"Chuck Snyder" wrote in message
[…] The $64 question is: is the dpi
of a printer counting the individual dots of color in an inch of print, or is it somehow a ‘pixel equivalent’ number where the actual number of dots per inch is much larger?
Printer resolution is reported in the maximum number of dots printable, and does not take into account the dithering required to print 24-bit color on what is effectively a 4- to 8-bit device (depending on how many colors of ink the printer uses).
*However*, just because you lose effective resolution due to the dithering (just as you would when using a halftone screen in a conventional printing process), that doesn’t mean that providing the printer with as high a resolution image as possible isn’t useful. There are actually a variety of dithering techniques in use (one common one being "error diffusion"), but all of them can benefit from additional image data.
The reason comes down again to the question of where the resampling is done. The printer driver, being intimately familiar with the hardware, can do THE most effective job of resampling in order to produce an image that most closely matches the original digital data. It knows the exact shades of each of the inks being used, and it knows how it to blend those shades just so to match the original image. It can also adjust the dithering so as to preserve apparent lines in the image, helping to reduce the appearance of aliasing in the printed image.
Let’s take an extreme example and look at it. Consider a 300 dpi B&W laser printer using half-toning (which is essentially just another kind of dithering). Let’s say you choose a half-tone screen that drops the effective resolution down to about 50 dpi (I guess we want a lot of gray-scale resolution 🙂 ). Now, print an image that has a single diagonal line on it, first at 300 dpi, and then at 50 dpi.
You *will* be able to see a noticeable difference between the two printouts. While the effective resolution for both is still just 50 dpi, you will find that the image printed at 300 dpi still has less "jaggies" than the one printed at 50 dpi. Yes, the "grain" of the half-tone screen will make the edge of the line a little fuzzier, but you’ll get more of an anti-aliased effect rather than a hard edge at the edge of the line.
Another way to think about it is to realize that there’s "real" resolution and "effective" resolution. "Real" resolution is how many dots that can be reproduced, and "effective" is how much detail can actually be reproduced. What "effective" is describing is, what is the smallest "feature" that you would be able to resolve at that resolution? Sort of like the resolution of spy satellites, when they say they can resolve down to 10 meters or 1 meter or whatever. But with a spy satellite, just because you can only resolve down to 10 meters, that doesn’t mean that a car only 5 meters long won’t show up. It just means that you won’t be able to tell exactly that it’s a car.
Same thing with printed output. At the "effective" resolution, you’ll find the limits of being able to identify a feature in the image clearly. But that doesn’t mean that there won’t be any additional detail smaller than that. It just means it will be harder to tell what that additional detail is.
Finally, as far as "the ones that come free with a computer", I think you’ll find that, other than the maximum resolution the printer is capable of (the "free" ones are usually down in the 720dpi range, while the expensive ones are up in the 2880dpi range), they are using substantially the same software to run the printer, and thus would take the same advantage in the dithering algorithms that the more expensive printers do.
In fact, the drivers that came with my Epson 2200 appear to be used for all (or almost all) of the current Epson line, and likewise the HP drivers I downloaded a few weeks ago just to play with also appear to cover an entire HP product line (their "photo" printers).
So I would still send as high a resolution image as possible to the less expensive printers. Obviously you’ll reach a point of diminishing returns. The driver can only do so much, and can’t create resolution in the printer where none exists. I doubt sending a 2880 dpi image to a 720 dpi printer is going to be much better than sending a 720 or 1440 dpi image, if at all. But up to and including the actual resolution of the printer, more dots are better.
If all one is trying to do is print a picture at a particular size on the printer, I do not think it is EVER useful to resample the picture in Elements or a similar editing program. All such programs allow you to specify the final printed size without changing the underlying pixel data, and that’s the right way to print a specific size.
All IMHO, of course.
Pete