Gimp stuff, long
2002-12-16 by Tyler Boley <tyler@tylerboley.com>
Some thoughts on the GimpPrint drivers, another bothersome cross post. There is a quad ink option built into the driver, it does in fact partition the inks properly for the common quad ink sets available. It prints too light, but it works quite well with the following adjustments. The Photo Quality Glossy Paper seemed to give the deepest black, but some others were so close it may have been instrument tolerances. In the advanced settings you must max the density setting to 2. Other options seemed to either create problems or not help. Unfortunately the intriguing settings for individual C, M, & Y densities seems to be disabled when the quad ink option is selected. If they were enabled, this thing could work much better. The different dither options created some problems or artifacting in parts of the scale, the one that worked well was "ordered". Maxing the density brings the Dmax up to workable and brings a middle gray more in line. Unfortunately the (otherwise correct) quad partitioning still doesn't bring the light inks in hard enough and shadows get too dark, but there are no flat spots or non-linearities. A custom gray scale dot gain curve was made and used as a destination space in Photoshop when printing. This solved almost all of the problems and results were very good. Dither is good, visually dotless, tonal transitions were smooth. There were a couple of barely detectable single percent jumps in a grad but images did not show them. I don't know if you can send it 16 bit files or how it would process them. I believe the internal ramping in the driver is 16 bit though. This was done with a 3000, I doubt the Gimp people have set this option up to properly work with the common practice of doubling up two of the four inks in six ink printers, so I can't say anything about that. I imagine an 1160 would work very well. It looks like some sort of hex set would be necessary in a 6 ink printer from the color ink test I did. There is a raw CMYK mode, more of interest to me. Unfortunately it's not quite there yet. 100% GCR occurs, and it permits no CMY under 100% K. However C, M, Y, CM, CY, MY etc. go through unmuxed. Again the intriguing ink control options do nothing in this mode. However for quads this could all be made to work by partitioning that avoids any equal CMY densities (probably a natural consequence of good partitioning anyway). Also 100% K will have no support from other inks so those concerned about that will have to pick their K ink accordingly. This mode could therefore be of use to those wanting to work with their own seps in CMYK. Of course your file size will quadruple. I'm sure RGB workflows could easily be worked for this driver as well. I didn't take any of this very far since I'm not actually using it. But it looks very promising, it's free, and the developers are very bright and responsive. They could use help from anyone who knows about printing AND code. I have no idea if it is workable on a PC. This is a very promising driver, all of the Epsons of interest are supported, did I mention it's free? I hope anyone using Jaguar will look into it and support it, it opens up the options considerably. John Labovitz, a long time quad printer and much brighter person than I, has demonstrated to me that good hackers could easily build a custom driver from this thing and have a great deal of ink control to accomplish any number of desired results. Tyler