Silverace - dottyspotty
Issue 3: Enlarged Inkjet Dots Samples

11 April 1999

In this issue, you can see for yourself, and probably for the very first time, the actual dots placed onto paper by the various popular inkjet printers in use today.  This will let you see exactly what people are talking about when they compare dot sizes between competing printers, and let you guage the quality of inkjet prints on a more objective basis. These images will help you choose the very best printer(s) for your needs.

These images are included in my Printer Sample Output Chart available for download - 300KB.

Given my own limited means (likewise for most of you as well), I would like to say immediately that these are not the very best samples that can be produced.  In fact, they are not even very good samples in my own opinion!  Instead, these are merely good/average samples produced using the limited equipment and print samples I've had access to.  If I were to do this Printer Output Sample Chart properly, I would certainly introduce the following:

The actual printers at hand.
Identical images for each print sample.
Top quality manufacturer inkjet papers.
Identical plain papers.
Top quality microscope with high-resolution color video camera and color controlled lighting.
High-resolution scale/ruler marked in millimeters.

Instead, I've had to use a poor man's approach which includes:
30x Radio Shack Micronta pocket microscope.
Sony TRV-22 Video8 camcorder.
ATI All-In-Wonder capture card (capturing images at 640x480).

The setup I used was as follows:
These samples were taken from actual print samples generated by these printers over the years I've collected them.  When mentioned, the inkjet or photo paper used are those specified by the manufacturer for that printer. The Sony camcoder was placed on a short tripod and placed on a desk.  The desk was illuminated from sunlight passing through the window next to it.  The sunlight came though the window at a low angle (30-45 degrees above horizon) and illuminated the print samples placed on the desk.  The camcorder was angled to face downwards, and the microscope was placed between the print sample and the camcorder lens.  The camcorder lens was zoomed to allow the lens to capture what it saw through the microscope and left at that zoom level.  The tripod was left at the same height.  The microscope was moved away as necessary to swap print samples.  When a new print sample was placed under the camcorder, the microscope was placed on the paper and refocused and moved so that an image appeared in the camcorder.
Due to the extremely limited depth-of-field in the microscope combined with the limited depth-of-field in the camcorder zoomed lens, it was often not possible to achieve and maintain crisp, sharp focus during the duration of the video capture.  However, these samples were taken from the many attempts to capture a sharp image, and in my opinion, very similar to what you would see yourself had you peered through my microscope.
All images were transfered through a Moster Cable composite RCA cable into an ATI All-In-Wonder capture card in a PC at a resolution of 640x480 in 24-bit BMP files.  Then, the images were resized by 1/3rd to their current size, a few had their brightness and contrast adjusted to make them clearer to see, and all were stored as seperate new BMP files.  The final images were combined into one chart and stored as a TIFF image, which was then used to create the JPEG Printer Output Sample Chart you have now.
The compression ratio for the JPEG image was carefully adjusted until the usual JPEG compression artifacts were of no concern or significant note in the samples images when compared to the master TIFF original.  At this point, a number of the samples in the JPEG chart were compared once again to the orignal images as viewed by myself under the microscope.
I feel that there was no significant artifacts or differences between the images in the JPEG chart and what you would see under the microscope with the exception of the noted increased sharpness seen during actual viewing.  Also, given the wide dynamic range of the human eye, the contrast in these images are not as great as what you would actually see.  You always see a wider range of colors, brightness and detail in real-life vs. digital images. Thus, minor details such as paper grain/fibers wash out when the brightness and contrast are adjusted such that the dots are nicely visible and clear in my output chart, and are often invisible even in the imager of the camcorder itself.

When you compare these samples to those of others taken by other means, usually drum or flat bed scanning, you'll note that these samples are superior in quality and resolution.

Let me explain.  If you have a printer that outputs at 1200x1200 dpi and a scanner that scans at 1200 dpi, you can guess that one dot on a print sample will appear as one pixel in the scanned image.  As you all well know, one dot, the size on a period on the screen, doesn't say much at all about the quality, size, or characteristics of the printed dot at all.  Also, due to the limited scanning resolution, you won't pickup on other important details of inkjet prints such as dot spray, bleeding, and so forth as they are usually even smaller defects.

In fact, even if you have a 5000 dpi drum scanner, you will only capture around 3.5-8.3 pixels (1440 to 600 dpi) per printed dot!

Clearly, these other means of capturing dot samples are inferior and usually produce fuzzy, blurred samples from which little of note can be said between competing printers.

Happily, the method I've used captures quite a bit more and allows, even, for the reduction of samples.

I would expect that at the minimum, the camcorder used would capture a paltry 200 lines of resolution.  Nevertheless, dots would still occupy dozens of pixels across the image, and when reduced by 1/3rd these dots are still a dozen pixels large.  More than double the resolution of the drum scanned samples at minimum, and actually far higher.  You can even make out defects such as dot spray quite easily in these samples.

The following are notes on the various samples in the chart.

Ruler - A steel ruler ruled to 1/100th of an inch and used for graphic design.  This provides a clear guage of the actual dot sizes you see.  Very high quality ruler, but unfortunately, not in metric.  As you can see, inkjet dots are quite smaller than the spacing of this ruler (with the ol' HP Paintjet coming quite close with its huge dots; the 'first' color inkjet printer that you could actually afford for home and business use; my first start as well).

Alps 1300 - This is output from a thermal sub-dye printer.  It fuses solid inks into the paper, mixing them like hot wax.  As a result, there are no visible 'dots' to speak of, and only the slight banding between different lines of the output.  However, given the broader dynamic range of the colors of a sub-dye printer, and this blending effect, the output is effectively identical to photo quality at any distance to the human eye.  Along with the Alps 5000, these two printers are the very best under $500 USD home photo printers that you can buy today.

Commercial 4-color press - These samples are from the cover of a high-quality Japanese magazine cover.  It is representative of the type of dots you can expect when viewing most commercial publications of the same nature.  As you can see, the lpi (lines per inch) are not very high at all and far less than that of an inkjet printer.  However, the reason commercial publications look so good is because of several factors:

1) Variable dot sizes that go from huge to tiny.  As you can see from these samples, the variations allowed in commercial printing effectively covers everything from 100% to 0% coverage, with dots that vary to cover.  This wide range allows the commercial press to beat inkjet printers easily for most output today.  The smallest dots of a commercial press are several times smaller than even the smallest dots that can be produced by an inkjet printer today.
2) Broader gamut.  Commercial 4-color presses can produce output with a broader coverage of the color gamut than an inkjet printer in most cases.  Add to that spot colors and 6-color (Hexachrome) and you'll have even better output than any inkjet today.
3) Non-water based inks.  Don't use water and you don't get the bleeding effects you see when wet inkjet inks bleed into the cotton fibers of the paper.

Other factors contribute as well.

Canon BJC-7000 - Note how the combination of droplette modulation (variable dot size) and the use of seven inks (photo inks), allow for the production of subtle changes along the line.  (see my second article for background information and references to patent articles)

Anything on plain paper - Note how the bleeding of inks into the cotton fibers effectively destroys any semblance of a round dot.  Plain paper has to be the worst enemy of color inkjets today due to the use of water/alcohol-based inks in all color inkjet printers.

Epson on photo/inkjet paper - Note the exceedingly small, crisp dots produced by the Epson printers, espeically the latest models, on their best paper.  This and their higher resolution results in the superior quality of Epson photo prints vs. other inkjet printers.

HP Photosmart - While the red/orange samples may leave you wondering what is going on, the blue/white sample provides the clue.  The HP Photosmart uses a combination of 'tricks' to improve photo image quality.  First, the photo paper allows the inks to blend together, similar in effect to a degree, like a sub-dye printer.  This reduces the harshness of visible dots on the output.  Also, their dots are teardrop shaped at times and this creates the illusion of a better blending of dots along the length (vs. normal round dots).

HP 7xx/8xx/11xx/2xxx - The latest RET II (resolution enhancement technology) HP printers produce dots of similar shape, but with only 4-inks vs. the six in the Photosmart and possible greater dots sizes (16 in the RET II, unknown in the Photosmart).  As a result the output of these printers are similar to the Photosmart printers, but of lower quality as can be expect with the use of only 4-color inks.

Text output - As annoying as it may sound, some print drivers do not use black ink when printing black areas on photo/inkjet paper.  A clear example of this are the HP 7xx/8xx/11xx/2xxx printers.  I have a set of print samples of the same image printed on plain, inkjet and photo papers by the HP 7xx series.  If you take a closer look into the photo text of the HPs, you'll see that they're created from a blend of all the colors!  Thus, if you scratch your head and wonder why the HP photo paper text never looks as good as their inkjet paper text, or those of other printers, this is the reason.  The Canon 7000 has got to be the worst offender as it tries to lay down multiple smaller & lighter dots outside of the character itself in an attempt to create the illusion of sharper text.  This is not the same as printing gray or colored text with colored dots, as the Epson 600 samples shows, since then, you'll need to mix colored inks to achieve that effect.
  If you look at the Alps text, you'll note that the printer is so good at imaging the dots on their best paper that even the dots themselves are crisply defined!  Good or not, you do also see sharp staircase transitions between dots as a result.
  Many of these higher-quality printer texts will be similar to the output from a laser printer on photo paper.  On plain paper, bleeding and dot spray artifacts occur commonly.

Hopefully, you've had a good time looking at these samples and can see how different printers produce better output than others.  If nothing else, the output chart would make an interesting wall chart for the office.

I'm always looking for high-quality print samples myself to include in this chart and future articles, so if you have any you want to send my way, let me know.