This post contains background information for my Prototype Rails presentation in January 2004. It deals with calibration of the white laser printer. The purposes of calibration are to find color values for the decals that:
- result in a dense coverage of toner on the decal film so that the underlying freight car color doesn't show through
- has the desired shade of gray to match the prototype lettering, with due consideration of any weathering to be applied
- has a fairly even flow of toner out of the three white cartridges (so that you don't constantly have to be replacing the toner in a high output cartridge)
- all other things being equal, minimizes the total toner flow from the white cartridges because this is the more expensive stuff.
In several earlier posts I've described how I took an OKI color printer with CMYK cartridges and modified it to a KWWW printer for printing white/gray/black on decal film. Those were the hardware aspects of this decal printing project. In this post I want to discuss the calibration process, particularly why calibration is necessary and how I went about developing and implementing my approach to calibration.
As I remarked earlier, there are four cartridges in a typical color laser printer (cyan, magenta, yellow, and black). The printer works by allowing a small amount of powdered toner to come out of each of the four cartridges, where it is transferred onto the paper and fused in place. The small amounts of each powder are selected by software inside the printer to produce the desired color spot. So the printer has a computer and software embedded inside of it, and that computer software is predicated on the expectation that the 4 cartridges are C, M, Y, and K. The computer software inside the printer doesn't know that the user has physically replaced the C, M, and Y cartridges with white and that the user is expecting to see a shade of gray.
Further complicating things, the laser printer is an RGB machine. Regardless of the colorspace of the artwork (i.e., whether the user used RGB or CMYK or HSB or Lab to specify colors for Photoshop or Illustrator or Corel Draw or whatever program), once the Print... command invoked, the software, the system color matching software, and the printer driver combine to produce a string of RGB values to send to the printer. The printer software has a set of functions inside of it that are designed to take a certain RGB value as input and figure out the right valve settings for the CMYK cartridges to produce the correct color corresponding to that RGB value. And since each manufacturer's toner is a bit different, this set of functions is buried inside the printer and is considered proprietary, and there's no way for a user to get at it and modify it for the KWWW process.
An overview of the 3 step printing process:
- The user prepares the artwork in some computer program of choice (e.g., Photoshop or Illustrator) in some color space that is convenient for the user (e.g., CMYK or HSB or RGB or Lab)
- When Print... is invoked, the software, including the system color matching software and the printer driver take each pixel in the artwork and create an RGB value for that pixel which is sent to the printer.
- The printer takes the RGB value and figures out "valve settings" for the CMYK cartridges which lets the precise amount of powder out of each cartridge to reproduce the color of the pixel.
So with my modified KWWW printer, I needed to find out what RGB values in the artwork turn into the desired shades of gray with acceptable coverage density on the printed page. Because the details of steps 2 and 3 of the printing process were hidden to me, I had no easy theoretical way to approach this, so I developed a brute force calibration process. Inside Photoshop, I made a document in RGB space that has 4096 colored dots, each with a different RGB value. There are 8 pages, each page has 2 arrays of dots, and each array has 16 rows and 16 columns of dots. Then I printed each of the 8 pages onto red sheets of paper using the KWWW printer. Here's a low-resolution scan of page 1 (you can click to enlarge):
The reason I printed this on red paper is that it is a good background on which to evaluate coverage density as well as the color. Here's a higher resolution scan of the lower block of page 1 that illustrates some of the variability in the effects that can be produced:
So the user searches through the 8 pages and finds the coverage density and shade of white/gray desired, and then the page, block (upper or lower), and row and column of the desired dot easily translates into an RGB value (or values) that are used for the artwork.
Richard Brennan is a TT-scale modeler who has been following these decal posts, and he was kind enough to help me with a beta test. We exchanged a few eMails as part of the color selection/calibration process, and he prepared the artwork to letter a variety of freight cars he is interested in for which TT-scale decals are not available. He created a pdf of his artwork and eMailed it to me, and I printed out the pdf on the KWWW printer. In this little test project Richard acted as the content creator and I acted as the print service bureau:
This is a photo of his proof-page (red paper, black boxes separating different freight cars, and white/gray lettering at top) and the final 8 1/2" by 11" printed decal sheet (at bottom). You can barely see the white lettering in this photo, but I looked at the proof sheet and the decal sheet under good light with the Optivisor and I think he chose RGB values that resulted in pretty good coverage and his shades of white/gray look pretty compelling to me. The proof of the pudding will be when they are applied to actual models...
I sent him the printed sheets yesterday and they should arrive at his place on Tuesday. In return, Richard has agreed to evaluate these decals from the point of view of an end-user and report on his findings, which will be posted here when they become available.
Thanks Richard!
Charles Hostetler
Goshen, Ind.