Hardware - Monitor2

Please be patient
(LARGE SECTION - to explain Monitors thoroughly - I had to include many images in this section - make sure to read Monitors - Part1 first)

I took a close-up snapshot with the shadow mask and aperture grill monitors. First, I created a few tiny squares (8x8 pixels) and filled them with pure colors such as Red (RGB = 255,0,0) and a few combination colors such as Red-Green (RGB = 255,255,0). Then lower end rectangles are 3x8 pixels.

Shadow Mask - the following is a close-up snapshot of the original image, taken on a 17" shadow mask monitor set to a resolution of 800x600. The actual pixels are outlined over the green square for comparison. Note that is slightly more than one dot triad per pixel (you can only see the green dot, but the others are there - the number of dot triads per pixels is calculated later) :

With the pure colors - the close-up reveals that, the other two dots are turned off, and you see quite a bit of black - the black actually dominates. But, when viewed at normal, small size - you see no black. Black is the absence of light and color, and what happens - is the colored dots are so close together that they blend from afar. There is no way that you can separately see any of the dots - they are too small. So, what you see is a blend of all dots - including the black. For example, the pure green square has black and green dots - 2 black for every 1 green. But the green is the only "color" available, and your eyes can only see color - they can't see black, per se. Your eyes simply see the green color emitting from that area of the screen as well as the slight amount of green that bleeds into the black dots.

You can see that the active color bleeds over to the adjacent dots slightly. This is especially evident in the dots that are supposedly off (Black). Instead of pure black - the phosphor has received some energy from the electron beam extends over to the adjacent dots.

Aperture Grill - as a comparison, the following is the same image, with a close-up snapshot of a 22" aperture grill monitor set to a resolution of 1024x768 :

All we ever hear about with regards to monitors is pixels, pixels, pixels. Oddly enough - your eyes cannot detect pixels. Instead, they are just a unit for measuring image size and resolution - to aid your system in displaying what you see. The only thing that uses pixels, is your system - not you.

Notice that the pixels from the above aperture grill example, outlined on the green square, all contain varying patterns of green stripes. It is evident that the pixels themselves are not at all identical. This shows that the amount of detail of the screen is not discernable to the level of pixel-by-pixel. If it was, your eyes would detect a difference between each pixel. Here are the first six green pixels, shown separately :

Each one is completely different than the next. But your eyes do not seek out individual pixels. Your eyes simply see millions of colored particles on a phosphor screen, and they have no idea of where a pixel begins and ends. They do not see pixels boundaries. Therefore instead of seeing the 6 different pixels shown above, your eyes see this :

In addition, pixels are so tiny, that the human eye can only discern large groups of pixels. The contents of a single pixel cannot be seen. If a pixel has 1 aperture stripe as opposed to two stripes, your eyes could not tell. Pixels are only a means of measurement for monitor settings and resolution. You cannot make out any detail at all in a single pixel.

Below, a series of groups of pixels is shown. You can see that there is no way to see the detail when only one pixel is shown. Even with the larger groups of pixels - it appears to be a constant value of red. In reality, the pixels contain either stripes or dots as shown above - and each pixel contains varying patterns of stripes or dot as we have just shown. It is the overall pattern of these stripes or dots that your eyes see . . . and the overall pattern is a steady, repeating pattern :

Even though each pixel in this example, has widely varying patterns of red within it (the same patterns that you saw above with the green pixels), your eyes don't care because they do not see pixel boundaries - instead, they see the continuous repetitive pattern of the stripe triads (aperture grill), or dot triads (if using a shadow mask) which results in a constant red shade.

Using the shadow mask example from above, I counted approximately 50 triads across one line - so about 100 dot triads per inch since the image is 1/2 inch on my daughter's screen.

What about the separate Dots ?? - this is actually 300 dots per inch, since each triad is 3 dots. But, forget about the 300 dots per inch. There actually is 300 dots per inch - but when you hear the term, "dots per inch" for physical screen dots, they are talking about dot triads, and therefore, for all practical purposes - there are 100 dots per inch (which is 100 dot triads per inch). Always assume that dots per inch - when referring to physical phosphor dots - refers to dot triads. BUT - beware, since more often than not, "dots per inch" is referring to pixels per inch. I warned you, it is so *%$^% confusing !!!

I won't do that to you. When I refer to the physical dots, or the phosphor dots on a shadow mask screen, I will say "Dot Triads", or if it is an Aperture Grill screen, I will say "Stripe Triads". When I refer the the screen resolution units, I will say "Pixels", and when I refer to the printing resolution, I will say "Dots".

It is amazing that every where - and I mean everywhere - the only description for the 3 color elements of a monitor, focuses on the dots in shadow mask monitors. Since no book or website I have found, so far, taks about the RGB elements with the now common aperture grill - I had to invent a term. I have seen the term, "Stripe" to indicate one colored stripe on an aperture grill monitor. Also, the term "Strip Pitch" is an accepted terem. Since the stripes come in triples, I coined the term "stripe triad", which is probably correct, even though I can't find it anywhere. It makes perfect sense, since it's cousin, the "dot triad" refers to the triplets of dots on a shadow mask monitor. If you have seen an actual official name for the three strips (Red, Green, and Blue) - drop me a line.

For brevity, only the shadow mask elements are detailed here. The following are calculations of dot triads and pixels at various common resolutions. Her screen is 12.5" across. No matter what the screen resolution is changed to, the dot triads will stay the same size - they are physically fixed by the holes in the shadow mask, fixed at the value I counted earlier - 100 triad dpi.

Screen Resolution of 640x480 :
Pixel per inch (using width) = 640 pixels/12.5 inches = 51.2 pixels per inch.
Dots per pixel (using width) = 100 dots per inch/51.2 pixels per inch = 1.95 dot triads/pixel

Screen Resolution of 800x600:
Pixel per inch (using width) = 800 pixels/12.5 inches = 64 pixels per inch.
Dots per pixel (using width) = 100 dots per inch/64 pixels per inch = 1.56 dot triads/pixel

Screen Resolution of 1024x768:
Pixel per inch (using width) = 1024 pixels/12.5 inches = 81.9 pixels per inch.
Dots per pixel (using width) = 100 dots per inch/81.92 pixels per inch = 1.22 dot triads/pixel

*** all value's are for my daughter's 17" monitor - this will vary depending on the monitor

Screen Resolution	Dot Triads per Inch	Pixels per Inch	Dot Triads per Pixel
640x480	100	51.2	1.95
800x600	100	64.0	1.56
1024x768	100	81.9	1.22

This table only applies to a 17" monitor - it is shown here to show the relationship between screen resolution, Dot Triads, and pixels. Other 17" monitors may have different densities of Dot Triads - that depends on the manufacturer. The size of the monitor will also have an effect on the number of Dot Triads per Pixel. A 21-inch monitor at 800x600 will have much more surface area per pixel, than a 17-inch monitor at 800x600.

NOTE: for decent clarity of images - manufacturers design monitors so that there is at least one Dot Triad per pixel. For this 17" monitor, if you could set the resolution to 1600x1200, there would be only .78 dot triads per pixel, which would result in grainy looking edges and poor fine detail. That is one reason why 17" monitors are never set to 1600x1200. In my opinion, even 21" monitors should not use 1600x1200 . . . it is very hard on the eyes.

Dot Triads and Strip Triads per inch are fixed - but per pixel, is a function of: screen resolution setting, dot pitch (which correlates to the density of dots), and monitor size

It seems that if each pixel has 1.56 dot triads in it - that edges and boundaries of text and images would be blurry. Apparently, the Monitor and video card is able to interpolate partial dot triads and still have a sharp picture. I am not certain, but I believe that the monitor circuitry has the ability to break the dot triads down to separate dots, and not treat it as 3 dots all the time. For example, in the large areas of blue, each dot triad, would be RGB=0,0,255 - but on a line where blue changes to red, and it occurs right down the middle of a dot triad - the 3 dots could be RGB=255,0,255 :

The example above, would be where you would have blue color on the right, and red on the left. It may even be, that the monitor never looks at dot triads, rather, works with each dot separately, so that it is not bounded by the triangular boundaries of a triad.

In addition, the pixels contain fractional triads. How could the monitor display one pixel if it is sized to equal 1.56 triads ?? It can't, of course - which means that the images do not depend on accuracy of each pixel being painted on the screen in an exact form. Algorithms interpolate the dots and the pixels, to give a fair representation on the image.

I have been unable to find info at this detailed level. I know that each manufacturer has different methodologies, most are guarded and secretive.

From the close-ups, it sure looks like the Shadow Mask would have much better clarity. But, surprisingly, today's Aperture grill monitors are very good, and very sharp. The image you see is my own monitor . . . an $1100 NEC 22" aperture grill. The slots to me, look like the detail is soft - but when I look at it in the normal viewing mode, the slots are so extremely tiny, that the picture is crisp and clear.

Shadow masks were going the way of the dinosaur, because they let much less light through - and this gives the aperture grill the advantage of brightness and vivid colors. However, recent improvements have made both monitors very competitive with one another.

Dot Pitch and Stripe Pitch - A measurement that indicates the distance between like-colored phosphor dots or stripes on a display screen. It is never measured between adjacent dots or stripes, since they always have a different color.

NOTE 1: since the two technologies are completely different, you cannot compare Dot Pitch to Strip Pitch. So, a .26 mm dot pitch monitor may be better, or worse, than a .26 mm stripe pitch monitor. If you need to compare, compare the same with same.

NOTE 2: Most people have never heard of stripe pitch. Some aperture grill manufacturers call their stripe pitch - dot pitch, simply because the public does not understand, and they feel if they confuse the public - it will hurt their sales. NEC sometimes calls it a "Mask Pitch".

For shadow mask CRT's, dot pitch used to be measured diagonally - exclusively. Then a few companies became very insidious, and realized that the horizontal dot pitch is always smaller, and began quoting that distance instead. Be careful of this very unscrupulous practice. For a standard CRT, the horizontal dot pitch equals 0.866 times the diagonal pitch, which comes to 0.2252mm horizontal measurement for a 0.26mm diagonal pitch; this calculation is derived from the geometric properties of equilateral triangles.

Aperture grill strip pitch can only be measured one way - horizontally - from one color to the next occurrence of the same color.

Measured in millimeters, pitch is one of the principal characteristics that determines the quality of display monitors. This is the most common measurement used by buyers to help them decide, although it's importance is questionable. Dot pitch typically varies from .22 mm to .28 mm (the earliest IBM monitor had a whopping .43 mm dot pitch !!! ).

NOTE: some companies will list "Vertical Dot Pitch". As you can see in the diagram, vertical is identical to diagonal dot pitch, since the like-colored dots form a perfect isosceles triangle !!

Slot Mask CRTs

You probably won't see these, but let me briefly mention them in case they get hot later. Slot mask CRTs are a new and still rare variety that has been developed to improve the brightness of CRTs using a slot rather than a round dot for the mask. Because the slots are larger than the dots on a conventional dot trio shadow mask, more of the beam reaches the phosphor, giving higher brightness.

The CromaClear CRT from NEC and the PureFlat CRT from Panasonic use slot mask systems.

The CromaClear pitch is quoted as the diagonal horizontal mask (rather than dot) pitch between adjacent dots of the same color and is typically 0.25mm. NEC quotes the horizontal dot pitch as 0.255mm.

It was found that the faster the entire screen was "painted", the less flicker humans could detect. They found that the best way, was to scan every odd line, from left to right, going from top to bottom (like reading a book), and then immediately going back up to the top, and scanning every even line. This is called "interlacing", and virtually every monitor uses interlacing. North American monitors follow the NTSC standard, and use a total of 525 scan lines, which includes the retrace lines (explained below).

Let's imagine what would happen with no interlace. Non-interlaced monitors were common 10 years ago, when it was thought that Monitors were pricey and non-interlaced was a cheap and viable alternative. Later they discovered no one could stand the flicker. With non-interlaced monitors, instead of 60 fields per second and 30 frames per second - there would be 30 fields and 30 frames per sec. So, actually, a field would be the same thing as a frame. The human eye definitely detects flicker at refresh rates lower than 60 Hz.

With interlacing, each pair of fields combines to make one frame. There is no change in the image content between the two fields !!! It is simply one still image, broken down into two sub-images . . . the first sub-image contains the odd lines, and the second sub-image contains the even numbered lines.

Line Retrace - after a line has been scanned across the screen, the electron beam needs to return back the the left side, and down two lines so that it can begin scanning the next line. This is similar to a typewriter carriage return. Of course, you cannot continue painting the screen while the beam is returning. Therefore the gun (the cathode) voltage is turned high during the retrace, so that it has no effect on the screen - it basically becomes invisible. That takes a very short time - the deflection yoke has a quick, strong signal sent into it, so that the beam returns back from right-to-left, much quicker than the time it took to scan the image line from left-to-right.

Field Retrace - once the last line of a field is traced onto the screen, the electron beam must return to the top-left portion of the screen. Since this is quite a bit farther than just returning to the left and down 2 lines - it takes more than one cycle to get back to the top. Again, the gun is turned off during this retrace so that it becomes invisible.

The following is a simplified example of interlaced scanning and retracing. For simplicity, imagine a monitor that has only 8 lines, and a full red screen image is being shown. The lines are show with exaggerated seperation so that you can see the pattern of tracing the first field, the odd one (lines 1,3,5, and 7) and the the even field (lines 2,4,6, and 8). The line retraces are not shown since they happen so quickly. The field retraces are shown as black dotted lines - but be aware that the field retrace is actually invisible, since the gun is turned off during these scans.

These are Television standards - not monitor standards, but are mentioned because PC video applications use the NTSC standards. It was always felt, that the closer one could approximate NTSC - the more pure the resulting video. PAL is the European standard, so we will concentrate on NTSC, which is also called "Broadcast TV" quality video.

PAL (Phase Alternating Line) delivers 625 lines at 25 frames a second, using 50 interlaced fields (half frames) per second half-frames per second. The aspect ratio (width:height) is 4:3. Another popular standard overseas is called "SECAM".

NTSC (National Television Standards Committee) covers a broad range of topics, but as far as television, it defines interlaced scanning at 525 lines - 30 frames a second, 60 interlaced fields a second. Actually there are 29.97 frames per sec and 59.94 fields per sec. - but almost everyone just rounds it off. The aspect ratio (width:height) is 4:3

Monitors, unlike TV's - do not follow the NTSC refresh rate of 60 fps. Instead, the monitor vertical refresh rate is adjustable, typically from 65 to 85 Hz (Hertz is cycles per sec, which is the same thing as frames per sec). They do, however, use the same 4:3 aspect ratio.

This is so, so, so confusing. You really want to read and re-read this section. There is screen resolution in display dimensions (Width x Height in pixels), there is Screen Resolution in pixels per inch, there is Print resolution, and there is image resolution. All of these are variable, and can be changed on the fly.

Dot pitch cannot be changed - that is a constant. Fortunately, you will deal with dot pitch only once . . . when you go shopping for a monitor. After that, it's pixels pixels pixels . . . and unfortunately, manufacturers list "dots per inch" for printing instead of pixels per inch - while Adobe Photoshop lists pixels per inch for printing and you need to realize that it is unrelated to the dots per inch of the printer .

For example, you can take an image and set the print size in Photoshop to 100 pixels per inch, and then when you click print, a box opens up that lists the printing quality at 300 dpi (dots per inch). In this case, the resolution of the image is indeed 100 pixels per inch, but the printing is done at 300 dots per inch. Therefore, every 9 printer dots equal one pixel (you must square the ratio, since area is the square of length). For example, - if the image is 1 inch, then the total area in pixels is 100x100 = 10,000 pixels. The total area in printing dots is 300x300 = 90,000 dots which comes out to 9 dots per pixel.

Since the actual physical dots on the screen are rarely ever referred to - when you hear the term, "dots per inch" it is usually speaking of print resolution - or it can wrongly mean pixels per inch ( ppi ) since the term is so loosely thrown around. PPI is not DPI !!

Screen Resolution is the total number of pixels - often wrongly called "dots", horizontally and vertically, that it takes to fill your screen. RESOLUTION IS NOT Dot Triads !!! Another measurement of resolution is "pixels per inch" - NOT Dot Triads PER INCH !!! Resolution is not equal to the Dot Triad we mentioned before, nor is it equal to dot pitch - both of these measurements are fixed, and are dictated by the way the shadow mask or aperature grill was made. Keep in mind that few people use pixels per inch to describe screen resolution (pixels per inch is used for image resolution), so simply use the Width and Height in pixels when referring to screen resolution.

Your video card and the software/hardware drivers that it uses, specifies 2D units on the screen called pixels, which stands for "Picture Elements (Pix els)". Pixels are square, not round. Do you recall that we calculated for my daughter's system - each pixel contained 1.56 dot triads. This is true at her current resolution of 800x600, but will change if she changes her resolution, to, say . . . 1024 x 768. Since the pixel-to-dot tirad relationship will change whenever you change screen resolution - a pixel has no direct relationship to the Dot Triads, or Dot pitch mentioned earlier.

Screen Pixels per Inch (ppi) - you can calculate this from the dimensions of your screen., but since it requires you to perform a calculation - just use the dimensions of width and height to state screen resolution.

The 72 and 96 dpi Monitor Myth - don't know where this came from - but it is out there . . . man, is it ever out there !!! Everywhere !!! Many people, manufacturers and even technical books, state that most monitors are 72 dot per inch or 96 dots per inch. In the first place, they should say pixels per inch, since that's what they "probably" mean. In the second place, pixels per inch is directly related to the screen resolution - and that varies !! Even if they mean dot triads per inch - that also varies with the manufacturer and the model.

Each pixel will always have at least one Dot Triad within it. For example, my monitor is 16" across. So, if I set the resolution to 1600x1200 (width x height) then there is 100 pixels per inch. If I go into display properties and change my resolution setting to 800x600, I now have 800 pixels spanning my 16 inch monitor width, or 50 pixels per inch. Notice I had to calculate pixels per inch - and that s why no one uses it except for images where the software calculates it for you. So, use W x H to describe your screen resolution - not pixels per inch.

As a rule of thumb, here are the best resolution settings for various size monitors. Stay away from 1600x1200, it is just too hard to read - even with a 22" monitor, the desktop icons and text becomes microscopic. Also, try to stay away from 640x480, since the standard for web pages is 800x600, and the lower setting will force you to scroll a lot. Unfortunately, with a 13" monitor, it will be too hard to read at 800x600, so you are forced to use that resolution and deal with the scroll bars on the web :

Diagonal Monitor Size	Resolution setting (W x H)
13"	640x480
15"	800x600
17"	800x600
19"	1024x768
21"	1024x768 or 1152x864

Resolution can be easily changed by configuring your display settings - but dot pitch is fixed.

Yet another cool Exercise with Close-up Screen Photography - In the case of my daughter's shadow mask monitor, I created a tiny image, magnified here, and took some screenshots of it, to show the effect of 1 pixel having 1.56 dot triads in it. By making a tiny image with exact R, G, and B colored suqares in it, we can see what the monitor does to approximate and interpolate the edges, since the do not fall on exact borders (1 pixel = 1.56 dot triads).

First, I took the following two tiny images, the first one is 6x6 pixels and the other is 12x12 pixels :

Notice that you cannot make it out at all with the 6x6 image, but you can see color variation with the 12x12 (just barely). Both images have the following color pattern - the following is a 98x98 pixel image just to allow you to see what the actual patterns are. The 6x6 tiny has one pixel for each square. The 12x12 has 4 pixels (2x2) for each square. :

The following is a close-up snapshot of the 6x6 (1 pixel per colored square) on a Shadow Mask monitor at 800x600 resolution. Note that each square contains 1 pixel, and therefore is made up of 1.56 dot triads - you can see the difficulty the monitor has in representing this, and actually - it cannot represent the image satisfactorily.:

This next picture is a close-up snapshot of the 12x12 (4 pixel, 2x2, per colored square). Note that each square contains 4 pixel, 2x2, which is 4x1.56 = 6.24 dot triads - you can see the image is represented much more accurately, although the edges still have to be interpolated, since the pixels do not fall on the exact edges of the dot triad borders :

Here is the same 12x12 pixel image (each color square has 2x2 pixels in it), with a close-up snapshot taken on an aperture grill monitor. It looks as if there are 3 stripes for each color square. Since each square is 2 pixels wide, the ratio is 1.5 stripes per pixel :

NOTE: if you stand back 10-15 feet from the screen, these images take shape.

These images show that pixels are square - not round, and that the monitor is required to interpolate dot color values at borders where the colors change. It also show that pixels and dots do not have an exact relationship. If the pixels always contained either 1 dot triad, 4 dot triads, 8 dot triads, etc. - then the edges would be sharper, but it is doubtful that the human eye could notice the difference.

As mentioned, dot triads are constant, and fixed by the manufacturing process or the monitor. Pixel sizes change with the screen resolution. Here is a 6x6 white image, with close-up snapshots taken at 3 different screen resolution. Notice that the size if the dots are basically unchanged (the 640x480 image looks as if the dots are smaller, but the snapshot simply came out much more clear than the other two, which blurred when I snapped them) The size of the 6x6 image becomes larger when the screen resolution is decreased. Also, the lower the resolution, the more dots per pixel :

36 Pixels at 640x480 36 Pixels at 800x600 36 Pixels at 1024x768
70 dot triads 56 dot triads 44 dot triads

I used the dot triads per pixel table that I had calculated earlier for her monitor, so that I could list how many dot triads appear in each 6x6 pixel image. Now for the aperture grill :

36 Pixels at 640x480 36 Pixels at 800x600 36 Pixels at 1024x768
14 slats - 3 strips per slat 11 slats - 3 strips per slat 8 slats - 3 strips per slat

NOTE: it would seem that the aperture grill has much less in the way of separate entities. For example, the 640x480 snapshot shows 70 dot triads, but only 14 slats. But remember, the dot triads multiply themselves in the vertical direction - whereas one slat goes all the way from top to bottom. Nevertheless, the colors change along each strip, as you work your way down from the top so the same amount of color changes occurs.

Screen resolution is adjustable, and all monitors have numerous resolutions that they can be configured for. It is actually the video card and display driver settings that determines the display resolution. The dimensions of the dot triads stay the same, and only the dimensions of the pixels change, when you change the screen resolution. The higher screen resolutions have smaller pixels, and therefore the screen components become very small. Take a look at how your screen resolution setting affects pixel size, as far as how many Dot Triads are used per pixel :

Images have no pixels per inch resolution of their own (that is dependent upon screen resolution) - all they have is dimensions in pixels (Width x Height).

Imaging software, such as Adobe Photoshop, does not care what resolution your screen is set to, and can set the image dimensions to anything. It can change the resolution of any image but will never alter your screen resolution.

You will need to compare the width and height of your image, especially if you are resizing it - against the dimensions of your screen. It is the only way you can tell how large the image will appear on your screen - or someone else's in case you are emailing a pic. A good standard size to use for pics that you want to send to someone, is 400x400 or in that neighborhood..

On the screen , the image always adopts the pixels per inch that your display is set to ! ! !

The image print resolution in ppi (pixels per inch), and the printer's actual mechanical dpi (dots per inch) - is preset to a default optimal value, but can be changed. Some software imaging packages call the internal image resolution dpi instead of ppi, so be aware of that - it is confusing if they do that, though, because you will then have two different dpi values to consider.

Typical printers have two primary options - usually 300 dpi, or 600 dpi, although it varies . . . for example, my old Canon BJC4300 Bubble-Jet used 360 dpi as the default setting.. 300 dpi is very detailed, and there really is no reason to use 600 dpi. Typically even if you set the printer to 600 dpi, the image itself does not have a high enough resolution to feed that much info to the printer.

The printer's setting will print that many dots per inch, regardless of the image print resolution. Remember that you cannot change the resolution of the image itself - only the dimensions - but some applications, such as Photoshop, allow you to alter the image's print resolution and unfortunately they confusingly list it as pixels per inch - which has no relationship to the pixels per inch that you see on your screen when you look at the image.

Applications, such as Photoshop, embed info within an image file, to tell the printer what resolution in dpi (dots per inch) to print it at. It is not clear if other applications can pick up on this info - but I took the following two images, and printed them in both ACDSEE, adn in Netscape, and they printed out the same size. But when I printed them out in Photoshop, the 300 dpi image printed very small. Right-click and download these files if you would like to try it yourself. If you open these files in Photoshop, and click Image/Image Size . . . you will see that the dimensions are both 200x200, but that under the "Print Size", the resolution in Pixels per inch is 72 and 300 respectively. Obviously, since they both display on the screen as the same size, the displayed pixels per inch are identical.

But again, image files have no display pixels-per-inch information. They only contain dimensions of width and height, and print pixels per inch. If you view these two images on a monitor set to 640x480, they will be very large, whereas if you change the display properties to 1024x768 they will be smaller.

200x200 72 print Pixels/Inch 200x200 300 print Pixels/Inch
File Size = 15 kB File Size = 15 kB

You can see that the two images are identical on the screen, even though their ppi varies greatly - that is because this particular ppi is only info for the printer. Fell free to download these and see for yourself. If you open them in photoshop, they will print at two vastly different sizes. I printed both, tore off the excess paper, and took a snapshot - as you can see, the printer does care about the dpi setting :

Now here is where it gets even more confusing. The main benefit to changing the ppi print resolution of an image, is so that you can alter the size on the printout. For example, you have a standard image that you have downloaded from the web (and almost all images on the web have a print resolution of 72 pixels per inch). Then you print it with the normal Printer Resolution set to 300 dpi. Well, the printer does indeed print at 300 dpi, but the image is sent to the printer at 72 ppi, so it merely prints several identical dots. The actual resolution of the image on the paper is 72 ppi - even though there are 300 dots per inch that were actually printed.

Now, lets say that it comes out taking up a 3" x 3" square on the paper, and you want it to print as a 1.5" x 1.5" square. You can then double the print resolution to 144, without resampling the image (uncheck the resample image box in Photoshop). When you do that, Photoshop will now list the Print size of the image as 1.5" by 1.5". Finally when you print it - the printer once again prints 300 dpi, but the actual detail of the image is at 144 ppi.

Let us take a simple example of a very tiny checkerboard image (2x2 pixels) - magnified here :

Assume the Printer's default mechanical print dpi is at 300 dpi, and the image in Photoshop is set to 150 pixels per inch in the Print option box under Image/Image Size . . . It is obvious that the image will be sent to the printer, with detail of 150 pixels per inch. The printer will indeed print 300 dpi, but will simply double the same dot twice on each print line across each time, since it is only receiving info at a rate of 150 ppi. The image below shows the print dots as small white squares :

Now, change the print resolution within Photoshop to 300 ppi, to match the actual printers dpi. Make sure you uncheck the "Resample Image" box. On the screen, you will now see the same image, since the display only shows image dimensions, not pixels per inch. However, the printer will now receive info at the rate of 300 dpi, and therefore, does not have to double-up dots :

Note that the quality of the final image is identical. The printer could not increase the quality of an image that is already set, at 2x2 pixels. However, if the print resolution within photoshop is set too high, the consequences can be drastic. Let's take the same 2x2 image and set the print resolution to 600 pixels per inch. When you send the job to the printer, the first two squares (2 pixels in this case, since the image is tiny at 2x2 pixels) will be sent at a rate of 600 pixels per inch. The printer can only print at 300 dpi, so it must take the first two pixels of both rows and "average" them together, printing out one dot for every 4 dots sent. Not good !!

NOTE 1 : smaller simplistic applications such as viewers and even web browsers - do not have the ability to adjust the print resolution, and they simply accept the default resolution of the screen - not the printer. For example, if you have a 450x450 pixel image, and open it in a simplistic viewer such as ACDSEE, and then print it - it will send the info to the printer, pixel-by-pixel, and the printer will print at its default of, perhaps 300 dpi - but the image will print on the paper at 72 pixels per inch - if that's what your display is currently set to. The image viewers try to print at a size which approximates the size you see on the screen. This is known as WYSIWYG (What You See Is What You Get). Since the print job is sent to the printer at 72 dpi, and the printer is actually printing at 300 dpi - it will print repeats as explained above.

NOTE 2 : high quality print jobs require high resolution images. 150 ppi is usually the best setting for high quality images, because 300 ppi images will result in huge file sizes. Try a comparison and print a job at 150 ppi and then again at 300 ppi. You will be hard pressed to notice any difference at all.

With professional imaging software - you always want the print resolution setting in your software application to be equal or lower than the printer's dpi resolution. In fact, if you have a lower resolution within your application, it is best to make it even divisible into the printer's resolution, so that the rate it repeats identical dots is constant, and that way the printer never has to average dots together. For example, if your printer setting is 300 dpi, try to set your high-quality image print resolutions to 300 ppi, or 150 ppi. When you get to the lower values, you don't need it to be exactly divisible. So you can go ahead and print a 72 ppi image and it should look just as good, as if you had use the exact divisible amount of 75 ppi. If you send a print job to the printer at 250 ppi and the printer is set to print at 300 dpi, then it will start to print between dots of the original image, and will have to interpolate which degrades the quality slightly.

Notice that I have repeatedly said to uncheck "Resample Image" when you change the ppi print resolution within Photoshop. This is true for printing, because the print size can only be adjusted this way.

The Golden Rule - never resample up (i.e. larger) unless you absolutely have to, always resample down. Resampling up increases blur - always increases blur. Never resample more than once - always save your original large image. If you resample, and it is not what you want, undo the resample and then redo it. If you have already saved the resampled file - delete it and start with your original again.

But what if you truly want to change the dimensions of the image so that it appears on the monitor at a different size? To do this you need to perform "resampling", which is a fancy word for resizing. Again using Photoshop as the example application. you simply select image/image size . . . and type in different pixel dimensions, and click OK. Ordinarily you will leave two boxes checked : Constrain Proportions, and Resample Image. When you constrain the proportions, it keep the ratio of width and height the same, which will result in a truer resize, and a clearer result.

The algorithm used to resize images actually is done by "resampling". Each pixel is sampled, and combined with it's neighboring pixels. If you take a 200x200 image and resample it down to 100x100, the algorithm is simple . . . average every tow pixels into one. If you take a 200x200 image and resample it down to 170x170, then the algorithm has to average and interpolate in a more complex fashion - the results are not as sharp either.

When resampling images, try to cut them in half, or a fourth, or an eighth. Of course this is not always possible but it does give the sharpest resulting images, due to the simplicity of the algorithm. If you must cut an image down by an odd amount, such as 87% of it's original size, keep in mind that this usually works fine with large images, since there are so many pixels to work with. For example, reducing a 1000x1000 pixel image to 870x870 will look very good. But if you resample a 100x100 image to 87x87 there will be a bit of degradation in image quality.