johnson@horta.rice.edu (Bruce Johnson) (12/05/90)
Our local collection of research groups were in the market for a color printer that could provide hardcopy for a bunch of SGI Personal Irises, and color postscript naturally presented itself as a way to go. Judging from the reviews of the Model 10, and the current cost of (the otherwise preferable) dye-sublimation printers, I started trying to find a solution in the QMS direction. Other ps- color-printers are also available now, I can't comment on differences. However, I'll pass along our final configurationn for what it's worth. We purchased a QMS and connected it to a color Mac running A/UX 2.0. The Mac is a IIci with a 200 MByte disk and an ethernet connection. This way we are able to spool files via lpr through the IIci. It costs a few thousand more than the big hard disk PrintLink UNIX-ether connection that can be bought for the Model 10, but we also have the capability of using Mac applications for drawing, painting, retouching, etc. For people interested in such hard copy from UNIX workstations, one of the pitfalls is the sparsity of color-ps drivers. Several expensive packages produce the vector (procedural) color code, but the typical bitmap -> color ps conversion routines produce large files which take a long time to print and ultimately produce ditherings on the Model 10 resembling paintjet quality, a waste of the 300 dpi resolution. I saw last week that Seth Teller at Berkeley had written a feedback-driver for certain SGI machines (not ours, alas) which converts 3D IRIS graphics into procedural ps code. Something this sophisticated is necessary to get the full value out of screendumps even from megapixel displays. For typical Mac (and presumably pc) applications, the QMS seems to be nicely set up. Connects directly to Appletalk, for instance. Takes about 60 seconds to process an image (reported to be much slower over serial and parallel ports). The colors can turn out nice (good saturated colors), especially on transparency. Have to use special paper, otherwise things turn out blotchy. Cost per copy is quoted as about $.67 per paper copy, $1.80 per transparency. Since the basic unit of special paper comes in quantity of 2000, I expect most of that $.67 is the amortized cost of the ribbon from which the wax is transferred. I checked into the possibility of cheaper transparencies and was told to use at least 3 mils thick (e.g. from 3M). Apparently the bottom line is that one could save at least half of the price on transparencies, but perhaps risking violating the warranty should something untoward happen. Maybe someone else will do experiments(?). Oh, yes, another word of caution for all thermal transfer printers -- avoid big dark areas in your images unless you plan to display them on an overhead projector well-shielded in the infrared. A story at a Gordon conference was that the guy ended up having about 60 seconds per image before his colors started melting together (probably an extreme case). Still haven't put this thing through it's proper paces yet, so I don't have an exhaustive report. I'll be interested myself in other people's opinions. -- Bruce Johnson Chemistry Dept, Rice Univ johnson@horta.rice.edu