I've been working on improvements to the print plugin, specifically for the Epson Stylus Photo EX printer. This printer is capable of 1440 dpi, using 6 colors (CMYK + light C + light M). It's one generation behind Epson's current top of the line (the Stylus Photo 750 and 1200). The current print plugin supports a somewhat generic Epson printer model, which is based on just CMYK. My updates don't support all of the printer's features, most notably 1440 dpi and possibly super micro dots (I'm not clear on exactly what the printer's capabilities are in that area). Some of the stuff, such as 1440 dpi, are not publicly available and I'm still working on getting someone to print something to a file to hopefully allow me to reverse engineer this stuff. In any event, I can now get high quality 720x720 dpi prints. I wouldn't quite call them photo quality; this printer is not really capable of that, at least at 720 dpi using techniques I'm aware of (the 1200 comes a lot closer). I've put a gamma control in the dialog, but it's not really working correctly and I haven't had time to learn gtk to fix it up. The other structural change I've made is to make it capable of processing 16 bits/color in the back end. I've found that this significantly improves quality in the very light tonal ranges; it eliminates a lot of quantization caused by range compression in that region. The implementation is ad hoc right now, because I haven't had time to fix it up. I'd like anyone who cares to to play with it some, make suggestions, rewrite code, and so forth. Presumably this isn't going to make the freeze. Projects I have left here: 1) Improve the color balance -- right now it may be a bit weak in the magenta. It's really hard to tell for certain. It looks like it's a bit weak there compared to my screen display and the slide I scanned for test purposes, but there are so many places that things could break down that I just don't know. It may in fact be just fine, or a matter of opinion. A simple test pattern I used was quite good, although rather different from the same test pattern printed on Windows on a 1200 (it's arguable which one's "better"). That *might* just be a matter of getting the gamma right, but I suspect there's more to it than that. I'm not a color expert! 2) Reverse engineer 1440 dpi (the original author cannot provide me this info for NDA reasons). Likewise any smaller dots that might be available. Also, any improved use of microweave (currently it can't use the entire print head). 3) Figure out a good transfer curve, preferably in the form of a gamma setting. If that's not good enough, figure out what else I need (per-channel curves?). 4) Run more tests. 5) Clean up the code, particularly as regards the 16-bit stuff, and integrating it with the PostScript and PCL drivers. 6) Get the positioning of the image on the page correct (right now things are very badly wrong). 7) Someone with a 750 or 1200 want to take a crack at it? I have pointers to Epson's level 1 developer documentation (the less exciting stuff) that I can offer. 8) Merge this stuff into gs/CUPS??? As of Friday, a lot of my stuff looked better than Stylus Color output from Windows, although there were some glitches. I think I've basically fixed all the nasty glitches. The code's at http://www.tiac.net/users/rlk/print.tar.gz (there's no link to it from my home page). -- Robert Krawitz <rlk@xxxxxxxxxxxx> http://www.tiac.net/users/rlk/ Tall Clubs International -- http://www.tall.org/ or 1-888-IM-TALL-2 Member of the League for Programming Freedom -- mail lpf@xxxxxxxxxxxx "Linux doesn't dictate how I work, I dictate how Linux works." --Eric Crampton
