|
REVIEW OF THE EPSON R800 PRINTER John Hart Program in Atmospheric and Oceanic Sciences University of Colorado Boulder, CO 80302 nimbus.colorado.edu/hart/science.htm May 16, 2004
In the production of "stereo cards", it is necessary (or desirable) to print high resolution images on small size paper sheets. One typically constructs a stereo pair by printing two 3.5" wide images ("Left" and "Right") and mounting them on a stiff "card". Such cards are viewed with a lens viewer that provides a modest amount of magnification. Ink-Jet prints have previously been considered inferior in this application because the resolution on the print was not sufficient to provide adequate detail across a 3.5 inch print. In addition, ink-jet printers work by laying down a splatter-pattern of dots to generate colors. Thus a 1440 dpi printer makes shades by variably spacing tiny color dots at a maximum density of 1440 dots per inch. For example, the blue dots have the closest spacing for dark blue, wider for light blue, etc. Unfortunately the dots, if too large, become visible under magnification. In 2004 Epson introduced its R800 photo-printer. This may change the impression that ink-jet printers are not good for producing high-resolution small-size prints. It features an astounding 5776 x 1440 resolution (dpi) and a 1.5 picoliter droplet size. These numbers may be compared with an ink-jet printer of a year ago, the Epson C80, which has 2880 x 720 resolution and "3+" picoliter drop size. Can a factor of 2 in these specifications make much difference? ABSOLUTELY. In addition, the R800 features 8-color archival-ink printing.
The R800 Printer and a colorful 8.5 x 11 of a crystal (photo-micrograph).
8-Color (actually 7 + glossy overcoat) print cartridge (archival ink). RESOLUTION TESTS: We digitized an Edmund Scientific resolution test pattern (UAF 1951) using an Epson 2450 flatbed scanner. The digitization process creates a 2400 pixel per inch file that resolves out to about 12.7 lines per millimeter (lpm). The original chart goes out to 14.3 lpm. We then printed smaller sections of the test pattern on the R800 and on the C80. In both cases premium photo quality Epson paper was used and the print heads were cleaned prior to the tests. As a qualitative test, using a 135mm Pentax lens as a high quality magnifier, I could resolve about 10.1 lpm while looking at the original chart itself. When looking at the R800 print, I could see detail in the 8 lpm block. For the C80, I saw spacing remaining in the 5 lpm resolution block. OK, the R800 claims twice the dpi, and seems to retain detail in half the line spacing. However, the R800 seems to be approaching the limit of human vision at a magnification typical typical of stereo viewing. Using a standard plastic-lens print viewer, the viewing angle is about 30 degrees (3.5" wide print at a distance of about 7 inches). Looking at the original Edmunds chart, I can see structure out to 5 lpm. But the R800 holds detail to 8 lpm, and has quite good contrast at 6 lpm. The printed images were studied under a microscope, and one pair is shown below. Here the R800 resolved about 9 lpm (3-2), while the C80 resolves about 6 lpm (2-5). The point of this image is to illustrate that there is more to it than resolution alone. For reference, block 2-4 is 5.6 lpm.
Microscopic view of print output (4X objective). Left - C80, Right - R800. For reference, block 2-4 is 5.6 lpm. DOT GRAIN Ink-jet prints look grainy, reflecting the dot splatter patterns. In a way their images are reminiscent of film, with the associated film grain. The above resolution tests show how much more ragged (or "noisy") the C80 image looks. Another comparison of the rendering of a pure blue sky from the R800 and the C80 is shown below.
C80 Sky (L) vs. R800 Sky (R). Reflected light photo-micrographs of prints made at highest quality on Epson Premium Glossy Photopaper. The C80 claims twice the volume in its ink droplets ("3+ picoliters, vs. 1.5 picoliter for R800). If the spots are the same thickness, the C80 dots should be about 42% larger in diameter. They look much more prominent above, suggesting less momentum and smaller drop spread on impact for the R800. DETAIL The combination of higher resolution and lower dot-grain leads to substantially better print quality in the R800. Below we compare some very tiny blades of grass in a field of vegetation. The photomicrographs show a section of the print about 1.4mm across.
Photo-micrograph C80 plant (L) vs R800 plant (R). Image is about 1.4mm across. DRIVING THE R800 The downside of having a printer with as much resolving power as the R800 is that you must drive it with a high-resolution data stream in order to obtain every last bit of detail. We repeated the rez-chart print tests by feeding the R800 with different data streams. Not surprisingly, a 300 ppi (pixels per inch) feed only leads to 5 lpm in the print, 600 ppi gives ~ 8 lpmm, while 1200 ppi and higher are printer limited (rather than input limited) to about 10 lpmm. When making optimal 3" prints, then, you need to drive the printer with 1000ppi (say). This requires a source of 3000 x 3000, or 9 Mpixels. Since the print resolution we are talking about does not have 100% contrast at the finest ends of resolution spectrum (see samples above), this feed density can be obtained by first rezzing up a slightly lower resolution image. A 5MP camera would appear to be just about ideal for making 3.5" prints on the R800. Many commercial photo labs "prefer" 300dpi files for printing on such machines as the Fuji Frontier. Some will not accept higher resolution files, or will reformat higher ppi files down to this level before making a pint. This is OK for large prints, but is undersirable for small ones to be viewed under magnification. Clearly one advantage of individual printing is that the photographer is in control the whole process. The R800 appears very well suited to making superior quality small prints. The resolution, color range fidelity, nearly invisible dot pattern, archival inks, and relatively high speed (full color 8.5 x 11 in ~8 minutes), are impressive.
|
