" screens and
 inconsistent angles applied.  Irrational screens and inconsistent angles
 are used to increase the quality of the printed piece for the type of
 press being used, they are called irrational and inconsistent because in
 the case of the press, they are adjusted mechanically by the press
 operator (If they were generated by a high-end four-color system--Scitex,
 Chromacom, etc.--the irrational screens were computed elaborately).  To
 properly eliminate potential moires your scanner would have to adjust its
 resolution, tracking (pickup alignment), and linearity (the range of
 grey scales) dynamically across the entire image.

    Granted, this may not be a big deal if all you intend is to scan the
 cover of SI's swim-suit issue to embarrass your colleague (of course, I
 would never advocate this type of blatantly criminal activity), but
 assuming your goal is publication quality work, Halftone scanning should
 always be used as a last resort if at all.



 Halftone Scanning vs. Halftone Screening
 ----------------------------------------

    Halftone scanning should not be confused with halftone screening.  Sound
 confusing?  Halftone scanning is scanning an original from an already screen-
 ed source such as a magazine, coffee table book, newspaper, etc.  However,
 many scanners now come with plug-ins, or custom software that has an option
 labeled "Halftone."  Check your manual about this, but most of these options
 are for creating a custom halftone screen of the image being scanned not for
 scanning a pre-screened halftone.

    As an example, Microtek's Adobe PhotoShop plug-in has a halftone option
 that allows you to screen an image to a number of preset patterns as you scan
 the original.  This is important because Microtek, and most other scanner
 makers, have done their homework on their scanners and know how to generate
 a halftone screen that will best take advantage of their machine's capabili-
 ties with your chosen output device.  Another advantage is that once the scan
 is completed, you don't have a large grey scale file, but rather a simple
 1-bit black and white file that approximates the grey range of your selected
 output device and takes up only a fraction of the space that a grey scale
 would.

    The disadvantage to this, is that your screened scan is now appropriate
 for really only one output device.  For instance, If you perform a halftone
 scan with the Microtek plug-in set for a "5x5 60 dot #1" pattern (the recom-
 mended pattern for the Apple Laserwriter IIf photograde) it will look great
 when printed to the Laserwriter, but appear like a strange rippled newspaper
 photo when printed at 1200dpi on a Linotronic.

    Halftone screening a scan is not entirely without merit, however.  Many
 manufacturers will give you some screen patterns (such as horizontal or
 vertical lines) that can create some very artistic effects regardless of
 output device.


 Greyscale Scanning
 ------------------

    Greyscale scanning is most often used for original photographs, but can
 also be effective for maintaining the "soft" edge of some pencil sketches
 that use lots of shading.  Most good flatbed desktop greyscale scanners
 are now capable of scanning 256 shades of grey.  Don't be swayed by a
 manufacturers claim of 4096 shades of grey.  While this may be true, you
 won't be able to use them.  The current PostScript specification only
 allows for 254 halftone steps (the way a shade of grey is represented in
 print, 254 steps with pure white and pure black which makes 256 total).


    Principles

    Most of us scan at too high a resolution.  As we increase scan
 resolution, we capture too much detail--and the scan's file can balloon to
 outrageous size.  The imagesetter [ or laserprinter ] must render all of
 that data, using lots of costly RIP time [ RIP - Raster Image Processor,
 the specialized computer in your laserprinter or more expensive
 imagesetters.

    As you work with a scanned image, the computer constantly has to re-
 crunch the scan data.  Over the course of a project this time really adds
 up.  Scans with way too much data not only rob your time, they can even
 crash the RIP--in which case you don't even get a page.

    Take a few moments to calculate what you need rather than scanning at
 high detail "just to be on the safe side."  Capture just enough detail for
 what you need to do.

    Production Tips

 -   Do a little arithmetic before you scan.  Planning to reduce the image
     to 50%?  This doubles the resolution.  Reducing to 33% triples the
     resolution, and so on.

 -   For greyscales, a final resolution of 1.5 times the LPI works well in
     most cases.  For soft subjects you can get away with 1.3 times the
     LPI.

     (i.e. - Linotronic output at 1200 dpi, 133 lpi your scan resolution
              should be no greater than 200 dpi.  Final scan resolution
              means that you've calculated the change in spatial
              resolution when you size your original scan, as
              discussed in the above tip)

 -   Avoid data overkill.  Clean up scans in a retouching program before
     using them.  Adjust their size and resolution before placing them in
     a layout program.

     Crop the image in a package like PhotoShop or PhotoStyler as closely
     as possible to avoid having the RIP compute all that "white space."
     Avoid using PageMaker or other layout programs ability to "crop" a
     scan, this feature merely masks an area and does not decrease the
     data that the RIP must compute.

 -   Sharpening the image in a retouching program gives a better looking
     result than using a high scan resolution.  Make the on-screen image
     just a little sharper than you think is needed--it softens a bit in
     production.

     The CCD chip in your scanner that actually reads the greyscale data
     is linearily balanced, meaning that the transitions between grey
     areas tend to be minimized too much.  Sharpening algorithms in
     retouching programs like PhotoShop and PhotoStyler adjust for
     this.  Always run your programs "Sharpen" routine on a scanned
     photo to compensate for this.

     Another little known tip is to add "noise" to a scan to minimize
     the transitional steps between your grey scales.  For instance if
     you have created a blend in your illustration package that has only
     thirty steps, you can export the EPS file into your favourite image
     editor and add 1 or 2 units of random "noise" with the noise filter
     to take away hard edges and "cheat" your eye.  This technique works
     well for ultra-high and very low resolution scans too that either
     exceed the grey level of your imagesetter or produce very pixelated
     images.


    Common Mistakes

 -   Taking a big high resolution scan and shrinking it to tiny size--16
     Megabytes for a postage stamp--yow!  Reduce resolution by sampling
     down in a retouching program or re-scan at a lower resolution.

 -   Cropping a little piece of a big image in a page layout program--the
     RIP has to process that whole big image, then *throw away* what's
     outside the cropping window.  Crop first in a retouching program, and
     bring what you need of the image into the page at the proper size.

 -   Misuse of the text wrap feature of PageMaker. Be careful not to overlap
     your wrap points (ie - create a spider web of an irregular shaped text
     wrap where a wrap point is inside another wrap point, this mistake will
     crash the RIP because you are allowing PageMaker to define an illegal
     object (the object of the text wrap outline and associate it with the
     scan object)

 -   Thinking that white borders don't count--white creates data too.
     Crop the borders away while scanning to save the most time, or later
     in your image editing software.

 -   Using a page layout program to rotate big images.  The RIP likes its
     image data straight.  Twisting all those pixels is risky, lengthy
     work.  Rotate in a retouching program.

 -   Scanning Line Art as a grey scale because it has fancy stippled
     shadow detail.  Scan your artwork at a higher Line Art resolution
     rather than grey scale.  2x3 inch 150 dpi 8-bit grey scale is 2 Meg
     worth of data compared to 600K for a 2x3 inch 400 dpi 1-bit Line
     Art scan.



 Color Scanning
 --------------

    First, color scanning _is_ possible on the desktop regardless of what
 your stripper may have said.  And it would be possible for me to write a
 few thousand pages about how to not do it, but in this FAQ, I'm going to
 limit our discussion to primarily greyscale scanning.  But while I have
 your attention I may as well interject my opinion about color scanning.

    For most of us, color scanning is needed for color separations for used
 in four-color printing.  Your Howtek, Microtek, Sharp, or even Nikon color
 scanner does a good job if you have a skilled operator and an exacting
 output service bureau, but don't assume you are going to save any money or
 time. More often, the years of color experience a good four-color shop has
 is worth the effort to use traditional methods, and will save you time in
 the long run.  We have one of the best desktop color systems around
 (Quadra's with SuperMac displays, 128 Meg Memory, 2 Gigabyte Hard Drives,
 Nikon LS3510 slide scanner, and Optronics Colorgetter Drum Scanner System)
 and we choose not to use it for large format and high quality
 pieces.  We instead send these out to a local Chromacom-based Separator.

    If your bent on doing color on your desktop, try a hybrid approach.
 Traditional four-color separator scans the art and gives the DTPer a low-
 res image for placement.  Output is then sent back to the four color shop
 where the high res is used.  Hybrid methods are growing in popularity as
 users realize their "eye" for color and equipment isn't as good as expected.




-------------         PAGEMAKER/QUARK LISTSERV               -----------------
-------------   S C A N N I N G    F A Q    UPDATE 1/22/93   -----------------

                               Part 2

               Do You Really Need All That Resolution?


It seems even the scanner makers are now caught up in the "more-is-better"
routine.  A few years ago 300dpi 16-shade grey was an incredible scanner, to-
day it seems if you don't have a 600dpi 24-bit scanner you'll likely to get
drummed out of the local DTP users club.

But do you really need 600dpi?  It is quite possible that, depending on your
application, you could produce some very pleasing and fully professional
results with less investment.

I suppose I should explain my first question now.  I won't assume anything,
so don't let me insult your intelligence, just overlook any riduculously
elementary parts.

1.
A flatbed scanner is merely a series of CCDs (charge coupled device = light
sensitive integrated circuit) mounted in a stationary row that light
reflected from a piece of flat art is allowed to pass over.  These CCDs
register presence or absence of light (ON/OFF) thus producing a pixel
electronically.  Since they are mounted in a single row that is the way the
electronic file is created, row by row.  Essentially the CCDs are reflected
one row of the flat art at a time until the image is completely built.

2.
That being the case, resolution or the number of pixels written based on what
is reflected is controlled two ways.  The number of pixels horizontally is
controlled by how closely the CCDs are placed next to each other along the
single row.  The number of pixels vertically is controlled by how slowly the
light bar and mirror inch along the length of the flat art thus reflecting
onto the CCDs.  Therefore, the more CCDs and the smaller the steps of the
advancing light bar the greater the resolution.  Currently there are five
major scanner motor manufacturers (many re-packgers) and they all build
their systems essentially the same way.  The highest resolution flat bed
scanning system currently is physically limited to 400 dots (pixels) per
linear inch.  You may say wait a minute I've seen 600, 800, and 1200 dpi
ratings, but note I stated the physical limit is 400 dpi.  This is true,
there currently is a real physical limit as to how many CCD ICs can be placed
side by side in one inch and that limitation is 400 right now.

3.
This 400dpi physical limitation has been breached by what is known as
interpolation.  Interpolation is a software/firmware process whereby the
scanner essentially samples two pixels and averages (often times using more
complex formulas) the two pixels together to form an extra pixel (or more) in
the middle.  Better scanners now do this in hardware, but some still rely on
their scanning software to do it (often uninvolving the user).  But
nevertheless, this higher resolution is only psuedo-data.  That is, it is
data being create by averaging and not by actually sampling it from the
original art.

4.
Grey-scale scanners are scanners with CCDs that can differentiate between
levels of light falling on them, rather than just being on or off, the grey
scale scanner can determine if the pixel should be any number of shades of
grey.  Most scanner manufacturers have stopped R&D once they acheived 256
shades of grey because the current version of PostScript can only recreate
256 levels of grey, however there are many high-end systems that can produce
much more levels (4096, 32768, etc.)

5.
Color scanners are nothing more than grey scale scanners that have filtration
(most commonly Red, Green, and Blue) and make multiple passes to generate 256
levels of each RGB component.  The software then recombines the three passes
to create full color.  Higher quality scanners perform all three scans in one
pass at the same time to perserve registration (although this is not usually a
problem in good three pass scanners).

6.
Imagesetters are unable to produce continuous tones, that is that
are unable to make a pixel different shades of grey like the scanner "sees"
them, instead they use a very complicated screening pattern to simulate
shades of grey to the naked eye.  At higher magnifications you can easily see
that a photo is not really a photo, but is a series of variable size dots.
This is called "half-toning."  Half-toning, therefore, isn't using all of the
pixels to create the dot patterned image, in most cases it's using only half
or less of a high resolution scans original data.

7.
So the question becomes why scan so high if the data won't be used.  There is
a formula for this of course.  It is, scan at 1.5 times the lines per inch
(LPI) of the final output device.  Therefore if you are outputting to a 2400
dpi imagesetter at 150 LPI then the normal maximum resolution you need to
scan at is only 200dpi.  So if the un-informed user scanned his photo at
600dpi thinking he needed that high res capability because he was going out
to a high res imagesetter, he would be sending over 9 times too much data to
the imagesetter.  This would result in a very long RIP time and possible
crash of the RIP.

8.
Line art scans (black and white) could, of course, be a good candidate for
the high resolution scanner, but more likely if you plan to use a piece of
line art you can scan a large original and reduce it.  Or better yet,
autotrace it into your favorite illustration program and forget having to
work with large, slow bitmaps.

9.
High resolution color becomes a different story somewhat.  It is possible to
find a continuous tone color output device where it would be nice to output a
"true" continous tone modification that would rival the original (i.e.-
National Enquirer PhotoShop modification of a 35mm slide at high res and then
reoutput to 4x5 negative on a film recorder to produce a retouched print).
However, continuous tone output requires very high resolution to produce
satisfactory results.  This requirement pushes the upper envelope of flat bed
scanning (800, 1200dpi) and becomes a job for the slide scanner and drum
scanner.  Unfortunately at these ultra high resolutions the personal computer
becomes a liability.  Today I have been working on a photo retouch of a 35mm
slide that we intend to re-output to 4x5 on our Solitaire film recorder.
Unfortunately the full resolution file (4800 dpi) was over 250 Megabyte!
This is definitely not a job for PhotoShop.

10.
Finally, I have listed a few scanners in different catagories to show
their relative merits and estimated pricing, they are all units that we
use or have used internally and found them to be very reliable:

300dpi Grey Scale:
    Hewlett-Packard ScanJet                     $  800
    Apple OneScanner                               700

400-600dpi Color Flat Bed:
    Microtek 300Z (300dpi)                       1,200
    Microtek 600Z (600dpi)                       1,500
    XRS 3cx (AutoRads & X-Rays, 600dpi)          4,500
    Sharp JS450 (400dpi)                         6,500

600-2000dpi Color Flat Bed:
    Sharp JS600 (600dpi)                         8,000
    Imapro XL (1200dpi)                         14,500
    AGFA Imagemaster (1600dpi)                  22,500
    Artronic ViewScan (2000dpi transmissive)    35,000

3000dpi Slide Scanners:
    Nikon LS3500                                 7,500
    Nikon LS3510 (12-bit)                       10,500
    Leaf 35                                     14,000
    Leaf 45 (35mm & 4x5 12-bit)                 19,000

4000-6000dpi Drum Scanners:
    Dangraf DeskDrum (4000dpi, 12-bit)          22,500
    Optronics ColorGetter (6000dpi, 16-bit)     85,000




-------------         PAGEMAKER/QUARK LISTSERV               -----------------
-------------   S C A N N I N G    F A Q    UPDATE 1/22/93   -----------------

                       Questions and Answers


-- Where can I find information about specific scanning courses?

    Systems of Merritt, Inc.           (authors of "Inside PostScript"
    2551 Old Dobbin Dr. East            and original QMS color PS models)
    Mobile, AL  36695
    [phone]

    Peter Fink Communications, Inc.    (publisher of "Desktop to Press"
    120 Q Street NE                     and member of development team for
    Washington, DC  20002               Adobe's Accurate Screens technology)
    [phone]

    American Newspaper Publishers Association
    11600 Sunrise Valley Drive
    Reston, VA  22091
    [phone]

    TypeWorld/Color Publishing Systems Seminars
    Barrie Evans
    PennWell Publishing Co.
    [phone]

    QuarkXPress Users International
    P.O. Box 170
    Salem, New Hampshire  03079
    [phone]

    Graphic Arts Foundation
    4615 Forbes Avenue
    Pittsburg, PA  15213
    [phone]

    ACM/SIGGRAPH
    Conference Management
    401 North Michigan Ave
    Chicago, IL  60611
    [phone]



-- What are the specific copyright laws, et.al., governing scanning
-- for use in my DTP?

   Jeff Needleman provides the following information regarding this:

   Use a Scanner, Go to Jail
   -------------------------
   by Mark H. Anbinder -- TidBITS Contributing Editor

   Hewlett-Packard included a bulletin in a recent mailing to dealers
   warning them that, when demonstrating the capabilities of HP
   scanners, they must avoid scanning money and other "sensitive
   documents." Anyone who does scan such documents risks
   "Constructive Seizure" of their computer equipment, up to $25,000
   in fines, or up to fifteen years imprisonment.

   Apparently HP has learned of an incident where U.S. Treasury
   agents seized an HP ScanJet IIc scanner, HP DeskJet 500C printer,
   and an HP Vectra personal computer. The dealer and HP sales
   representative involved spent a considerable amount of time
   retrieving the equipment.

   HP provides this list of guidelines from the U.S. government.

   Unacceptable scanning:
   * Money
   * Federal Reserve notes
   * U.S. postage stamps
   * Foreign postage stamps
   * Revenue stamps
   * Other negotiable valuated articles (for example, checks, bonds,
     and securities)
   * Identification documents (for example, driver's license and
     governmental identification documents

   Acceptable scanning:
   * Photographs of people, places, or things
   * Pictures from magazines, newsletters, and calendars
   * Other similar non-sensitive documents

   We wish to add that, if you do scan photographs (or even text)
   from copyrighted publications, it's important to secure permission
   before using that material in any way.

   Perhaps the government is concerned that computer input and output
   devices are becoming powerful enough that counterfeiting is (or
   will soon be) a real concern. We've seen the latest output
   technology, though, and we're skeptical that counterfeiting with
   multi-million-dollar technology would be cost-effective! Of
   course, until output quality catches up with the government's
   concerns, we'll be unable to avoid the image of a crook lugging
   along a 24-bit color 1152 x 870 monitor, trying to convince
   someone that the scanned image on it is legal tender!

   [This article is being published simultaneously in TidBITS and
   Clicks!, the newsletter of the Ithaca Macintosh Users' Group.]

   Copyright 1990-1992 Adam & Tonya Engst. Non-profit, non-commercial
   publications may reprint articles if full credit is given.

   Until today, I was quite sure that scanning fell under the usual laws
   restricting reproduction of U. S. and foreign security obligations. The old
   laws were liberalized back in 1958 and again in 1968 to allow reproduction
   in some cases. The primary exceptions to the usual prohibitions are given
   in Title 18, U. S. Code, Section 504:

   "Notwithstanding any other provisions of this chapter, the following are
    permitted:
    [A-C provisions deal with postage/revenue/other securities of the U. S.]
     and
    (D) postage stamps, revenue stamps, notes, bonds, and any other obligation
    or other security of any foreign government, bank or corporation, for
    philatelic, numismatic, educational, historical, or newsworthy purposes in
    articles, books, journals, newspapers, or albums (but not for advertising
    purposes, except illustrations of stamps, and paper money in philatelic or
    numismatic articles, books, journals, newspapers, or albums)."

   The section goes on to spell out how the printed illustrations must appear.
   All have to be in black and white, except that postage stamps can be
   illustrated in color; all must be of a size less than 75% or more than 150%
   in linear dimension of each item illustrated (except that US stamps can be
   exact size if illustrated in b&w). The negatives and plates used have to be
   destroyed "after their final use."

   There's another section regarding film, microfilm or slides for projection
   upon a screen of such items; they're allowed completely except for
   advertising purposes (philatelic advertising is allowed), but you can't
   make prints or other reproductions from such films or slides except by
   permission of the Secretary of the Treasury.


-- Do I really need to keep the photographs or negatives from a client job
-- since I've already scanned the images and have them saved with the rest
-- of the job files for any future changes?

   Maybe not, but I'll answer that question by relating our situation.  Our
   department handles most of the medical photography for the university and
   its various medical research facilities.  This generates thousands of
   negatives a year for us to archive.  We currently have a vault of over 15
   years worth of medical photography negatives.

   We began to toss around the idea of scanning them to store them digitially
   then converting some of our Nikons to the new Kodak digital backs to record
   the imagery in digital form from the get go.  Problem though is two fold:

(1) No digital storage medium is as permanant as film.  Proof is the fact
    that we have film stored for over 40 years that is still in good enough
    shape to make a print and/or re-archive.  Hard drives we bought only five
    years ago are now dead or dying, tape backup is about as reliable over
    long periods of time as an 8-track tape, and optical storage systems once
    thought to be truly permanant have now been shown to begin substrate cor-
    ruption after only years service.

(2) Most digital storage medium mechanisms are bound in time technologically.
    That is to say, if you standardize on DAT tape whose to say that DAT tape
    will not go the way of QIC 20 format or 9 track format.  The mechanism
    may not be available in the near future let alone the distant future.

   So to compensate for this many organizations develop elaborate redunancy
   policies regarding digital storage.  This means that they get multiple
   copies of the image in digital form often on different types of storage
   systems. Redunancy and multiple mediums drastically increase the
   physical space requirements for image storage.  In short, save the
   negatives.


-- The question I have is where do they come up with setting the
-- LPI to 53, 83, and 106. Are these magic numbers or what?

   Almost!  The screen rulings (LPI) are based on a mathematical formula
   that takes into consideration, among other things, the occilation frequency
   of the video clock generator in the laser control circuit.  What?  Suffice
   it to say that the rulings are practically magical.  As a matter of fact
   most screen frequencies and angles used in four color separating in more
   expensive imagesetters are arrived at with a great deal of secrecy because
   these little differences mean the world to those little color rosettes.



-- What exactly is the difference between a PICT and a TIFF file format?

   For the most part they contain the same type data when it comes to bitmap
   information, that is each pixel in the image.  But the PICT file
   specification (created by Apple and based on Quickdraw) can also contain
   vector data.  Vector data is resolution independant.  For example, instead
   of a series of pixels making up a line across an image in a bitmap, a
   vector line would be an x,y coordinate where the line starts and stops
   and then a value for its thickness and color.

   As you can see the PICT file can be much more complex than a TIFF file and
   this is why Aldus along with Microsoft created the TIFF file specification.
   They needed a simple yet "rich" file type that could handle greyscale data
   from a scanned file and be exchanged between various machines and operating
   systems.  TIFF is the most widely supported bitmap file format to date.  It
   is now capable of handling not only greyscale data but color data as well
   and even has the ability to store data in a compressed form (LZW and
   Packbits in the 5.0 spec, CCITT IV and V, and JPEG/M-JPEG in the 6.0
   spec).  PICT currently has only one method for compressing a file,
   through the use of the Quicktime/Quickpress extension from Apple, this
   allows a PICT file to be stored as a JPEG compressed file and read by
   any application without conversion first.



-- Which format is best for a specific image (i.e. a simple B&W, a scaled
-- image, a color image, etc.)

   It depends entirely on your situation.  Remember, that more pleasing it is
   -- the more memory and the slower your system will be in manipulating them.

   In my opinion TIFF is the best file format for all of these examples
   mentioned.  The basic TIFF file (5.0) can handle 1-bit, 4-bit, and 8-bit
   greyscale data, and can handle 8-bit, 16-bit, and 24-bit color data.  The
   most common forms however are 1-bit and 8-bit greyscale, and 8-bit and 24-
   bit color.  With the two basic forms of compression recognized in the
   TIFF spec you can successfully compress all of these types of images so
   that they take up less space in storage without worrying about using
   some secondary compression utility.




-- Why are there so many different DPI scales?  What common laser printers
-- use what DPI?  What DPI do image-setters use?  What DPI can PageMaker
-- handle?  MicroSoft Word?  Quark?  What about an Image Writer or a
-- Stylewriter?

   DPI or Dots Per Inch is a clever marketing tool that the early laserprinter
   manufacturers came up with that really doesn't mean a whole lot.  A few
   years ago, imagesetter manufacturers and scanner makers were talking in
   the more descriptive SPSI (spots per square inch) and LPI (lines per
   inch) measurements.  It was very common for imagesetters to produce 1.6
   million SPSI at 133 LPI, or 1270dpix1270dpi with 256 grey scales.
   Marketing types couldn't compete because the laserprinter seemed paltry
   in comparison, 90,000 SPSI at 53 LPI, or 300dpix300dpi with 14 grey
   scales.

   Anyway DPI has become dominant because laserprinter manufacturers had more
   advertising money.  Most laserprinters are now showing their DPI uniformly
   anyway, so you can simply square the DPI and get an accurate SPSI.

   So DPIxDPI = SPSI which is much more useful in determining how much better
   one laserprinter is over another one.  For instance, the original Apple
   Laserwriter is 90,000 spsi (300x300), the NeXT Laserprinter is 160,000 spsi
   (400x400) which is 1.7 times better not 1.3 as the 400/300 ratio would
   suggest.  Where this really gets interesting is with the new crop of 600dpi
   lasers.  For instance, the new QMS 860 is 360,000 spsi, or 4 times better
   not 2 times as the 600/300 ratio would suggest.

   Below is a list of some common printers and their SPSI:

    Printer                         SPSI           Type
    ------------------------------------------------------
    Epson LX-80                       14,400    9-pin Dot Matrix
    Apple Imagewriter                 20,760    18-pin Dot Matrix
    Apple Laserwriter                 90,000    Canon CX Laserprinter
    HP LaserJet III                   90,000    Canon SX Laserprinter
    Panasonic KXI1124                129,600    24-pin Dot Matrix
    Canon BJ10e                      129,600    Canon 36-valve Bubble Jet
    NeXT Laserprinter                160,000    Fujitsu Laserprinter
    Dataproducts LZ960               160,000    Fujitsu Laserprinter
    Compaq PageMarq 20               320,000    Xerox Laserprinter
    HP LaserJet IV                   360,000    Canon BX Laserprinter
    Lasermaster L1000                600,000    Fujitsu Laserprinter
    Linotronic 170                 1,612,900    Infrared Laser imagesetter
    Linotronic 330                11,444,689    Helium Neon Laser imagesetter
    AGFA SelectSet 5000           21,622,500    Helium Argon Laser imagesetter

-- What are the best ways (if any) to compress a graphics file once it's
-- been scanned?

   The TIFF format provides a Compressed TIFF version -- that is usually
   recognized anytime that Uncompressed TIFF is.  Other formats can be
   manually compressed -- but will need to be manually uncompressed anytime
   you need them.  Manual compression is used more for archiving the file
   for storage and can save a significant amount of space.

   Without loosing any of the data, most forms of compression are based on the
   LZH (Lempel-Zimpel-Huffman) algorithm, they are:

    For the Mac:         For the PC:        For UNIX Machines;
        Stuffit              PKZip             Unix Compress
        CompactPro           ARJ               Zoo
        Disk Doubler         Zoo               PKZip
                                               ARJ

   They are all good and most will compress a file within 5-10% of the other.

-- My Hewlett-Packard ScanJet came with a rudimentary scanning package.
-- Is there better scanning software than DeskScan?  What are the main
-- features to look for in scanning software?

   Concensus favorite for grey scale scanning is Ofoto.  Look especially
   for software that gives you the ability to scan and save in all the
   common file formats that you will want to use, and your scanning software
   should definitely allow you to crop and rotate before or during saving.
   If you don't already own an image processing package, check out Photo-
   Styler for the PC and PhotoShop for the Mac.  These packages won't
   require separate scanning software but rather allow for the addition of
   a "plug-in" module that will allow scanning from inside these applications.


 Finally
 -------

    If you have any questions or additions to this FAQ, please feel free to
 send them directly to me:

    Jeff Bone, Electronic Imaging Coordinator
    University of Alabama at Birmingham
    School of Medicine
    933 19th Street South, CHSB-19 Room 320
    Birmingham, Alabama  35294-2041

    voice [phone]
      fax [phone]
    email [e-mail]