Black
Grey Yellow
Green
Blue
(Change background color on Mouse-Over above)
The Hewlett-Packard DesignJet® series are professional large format color printers that are capable of printing up to 54 inches wide ! To take full advantage of that resolution, would entail a very large file - even for an 8x10" print. For example: the number of pixels required would be:
8" x 600 dpi = 4800 pixels, by 10" x 600 dpi = 6000 pixels - - or 4800 x 6000 = 28.8 Million pixels !
ok..... so my digital camera is only capable of 1 million pixels or so. (say about: 1024x1024).... how can I even make an 8x10 inch print without seeing every individual pixel, much less produce even larger prints, you ask ?
Well, first of all, until CCD digital cameras increase pixel density ten fold, certain corners have to be "cut". This isn't as bad as it originally sounds however ! Highest quality offset printing is done at about 133 pixels/inch. So an 8 x 10 inch print matching this quality would require only the following..
8" x 133 = 1064 pixels, by 10" x 133 = 1330 pixels - - or 1064 x 1330 = 1.4 Million pixels !
So acceptable quality can thus be achieved at much lower pixel densities. Now as it turns out, perfectly acceptable results are obtained at even 72 dpi utilizing a printer technique known as "dithering". Thus an 8 x 10 printed at this density would require a mere 414,720 pixels. Even all but the lowest quality digital cameras should be capable of a half decent 8x10 print.
Even so, larger prints "eat up" even more pixels. A 30 x 40 inch print, printed at even a lowly 72 dpi would require 6.2 Million pixels.
and a 50 x 62" print at 72 dpi would still require 16 Million pixels. (1.1 Billion if at 600 dpi). Add to that 16 bit color depth for each red/green/blue channel per pixel (48 bits total), and the file sizes become staggering.......
Well, it's done in software. High end image processing algorithms, employ a "bi-cubic" scheme of effectively "adding" missing pixels. Simply put, the software "looks" at each surrounding pixel and adds the required number based on the color and luminance of the surrounding area. In this way, the final image will not "pixelate" - that is: exhibit a digital mosaic pattern (highly undesirable). Other processing software we employ, takes an even more sophisticated approach by selectively generating mathematical fractals to better approximate what pixels to add.
The usual software approach does not however, add any more detail information to what was originally captured. Thus a 50 x 60 inch print that originated from a 1.2 mega pixel camera will appear quite "soft" if viewed from close up. More sophisticated image processing can greatly enhance the outcome ! (See link to image examples below). Though a large print using these techniques can be printed from a low resolution imaging device, naturally much better results are obtained with higher pixel densities to begin with. (pictures of clouds and blue sky for example). Structures and signage are the most demanding, as exact image detail is critical. Aerial Photography places the highest demands on true image resolution. Everyone it seems wants to be able to read license plate numbers from 300 miles away !
A. Number of resolvable
elements (pixels) that were originally captured.
B. Proper focus/exposure/lighting etc.
C. Inherent subject detail.
D. Professional image processing software.
E. Paper and ink quality.
F. A professional quality printer.
These are the major critical areas that can be controlled. Other factors are "gamma" of the recording device (the dynamic range or how "contrasty" is the sensor), and what compression techniques were used to store the image.... which leads to the next topic.
File sizes for images can be immense. An 8 Mega-pixel mid resolution camera such as the Nikon 8700, will result in 23 Mega-byte uncompressed file sizes, while scanned media can easily be well in excess of 500 mb. All digital cameras employ compression software, to reduce the size of the captured image to a more manageable level, as well as allow more images to be stored on the memory card. Almost all employ the JPEG format, as it is capable of reducing the file size considerably. There is a price to be paid for such nice tiny files....... JPEG is a "lossy" technique. Although it preserves 24 bit color (all color information is retained) it simply tosses out redundant or closely redundant pixels. How close that redundancy relationship is, is determined by the quality setting you selected upon compression. (Some software will not allow you select a specific setting). The downside of JPEG is that after a file has been compressed and de-compressed a number of times, those losses accumulate...... much like a copy of a copy, of a copy etc etc..... Thus for the best image quality, files should be transferred as well as archived in the TIFF (.tif) or preferably PNG (.png) formats. Once they have been saved only once in a "lossy" format, the damage has been done ! TIFF performs no compression (lzw disabled) while PNG is a loss-less file compression format. Either will be a visually perfect copy !
Probably goes without saying, but NEVER use the digital zoom "feature" on your digital camera. Use optical zoom or better yet: get closer to the subject..... Enough said !
If you do a lot of wildlife photography, you're better advised to purchase a camera with a good 10x OPTICAL zoom lens or greater, rather than to rely on digital zooming.
One camera manufacturer (who shall go nameless) hyped about their camera having an 800 to 1 Digital Zoom....
WOW ! ........ Sounds impressive !!! This is even more powerful than most expensive serious amateur telescopes are capable of achieving, and photographing the craters on the moon, distant galaxy's, deep space objects and planets should be a snap !!....... Just think of all the money NASA could have saved on the "Hubble Space Telescope" had they consulted this manufacturer's marketing department first !!! ........... Course there is a "downside" the market'eers "forgot" to mention ........ It being a 3 mega pixel camera, that would result in a final image being about 3 by 4 pixels with it digitally fully zoomed in ! ........... (Maybe not such a hot "feature" after all ......)
Of all the scanner types, the flatbed is by far the most versatile. Seems like everyone and their brother is now in the business of manufacturing flatbed scanners. Only but a decade ago, flatbed scanners were affordable by only the professional user. Today, with fierce competitive forces, the efficiencies of high volume mass production & improved technology, what was once relegated to the professional ranks is now affordably available to even the casual home consumer. Flatbed scanners today range in price from about $39.95 to professional models running into the multi-thousands. Not surprisingly, a $1600 scanner has many more features/capabilities & will make a much better scan than the $39.95 model.
Sadly, most manufacturers "hype" the specifications, making their model sound much better than it actually is. After reading this section, you'll know what to look for when reading through the maze of specifications and be able to separate out the marketing lines of "bull" from the true specs. First we'll cover some basic scanner design concepts and begin by looking at the 2 dominant light sensing technologies. CCD's are used in better scanners, while CIS (Contact Image Sensors) are used exclusively in the low end scanners.
CCD (Charge Coupled Devices)
Basically, a light sensitive semi-conductor...... the same as used in digital cameras. Instead of a rectangular matrix of elements laid out in grid fashion as in a digital camera, a scanner CCD array is one dimensional... That is; it's usually a single strip of CCD's . A light source is passed over the document and the reflected light off the print is focused on the CCD array by a system of mirrors and a lens. The true theoretical optimum optical resolution of the scanner is determined by the number of CCD elements per inch.
CIS (Contact Image Sensor)
A much less expensive photocell array to manufacture, the CIS sensor negates the need of mirrors or a lens to direct the reflected light from the document or print to the sensor. Instead, the CIS is placed extremely close to the the print, whereby the light reflected directly off the document falls directly on the sensor. The advantages to a CIS, is that without need for mirrors or a lens, the scanner can be made very small and cheap ! Course, the downside is that resolution, image quality, dynamic range and color fidelity leave a lot to be desired.
Flatbed Scanner Optical Resolution
The ONLY specification that actually counts here is the true optical resolution. That is, the number of CCD or CIS elements per inch. When purchasing a scanner, beware how the manufacturers rate their resolution. Many scanner manufacturer's spec this as (for example) 1200 dpi x 2400dpi. The true optical spec is the lower rating of 1200 dpi (the actual number of ccd elements/inch), and not the 2400 dpi value you might suppose (and they lead you to believe). From a "marketing" standpoint, 2400 sounds a LOT better than 1200.... The so called 2400 dpi is achieved by "faking it"....... that is; stepping the scanner drive by only 1/2 the normal distance per vertical line. The simple reality is: the optical sensor still resolves only 1200 dpi no matter how you slice it (or step it !)....... Half stepping the linear vertical distance results in twice as many samples per vertical step than the CCD array can resolve. The result is that vertically adjacent pixels instead of being discrete values of each sampled dot, is rather an average between the two overlapping sample points. Thus absolutely nothing is gained as far as image resolution is concerned. Note: most marketing types will advertise their scanner as being 4800 dpi when in fact it's true optical resolution is only 2400 dpi. The 4800 dpi is again the result of 1/2 stepping the drive motor. This marketing ploy is at best misleading if not an out and out lie. Carefully read the technical specs ! - The "tip-off" to this misleading ploy is usually buried in the specs and is recognized for example as an optical resolution spec of 4800 X 2400 dpi...... Their main banner advertising boasts an impressive 4800 dpi (a lie), but when you dig into their specs you'll see it listed as being 2400 x 4800. Some even reverse the spec (4800 x 2400) as it "looks" better (4800 being the first number you see and thus making the most impact). Whatever the sequence, totally ignore the higher of the two, as the true optical resolution is in fact (theoretically) only 2400 dpi ! However, read on further..... even this reduced resolution might be stated as being grossly over optimistic.
Flatbed Scanner Interpolated Resolution
Another total meaningless specification as to scanner quality, is the interpolated resolution. A native optical 600 dpi scanner for example, often comes with software that may allow for example, so called interpolated resolutions up to 9600 dpi or even more. The 9600 dpi image for example, will be no sharper, nor will it contain any more detail than the originally scanned 600 dpi image, as 600 dpi was all that the image sensor was able to capture in the first place. The additional 9200 pixels/inch were derived in software, by averaging adjacent pixels from the original optical scan. Yet another marketing ploy to make the scanner sound much better than what it actually is.
Dynamic Range
Dynamic range is probably just as important a criteria for rating a scanner as is it's true optical resolution. The dynamic range is simply a measure of the scanners ability to record a wide range of luminance values ranging from pure black to pure white. The agreed upon scale ranges from a D (Dynamic Range) of from 0 to 4, with 0.0 being perfect white, and 4.0 being perfect black. The higher the number, the better the scanner will be able to resolve and record the subtle variances in luminance - especially in the shadows. Most flatbed scanners have a Dmax of around 2.5 to 3.0 (2.5 isn't all that great). A D of 3.2 or above is considered good for a flatbed scanner. (Note that high end drum scanners often have a Dmax of 4.2 or greater.) The images obtained from a low end scanner with a low dynamic range will be somewhat "contrasty". Highlights will often be washed out, while shadow detail will be lost in the "mud", especially if the original media was not properly exposed.
What determines the dynamic range you might ask ? Part of it is determined by the physical characteristics and response of the CCD array itself. Not all CCD's are of the same construction & response, and performance can vary widely. The other major influence is the total number of digital data bits available (known as the bit depth) for storing the values for each R, G or B channel. The fewer digital bits available: the less amount and range of information that can be stored, and thus a lesser dynamic range. What also plays a part is the quality of the optics system as well as stability and purity of the light source.
Another highly deceptive marketing ploy, is how some manufacturers typically advertise dynamic range specifications this way: - The specification is based solely on the maximum theoretical value of image range that can be stored. This calculated spec is based solely on the amount of available bit depth, as opposed to the true measured dynamic range which is substantially less. You'll often see "(calculated)" next to the specification.
24 bit scanners - specifications near 2.4
30 bit scanners - specifications near 3.0
36 bit scanners - specifications near 3.6
42 bit scanners - specifications near 4.0
48 bit scanners - specifications near 4.4Note: a 48 bit scanner actually describes a scanner whose array is sampled at 16 bit depth per each R G B channel (16x3=48). Thus each color is sampled at 16 bit depth (2℮16 or 65,536 discrete luminance values/color channel). Early scanners were designated and advertised as being 8, 10, 12, 14 or 16 bits - referring to the number of bits allocated for each channel. In today's terms, that equates to 24, 30, 36, 42 or 48 bit scanners respectively. A 48 bit scanner for example resolves 16 bit depth per channel. 48 just sounds like it's a lot more advanced. Once one manufacturer started advertising this way, the others were forced to follow suit.
Anyways, if the advertised "D" is theoretically calculated this way based only on the total bit depth, then you can be absolutely certain that the actual Dmax will be less - owing to the characteristics of the CCD array, optics & light source. In general however, anything less than a 36 bit scanner will be "crippled" by simply not having enough bit depth to store the maximum amount of image data to realize the CCD's maximum potential for dynamic range. Sadly, the actual D is often not published by many manufacturers, in which case you will have to rely on professional reviews of the equipment or better yet, taking some poor photos or a grey scale test chart along and actually see how the scanner performs.
Note: Though it doesn't sound like much, the Dynamic Range Scale is logarithmic and a scanner that has a D of 3.6 for example, will "blow a 3.2 D scanner away". Most inexpensive scanner manufacturer's don't even list the dynamic range of their product - either theoretical or true..... (and probably for good reason).
Transparency Scanning
Most flatbed scanners today come with a transparency adapter that enables the scanning of negatives or transparencies (slides). Though most flatbed scanners support scanning a negative or transparency, much better results are obtained by using a dedicated film scanner specifically designed for such purposes - especially the small formats such as APS, 35mm, 6x6cm, 6x7cm & 120 roll film. Film scanners typically have a much higher dynamic range than does a flatbed, and the optics and mechanical transports are designed out of necessity to much tighter tolerances. A good film scanner is also quite expensive in comparison to a flatbed.
Question: Ok.... I found 2 scanners both offering 2400 dpi optical resolution. One is $99.95 and the other is $1600.00 ! Why the HUGE disparity in price ??? ..... won't they both resolve 2400 dots/inch and make just as great a picture ?
Answer: Sadly, the answer is NO...... (not quite what you wanted to hear ?). The 2400 refers to the number of sensors/inch on the sensor array. True, both are capable of resolving 2400 dpi from a theoretical standpoint, but from a practical real world standpoint, the differences may be (read: probably will be) significant.
Just because a scanner has an sensor array of say 2400 dpi, doesn't necessarily mean that it will be able to resolve anywhere's near 2400 lines per inch on a resolution test chart. The sample spot size (each pixel) must be matched to the number of dpi. Sloppy manufacturing techniques and tolerances in the mirror and lens systems can often result in an excessively large sample spot size, such that there is greater than 10% overlap of adjacent samples. 50% overlap for example, would result in actually something less than 1200 resolvable lines per inch. Thus the 2400 dpi spec for practical purposes might be completely bogus. True, it will sample 2400 dpi, but might not be able to resolve anywhere's near 2400 dpi !
Also, like a camera, the image must be held steady while the image is being exposed (or scanned in this case). Any movement, and worse case, you'll get a "perfectly exposed blur" ...or variations thereof. A well constructed commercial grade scanner employs highly accurate, smooth stepping motors housed in a heavy rigid cast frame with good isolation dampening. The $99.95 special you can safely bet, will be constructed out of "Genuine Re-cycled lightweight Plastic" throughout. The entire unit shakes, hums and vibrates as the image is being scanned. Even if the scanner were to have the best CCD array and optics, it will be all for naught if they are housed in a glorified cement mixer.
Also critical is the back focus. Cheapie scanners are factory set to focus on the media side of the platen. One setting does all, but it is susceptible to getting knocked out of whack (plastic isn't all that structurally sound a material ). Better scanners use a computer controlled focusing motor that optimizes the back focus for each print scanned.
Optics are also a critical component.... Keep in mind that most scanners have a "sweet spot" where the highest resolvable number of lines is obtained. It is usually (though not always) a 4 x 5 inch area in the center of the platen or often times a relatively narrow strip down the middle, depending upon the optics design. Lens aberrations as well as astigmatism result in softening of images at the edges. It is for this reason that images (& especially transparencies) are to be scanned in the sweet spot. Keep in mind that some manufacturer's rate the resolution of their product only in this "golden spot"... Anything outside this sweet spot will probably yield results substantially less than gloriously claimed and touted. Note that inexpensive scanners employ "genuine" plastic lenses. A plastic lens is no match for a high quality coated 5 or 8 element glass lens - the glass lens itself costing more than the price of the entire cheapie scanner alone. Naturally, all this is a lot more expensive to manufacture. The optics system is a critical component of any scanner. The highest dpi CCD array will do nothing for resolution if the optics system is a piece of junk, as all you'll end up with is a "perfectly exposed blur"
A penetrating glimpse of the obvious perhaps, but It's also a pretty safe bet that an inexpensive scanner will have poor real world dynamic range. (2.7 or lower).
Alas, you tend to get what you pay for. Something else to consider is that the bargain scanners will soon be outdated, whereas a quality device will typically have 2 to 3 times the life before newer technology makes them obsolete. Thus in true dollar costs averaged over it's life, a better grade scanner may actually be cheaper or maybe be only marginally more expensive in the long run. In general, it's recommended to purchase the best scanner to match your highest level of anticipated use. But also keep in mind that If all you need is a small pickup truck, there's no need in purchasing a Kenworth W-900 Condo Cab 18 Wheeler with a 600 HP CAT Turbo Diesel hooked to a 53 ft trailer to make an occasional run to the town dump. (perhaps a bad analogy, but still fitting...) Thus for non-critical general scanning & pictures sent to Grandma, the $39.95 to $99.95 special will probably be just fine. Keep in mind that less than 20 years ago, a scanner offering the capabilities of the $99.95 special of today, cost then close to $10,000.
Surprisingly, TWAIN isn't an acronym for anything meaningful.
The word TWAIN is derived from Kipling's "The Ballad of East and West" - "...and never the twain shall meet...". It's a play on words reflecting on the difficulty of connecting scanners and other imaging devices from different manufacturers to personal computers and have them recognized by various software applications. Many folks believe it's an acronym for something logical, but such was never the case. A contest was held to come up with a meaningful expansion of "TWAIN", but none seemed appropriate. The closest anyone ever came was the entry "Technology Without An Interesting Name" which "stuck". It's not the official name and it's use continues to haunt the standard. Anyways, now you know where the name TWAIN came from !
So just what is it ? ......
TWAIN is a software protocol that serves as a link between the application you are running to import the image and the scanner. The TWAIN protocol is thus a standard to which the device driver is written and has been agreed on by the software as well as hardware manufacturers. Almost all applications are TWAIN compliant, and just about all scanner manufacturers' write their device drivers to be TWAIN compliant as well. TWAIN allows whatever application you are running to recognize the scanner and effect communication with the device. When you see advertising that states "TWAIN Compliant", it means that the device driver for the scanner (or other imaging device) is exactly that.
Purchasing a professional grade scanner makes no sense if your only intended use is for scanning snapshots to be included in emails to Grandma. In that case, even the $39.95 scanner will probably suffice.
Most scanners in the $150 to $250 price range, offer improved resolutions, dynamic range and a host of other improvements, features & controls, making them suitable for the scanning of family photos and general small office applications. Scanners in the $300 to $1,000 range offer improved quality and are targeted more to professional markets. Scanners selling above $1000 typically are devices usually offering larger scan areas and enhancements/features targeted to the graphics professional.
The standard scan area is typically 8.5 x 11 inches for scanning typical letter size documents (sometimes referred to as A4 size) or 8.5 x 14 inches for scanning legal size documents. Larger format scanners coveted by engineers, designers and graphics professionals, typically scan 12 x 17 inches. Naturally, the larger the scan area, the greater the price tag.
Many scanners support the capability to scan in transparencies, negatives and slides. Some have a built in light table, while others require an adapter. Some will only handle 35 mm slides, while others will handle almost any photo format such as 35mm, 120 film, APS, 6x6, 6x7 & 4x5 formats. Other specialty scanners will accept 9x9 aerial formats as well as transparencies up to 8x10 inches and larger. The offerings run the gamut. A flatbed scanner with at least 1200 dpi minimum will suffice for the occasional non critical scan of a slide for printing a small snapshot, but MUCH better results are obtained by use of a dedicated film scanner especially designed for the task. Even the best flatbed scanner is no match for a half decent film scanner - it's not even a contest. Professional ccd based film scanners are now offering resolutions up to 5600 dpi, while high end drum scanners offer resolutions in excess of 12,000 dpi.
Old damaged photographic prints are often the only images to survive... the original negatives having been long lost. Many of these early prints are in rough condition. If restoration of damaged photographic prints is your goal, then Kodak's "Digital ICE" technology warrants serious consideration. The technology is a marriage between the scanner hardware and software. Effectively, the scanner creates in addition to the R, G & B channels, an additional error mapping channel that maps the errors such as creases, tears, scratches & surface dust spots. The software based on the generated error map then applies corrections to remove the defects. It's quite effective on rips, tears, scratches and dust removal and can literally save hours of re-touching in Photoshop - especially on the "basket cases". Most corrections are virtually transparent with no softening of the image as is usually the case using software applications alone. Note however, it's effective only on reflective media and does not work with transparencies. Invoking the software (it is selectable) will easily quadruple scan times. However, considering that it may save literally hours in Photoshop, it's a small price to pay. Currently, there is only one flatbed scanner that has "Digital ICE", but soon the market will be swimming in others... The next generation of high end flatbeds will undoubtedly adopt the Digital ICE technology, which is already being widely accepted and supported in many of the high end film scanners. Note: Digital ICE is not a stand-alone software package, but works in conjunction with the scanner hardware especially designed for this implementation. It's not 100% effective..... you'll still probably have to do some Photoshop work, but this will greatly reduce the time required. If you have a large number of badly damaged photos to restore, this scanner for the moment, is the only way to go. Quite amazing what it will correct !
Photographic prints typically yield no more than 300 resolvable dpi. The rule of thumb for the best results, is to scan at double the inherent resolution of the document to be scanned (ie: 600 dpi in this case) Scanning at higher resolutions will not yield any additional detail. All that will happen is an increased time to scan with much larger resulting file sizes. Scanning an 8x10 at 3200 dpi would be better described as more of an endurance test than a high resolution scan.... If making enlargements from the original, then scan at a resolution appropriate to the degree of enlargement. ( ie: A 4x5 print scanned at 600 dpi would have to be scanned at 1200 dpi if the final print size was to be an 8x10). This is often better achieved by "adding pixels" through software such as Photoshop ® if you're already scanning at a higher resolution than that of the original document.
A better solution that will keep file sizes reasonable, as well as add the capability of image enhancement, is a software package by the name of Genuine Fractals. It's not cheap, but the results more than offset the cost. If making large format prints from low resolution sources, then this software is an absolute necessity.
Note that most optical character recognition programs like to "see" scans at 300 to 400 dpi.
Even the best scanner gets dirty over time. Accumulated debris and dust accumulates on the mirror (s) and lens over time, reducing resolution as well as dynamic range. A simple cleaning can often return performance to new !
Over time, the fluorescent light source grows dimmer as the tube ages, making for muddy scans, poor color fidelity and poor dynamic range. Simply replacing the bulb will do wonders !
Like their scanner counterparts, there's a tremendous amount of marketing hype in the way color printers are marketed. The most misleading claims are in the area of print resolutions. Some manufacturers are extolling the virtues of their printers being able to lay down in excess of 4800 dpi without going to any lengths to explain that the color print will not be capable of resolving anywhere's near that resolution. Most uninformed consumers equate a printer's highest dpi (dots per inch) capability with the resolution or sharpness of the print expressed in lpi (lines per inch). Though inter-related, the two specifications are quite different. Few 4800 dpi printers will in fact be capable of making a print where more than 400 lpi can be resolved. However, from a marketing standpoint, 4800 dpi sounds a LOT better than does 400 lpi, and that's the spec the marketing departments extol and plaster over the outside of their colorful boxes. A 4800 dpi color photo printer must offer over 8 times the image sharpness & quality of a mere 600 dpi printer it follows to reason..... Right ??? Alas, such is NOT always the case.
First of all, the published dpi (dots per inch) in the case of photo printers, usually does not mean the published dpi will translate into an output that can resolve anywhere's near that lpi resolution. Even with perfect optics, inkjet nozzles & operating in a perfect world, you would not be able to resolve in lpi the published dpi. Instead, the output will resolve something substantially less, and expressed as the resolvable resolution in lpi (Lines per Inch). The lpi specification is what you'll actually end up with. The reason for the significant disparity, is that photo printers, in able to produce millions of colors from only a few basic colors available from their ink cartridges, use a technique known as "dithering".
A single pixel may consist of 3 or sometimes as high as 36 more adjacent colored dots to establish the color for some pixels. The number of dots dithered to form the color of each pixel is not fixed, but is rather dependent on the color of the pixel. Other pixels, depending on their color and the number of colors available from the cartridge, could use more or less dots. So a lot of "dots" may be consumed in the dithering process - greatly reducing the potential resolution obtained in the final print.
Because all inkjets employ the subtractive color process, you cannot always literally spray the exact same dot with a mixture of different colors to achieve the desired luminance and hue. Unlike an artist who can mix paints from his palette in different ratios, the inkjet nozzle is an "on-off" affair - an "all or nothing" sort of proposition. It either sprays a fixed size droplet of ink or it doesn't. To get different ratios requires laying down a field of adjacent pixels. To "mix" 1 part Cyan with 10 parts Magenta for example, would require 11 discrete dots and then let our brains and eyes integrate the now multicolored pixel into what appears to be a single pixel of the correct color. Even if the droplets could be mixed, overlaying up to 36 droplets at the same exact location would result in a "flood" of gooey ink - susceptible to smearing, contaminating the print head and taking "forever" to dry.
Exactly how the internal printer microcode decides how many dots are required to make each pixel & then which dot should be which color, is dependent on the number of colors available, the number of dpi the print engine can support, and the dithering algorithm the programmers employed to achieve the most accurate color reproduction. In general, the greater number of color cartridges, the more accurate the color reproduction and the less dithering required.
Naturally, the manufactures' marketing departments all tout the much higher dpi rating, but usually quietly hide the lower real world lpi specification.
What is Subtractive Color you're probably asking ???
Remember all those stimulating high school physics classes that instilled a wonder and amazement for the way our world is put together ? In case you forgot, then here's the gist of it:
Additive color is the process of adding say three known colors together to make almost any other. From only three colors: Red, Green & Blue, we could make almost any other color by mixing each of the R, G & B together in different ratios. Those 3 colors are what's called the Primary Colors. Take for example a case where you have three flashlights, each with it's own R G & B color filters. Cross the R G & B light beams, and where they all intersect, you'll end up with white ! This is the way it works when "adding" light together !
However when you use inks and print on paper, it doesn't work that way, as we are not adding light. Instead, mixing 100% R G & B will result in a "grungy" looking composite black - since when we "paint" on an object, we are simply subtracting colors we don't want reflected. (Red paint for example, is just white paint without any Blue or Green). Thus when we paint (or in this case the printer does the "painting), we are actually subtracting colors - that is; removing the colors we don't want to be reflected from the white paper. The process is somewhat akin to carving, where the artist simply removes the unwanted material to "reveal" the hidden object. In this case, the white paper holds the image and all the painter (or printer does in this case) is to remove the unwanted reflections to reveal the "hidden image" ! - Perhaps a "twisted" way of looking at it, but it accurately describes the process !
Turns out that the primary subtractive colors that inkjets use are Cyan, Magenta & Yellow. (The color compliments of RGB). Mix equal amounts of Cyan, Magenta & yellow together and you'll get solid black or grays depending upon the equal amounts. I won't delve into what's so special about those colors, as this article is on color printers and not meant to be a science paper........ (does make for interesting reading though !)
So why not mix CMY together for each dot ? Well, not only do the colors subtract, but so does the resulting reflectance. Making a brilliant saturated red by mixing equal amounts of Magenta & Yellow would be difficult. "Slop em together" & you'll get a "dingy" saturated red, but not a "flaming" brilliant one. (Who ever saw a dingy red fire truck ?) Anyways, much better to lay down adjacent dots of Magenta & Yellow. Our brains will integrate the two small adjacent M & Y color dots & the bright intervening white space in between, into what the brain thinks is one brilliant Red dot ! (pretty clever, huh ?)
Though what looks like a simple spray of ink being laid to paper, is in fact a somewhat complicated process !
Anyways, that's not to say that a 4800 dpi printer is without merit, it simply means that there are other important considerations. Thus instead of focusing solely on the hyped up highest dpi specs, place far more emphasis on the following......
Beware that just because one model
costs more than another, it's not always safe to assume that it will produce
a higher quality print. A higher priced printer usually offers larger
print size choices, faster print speeds, networking capability etc etc etc,
but none of those features affect the actual print quality. Most color printers
today of the same class and number of inks, all use the same print engines that
employ the very same or similar dithering algorithms.
Don't equate print quality solely
on the basis of maximum dpi. Color saturation, gamut , color stability & the
availability of high quality durable inks should be the first consideration.
(Of course Dye sub printers do not use dithering which gives them a significant
advantage in the resolution and continuous tone departments).
As a general rule of thumb, the more
inks, the better will be the color gamut (the range of colors capable of being
reproduced). It's a safe bet that an 8 color printer will make a better
color print than a 4 color printer since with more colors available, it will
have to resort to less dithering.
Avoid any 3 color printer like the
plague if you're looking for photo quality. (Cyan, Magenta & Yellow only). A
3 color only printer will not be able to reproduce clean dense blacks and won't
deal well with shadow details. Printing large areas text or images that contain
black, will quickly drain the expensive color cartridge.
Some printers offer archival inks
that are UV resistant and not susceptible fading. If the availability of archival
inks is a consideration, this will narrow your choices considerably.
Media costs can quickly surpass any
initial hardware costs. Though the printer may be a bargain, the supplies may
send you quickly to the poor farm. In many cases, the printer is the "loss leader".....
they make their REAL profits on the consumables ! Less costly to keep
fed are printers that have individual ink cartridges. Simply replace only the
color cartridge that is depleted. Multiple colors in one cartridge, require
that you purchase a new cartridge when only a single color is depleted.
Some manufacturers are placing microchips in their cartridges. One the plus
side, they allow estimating the amount of ink left. On the Big negative side,
when one color runs low, the chip tells the system that the cartridge is empty
(in reality it probably has at least 10% left) but will not allow any further
printing. Consistently print one color and that one will quickly deplete.
Meanwhile you have to purchase an entire new cartridge - even though the other
colors may be hardly used..... Nice scam for maxing out the profits !
No manufacturer I'm aware of makes any reference to his printer making a lousy or sub-standard color print, no matter how low the price point. Thus advertising claims as to print quality are almost meaningless. The best way of judging print quality is to actually see the printer in action.
Tips on Purchasing a Digital Camera
To realize the maximum resolution, colorimetry & optical performance, the entire electro-optic system must be of the highest quality. In general, if engineers had their way, each product they designed would be the most advanced product ever conceived - as most live for the challenge ! But like everything else in life, there are always tradeoffs to be made - especially during the engineering review process. Management and marketing groups spend countless hours performing marketing analysis to properly position a new product to insure it's success in the marketplace. Performance, features and price are the 3 main defining parameters that always seem to be at odds with one another. The design engineering team could no doubt make their camera the most technically advanced camera in the marketplace, but at $70,000 each, it would be completely unaffordable. The manufacturer could produce the best camera on the planet, but if no one can afford it, it's a rather mute point. Thus the design team is always burdened with having to make tradeoffs.
In general, the more expensive products will have to suffer far fewer design tradeoffs. But no matter what the price point and target market, keep in mind that any product you purchase, whether it be a digital camera or a Chinese cooking wok, the end product is the result of design tradeoffs having to be made. Most engineering reviews are nothing short of "bloody" affairs - verging on all out wars. You can be certain your prized new camera is the end result of some heated battles !
Obviously a $79 digital camera is no match for a $7,000 SLR with an interchangeable quality lens. But no matter what the price point, there are several major points to look for when purchasing a digital camera.
1.. Pixel Density ! - In the case digital cameras, the pixel density is one of the major design points to be considered. If one is allowed to generalize for a moment; the greater the total number of pixels, the greater the POTENTIAL resolution and thus the capability to make larger sharper prints.All other things being equal, an 8 mega pixel camera will make a sharper, higher resolution picture than will a 3 mega pixel camera. But rarely are all other design parameters equal - even across a manufacturer's own product line, and thus the use of the word "potential".
2. Effective Pixels vs Raw Pixels- Many manufacturer's marketing departments list only the total amount of pixels their image sensor is capable of, in an attempt to make it look like there is more potential resolution than the camera can actually deliver. A much better rating is the Effective number of pixels. What's the difference you ask ? Effective pixels are only the number of those used to actually make up the image. For example: a 12 MP camera (actually 11,808,768 pixels (3968x2976)) if rated for Effective Pixels, would have all 11,808,768 of those pixels used to make up the actual image. Since no design ever uses every last one in effect to the very edges of the sensor, the actual sensor has on average 10% more than are actually used to record the image. That is: the raw pixel count is closer to 13 MP instead of 12 which is the actual useable amount. So why not use every available pixel ? Well to determine the black level, light is not focused on the extreme edges of the sensor, which by sampling those very edge pixels establishes the black level of the image allowing the proper exposure, and setting of the knee of the response curve and thus achieving the optimum gamma response. The other reason is that often the optics system often cannot focus properly the image onto the extreme edges of the sensor anyways. Thus for the purpose of forming an image, the pixels near the edges are not used.
But 13 mp sounds a lot better than 12mp, or 12 a lot better than 11mp etc... no matter that some have nothing to do with making up the image... The better manufactures that are more honest (or less deceptive depending on how you look at it) will rate their max theoretical resolution in Effective Pixels. The more deceptive will not mention Effective Pixels, which usually means a higher pixel count than what's available to form an image. But read on... it's almost a mute point: a 12mp effective pixel camera will not be capable of resolving anywhere's near 3968 lines anyways...
3. Optics System - The highest pixel density will be of no value, if the optics system can't sharply focus the image on the CCD. High quality lenses are in themselves quite expensive. To take advantage of the higher pixel densities, lenses with greater precision and resolving power must be incorporated. But here lies the crux of the problem..... The highest quality optics system could quickly price the end product out of the marketplace, and thus the lens and focusing electro-mechanics systems are in themselves tradeoffs. Also, there is no such thing as a perfect lens, no matter what the cost !
4. CCD Size - The larger the CCD, the less the noise and the better its' low light performance. It's just a physical fact of life that a physically larger CCD has a greater light collecting surface, and thus greater sensitivity. Another advantage of a physically larger matrix, is that the optics system doesn't have to focus the image as precisely on such a small area. The less expensive cameras will almost always have small footprint CCD's (typically 1/3"), while the professional cameras will have 2/3" CCD's. It may not sound like much, but the increase in terms of low light sensitivity and reduced noise when shooting in less than ideal lighting conditions is dramatic ! If you contemplate doing a lot of low ambient light hand held photography, having a camera that employs a larger CCD is a major consideration.
5. Optical Zoom - First of all, Digital zoom is to be avoided at all costs. (see above). Though obviously part of the optics system, optical zoom is such an important criteria, that it deserves special mention. If the subject is too far away, you have two options. The first obvious & preferred choice is simply to get closer to the subject. Since this is not always possible, then having a good zoom lens is the next best choice. Wildlife photography almost always demands a strong optical zoom capability. There are many new cameras on the market now with 10 or even 12x zoom lenses, as in case of the Panasonic Lumix FZ30. Some even offer up to 26x But any zoom lens in itself is more of an engineering tradeoff. It is a LOT easier, cheaper and technically feasible to make a fixed lens than a zoom. Aerial cameras that have to record as much detail as is technically possible, all have fixed focus lenses - each obviously designed for almost perfect focus with minimum aberrations at infinity. But for most photographers, the much greater capabilities of a zoom lens, outweigh the tradeoffs.
6. Image Stabilization - Having a powerful zoom, also results in some "powerful" motion ! A CCD, like it's film counterpart, requires a specified exposure time. While the exposure is being made, the camera/image must be held absolutely stable to avoid the dreaded motion blur... Attempting to hand hold a camera using greater than 6x zoom is difficult, and almost impossible at 10x or greater. Image stabilization is a system to look for that will greatly improve sharpness on hand held photography, and is just about a necessity for lenses that have 6x zoom capability or greater where use of a tripod is not practical.
7. Pixel Transition Ratio - Transition Pixel Ratio This is the real resolution slayer... Definition: The Pixel Transition Ratio is simply a measurement of a camera's ability to capture details accurately. The results are mainly attributable to the lens's resolving power and to a much lesser degree, the raw pixel count or even Effective Pixel count of the CCD (contrary to what most folks might believe). However other contributing factors include the amount of degradation in the image compression technique employed and noise immunity of the electronics. Thus the optical accuracy, not necessarily the digital image resolution potential is normally the prime influencing factor.
The test is performed by focusing on an ISO test target. Any blurring of the image results in grey pixels at the transitions from white to black and vice-versa. The ratio and hence it's name, is derived by counting the percentage of grey pixels at the transition points. This is one case where the lower ratio number equates to greater the effective resolving power. Thus the lower the Pixel Transition Ratio, the better ! Very good to excellent ratio's are below 2%. Perfectly acceptable ratios are in the 2 to 2.5% neighborhood. 3% is on the high side of acceptable, and anything over 5% should be considered unacceptable for all but "snapshot" type applications.
 |
 |
|
Perfect Pixel Transition Ratio of 0% |
Real World Pixel Transition Ratio (2.5%) |
There is a great misunderstanding between the pixel density of a camera and how many lines the camera can actually resolve. (refer to the Pixel Transition Ratio above) For example, one of the better SLR 8 MP cameras on the market, employs a CCD laid out in a 3264h x 2448v pixel matrix. It would appear on the surface, that the camera would be able to resolve 3264 discrete lines, since that is how many effective discrete pixels are laid out in the horizontal dimension of the CCD. But in reality, when measured on a resolution test chart, the camera can resolve no greater than 1700 lines over the entire width of the image. This is typical for even the best digital cameras. The point being, that there are many other factors other than just raw pixel density that determines the amount of detail that can be resolved. Lesson: the highest pixel count camera may not always be the best !
If $ is no object and you require a professional SLR 8 MP camera or greater, then don't cut final corners by "skimping" on the lens. It'll all be for naught if the optics can't accurately focus the image on the CCD !
For marketing reasons, none of the manufacturers specify how many lines their cameras will resolve (ie 3264 sounds a LOT better than does 1700). Most independent reviews that are actually meaningful, will however measure & publish those specs.
For happily clicking away and sending 4x6" prints to Grandma, - all taken in well lit conditions - even a $79 1.2 MP "wonder camera" with a Pixel Transition Ratio nearing "double digits" will suffice nicely. If you intend to make quality 16x20" prints or larger - shoot handheld in low ambient light etc, then be prepared to dig a lot deeper !
No matter how costly or sophisticated the digital camera, keep in mind that even as of today, none can compare to the quality achieved by the larger format 4x5 or 8x10 film cameras.
Last Modified: Apr 9, 2008
Made in Maine
Since 1978 - Copyright © 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 Video Interchange
Terms and Conditions on using this site
Video Interchange - PixelPrinters
92 George Luce Rd
Waldoboro, Maine 04572
207-832-5064
Northeast Region - New England
