This file is README.WRITE-W. It is extracted from a correspondence message sent during September, 1986. From: Neenie Billawala Subject: Re: Metafont parameters for Imagen 24/300 (This discussion is also relevant for the LN03) How to get the CM fonts to work better on a "write-white" machine ================================================================= The problem in getting a like result with write-white(ww)/write-black(wb) machines from the same pixel pattern lies in the fact that the effective size of each pixel differs; the size is typically smaller in ww machines. This means that you won't get the same variations in line thickness. Theoretically, one pixel on a 300dpi printer should have a width of 1/300 of an inch. On a ww machine, it's typically less; on a wb machine, it's usually a bit more. So the gradations of thickness may look something like the following: # of pixels "ideal width" ww width wb width 1 1 * 1/300inch .8 * 1/300inch 1.2*1/300inch 2 2 * 1/300inch 1.8 * 1/300inch 2.2*1/300inch 3 3 * 1/300inch 2.8 * 1/300inch 3.2*1/300inch 4 4 * 1/300inch 3.8 * 1/300inch 4.2*1/300inch 10 10 * 1/300inch 9.8 * 1/300inch 10.2*1/300inch So, you can see, the greater the number of pixels, the less important the .4 pixel difference becomes. The greatest discrepancy and problems occur with lines one pixel wide, where the % of difference is the greatest. Given these limitations, it is not possible to get exactly the same results on two such separate machines. Increasing the amount of "blacker" has the effect of adding that absolute amount to certain values; i.e., blacker:=2 would always add 2 pixels to the stem, no matter the resolution. A while back, someone came by with the same problem for a similar machine. We tried 1) some higher settings of blacker 2) selectively adding blacker to some values 3) setting a mimimum line thickness. Both 1) and 2) often created unwanted character distortions. The (tfm) widths wanted to be the same, and when e.g., each line in a character was increased by one; the shapes were often distorted, counter (inside) shapes seemed to suffer the most. A first attempt with 3) showed promise, and though the result was not as dark as the wb machine, the feeling of CM was retained. However, in the first attempt, we did not manage to change all of the thicknesses to 2 pixels. DEK made a global correction, which should keep minimum thicknesses to 2 pixels. Looking at the pixel patterns, this fix appears to work. However, the man who has the ww printer and who originally came by with this problem, is part of the University and has been traveling, so I haven't seen any concrete results yet. Such is the way of Academia! Appendix: (Pierre A. MacKay) Roughly speaking, the reason for this problem is that the black dot in a write-black machine has a roughly round shape, and naturally tends to overflow the minimum dot size. The dot on a write-white machine has an incluse square shape, and tends to be diminished by an oversize white dot unless the print engine is perfectly focussed. ..... ....... . . ......... ..... ......... ..... ......... ..... ....... . . ..... Write-black dot Write-white dot The ./cmfonts/mf directory contains an adjusted font_setup file. The changes are found in 2 lines of the def with the comment "% "WRITE WHITE" ONLY!". Add the line input white_setup to your file of mode_defs (or use ./utilityfonts/bases/modes.mf, which already has it), and in any mode_def for a write-white device, add the line let font_setup=white_setup; as the first line of the mode_def for any write-white device (see the RicohFourZeroEightZero mode_def in modes.mf). Directories [ ./utilityfonts/bases ./cmfonts/mf ]