- Notes on measurements
- MS Windows printers
- File formats
- HP Deskjet official drivers
- H-P 8xx, 1100, and 1600 color inkjet printers
- H-P 812, 815, 832, 880, 882, 895, and 970 color inkjet printers
- Other H-P color inkjet printers
- Canon BJC-8200 printer
- Other Canon BubbleJet (BJC) printers
- Epson Stylus color printer (see also uniprint)
- uniprint, a flexible unified printer driver
- Sun SPARCprinter
- Apple dot matrix printer
For other information, see the Ghostscript overview. You may also be interested in how to build Ghostscript and install it, as well as the description of the driver interface.
Several different important kinds of measures appear throughout this document: inches, centimeters and millimeters, points, and bits per pixel.
This section was written by Russell Lang, the author of Ghostscript's MS Windows-specific printer driver, and updated by Pierre Arnaud, the current maintainer.
The mswinpr2 device uses MS Windows printer drivers, and thus should work with any printer with device-independent bitmap (DIB) raster capabilities. The printer resolution cannot be selected directly using PostScript commands from Ghostscript: use the printer setup in the Control Panel instead. It is however possible to specify a maximum resolution for the printed document (see below).
If no Windows printer name is specified in -sOutputFile, Ghostscript prompts for a Windows printer using the standard Print Setup dialog box. You must set the orientation to Portrait and the page size to that expected by Ghostscript; otherwise the image will be clipped. Ghostscript sets the physical device size to that of the Windows printer driver, but it does not update the PostScript clipping path.
If a Windows printer name is specified in -sOutputFile using the format "\\spool\printer_name", for instance
gs ... -sOutputFile="\\spool\Apple LaserWriter II NT"
then Ghostscript attempts to open the Windows printer without prompting (except, of course, if the printer is connected to FILE:). Ghostscript attempts to set the Windows printer page size and orientation to match that expected by Ghostscript, but doesn't always succeed. It uses this algorithm:
The Ghostscript physical device size is updated to match the Windows printer physical device.
The mswinpr2 device supports a limited number of command-line parameters (e.g. it does not support setting the printer resolution). The recognized parameters are the following:
- -sDEVICE=mswinpr2
- Selects the MS Windows printer device. If Ghostscript was not compiled with this device as the default output device, you have to specify it on the command line.
- -dNoCancel
- Hides the progress dialog, which shows the percent of the document page already processed and also provides a cancel button. This option is useful if GS is intended to print pages in the background, without any user intervention.
- -sOutputFile="\\spool\printer_name"
- Specifies which printer should be used. The printer_name should be typed exactly as it appears in the Printers control panel, including spaces.
Several extra options exist which cannot be set through the command-line, but only by executing the appropriate PostScript setup code. These options can be set through the inclusion of a setup file on the command-line:
gs ... setup.ps ...
The setup.ps file is responsible for the device selection, therefore you should not specify the -sDEVICE=mswinpr2 option on the command-line if you are using such a setup file. Here is an example of such a setup file:
mark
/NoCancel true % don't show the cancel dialog
/BitsPerPixel 4 % force 4 bits/pixel
/UserSettings
<<
/DocumentName (Ghostscript document) % name for the Windows spooler
/MaxResolution 360 % maximum document resolution
>>
(mswinpr2) finddevice % select the Windows device driver
putdeviceprops
setdevice
This example disables the progress dialog (same as the -dNoCancel option), forces a 4 bits/pixel output resolution and specifies additional user settings, such as the document name (which will be displayed by the Windows spooler for the queued document) and the maximum resolution (here 360 dpi). It then finds and selects an instance of the MS Windows device printer and activates it. This will show the standard printer dialog, since no /OutputFile property was specified.
The following options are available:
- /NoCancel boolean
- Disables (hides) the progress dialog when set to true or show the progress dialog if not set or set to false.
- /OutputFile string
- Specifies which printer should be used. The string should be of the form \\spool\printer_name, where the printer_name should be typed exactly as it appears in the Printers control panel, including spaces.
- /QueryUser integer
- Shows the standard printer dialog (1 or any other value), shows the printer setup dialog (2) or selects the default Windows printer without any user interaction (3).
- /BitsPerPixel integer
- Sets the device depth to the specified bits per pixel. Currently supported values are 1 (monochrome), 4 (CMYK with screening handled by Ghostscript) and 24 (True Color, dithering handled by the Windows printer driver; this option can produce huge print jobs).
- /UserSettings dict
- Sets additional options, defined in a dictionary. The following properties can be set:
- /DocumentName string
- Defines the user friendly document name which will be displayed by the Windows spooler.
- /DocumentRange [n1 n2]
- Defines the range of pages contained in the document. This information can be used by the printer dialog, in conjunction with the following property.
- /SelectedRange [n1 n2]
- Defines the selected range of pages. This information will be displayed in the printer dialog and will be updated after the user interaction. A PostScript program could check these values and print only the selected page range.
- /MaxResolution dpi
- Specifies the maximum tolerated output resolution. If the selected printer has a higher resolution than dpi, then Ghostscript will render the document with a submultiple of the printer resolution. For example, if MaxResolution is set to 360 and the output printer supports up to 1200 dpi, then Ghostscript renders the document with an internal resolution of 1200/4=300 dpi. This can be very useful to reduce the memory requirements when printing in True Color on some high resolution ink-jet color printers.
These properties can be queried through the currentpagedevice operator. The following PostScript code snippet shows how to do it for some of the properties:
currentpagedevice /BitsPerPixel get == % displays the selected depth currentpagedevice /UserSettings get % get the additional options.. /us exch def % ..and assign them to a variable us /DocumentName get == % displays the document name us /SelectedRange get == % displays the selected page range % other misc. informations (don't rely on them) us /Color get == % 1 => monochrome output, 2 => color output us /PrintCopies get == % displays the number of copies requested
There are a few undocumented informations stored in the UserSettings dictionary. You should not rely on them. Their use is still experimental and they could be removed in a future version.
If the Windows printer supports the duplex printing feature, then it will also be available through the mswinpr2 device. You can query for this support through the /Duplex propery of the currentpagedevice. If it returns null, then the feature is not supported by the selected printer. Otherwise, true means that the printer is currently set up to print on both faces of the paper and false that it is not, but that it can.
The following example shows how to print on both faces of the paper (using the long side of the paper as the reference):
<< /Duplex true /Tumble false >> setpagedevice
Ghostscript includes output drivers that can produce Independent JPEG Group JFIF (JPEG File Interchange Format) files from PostScript images. Please note that JPEG is a compression method specifically intended for continuous-tone images such as photographs, not for graphics, and it is therefore quite unsuitable for the vast majority of page images produced with PostScript, which should be saved in a form better for graphics, such as Portable Network Graphics (PNG) format. If you get crummy-looking JPEG files, don't blame Ghostscript; instead consult a reference about uses and abuses of JPEG, such as the JPEG FAQ
http://www.faqs.org/faqs/jpeg-faq/
You can use the JPEG output drivers -- jpeg to produce color JPEG files and jpeggray for grayscale JPEGs -- the same as other file-format drivers: by specifying the device name and an output file name, for example
gs -sDEVICE=jpeg -sOutputFile=foo.jpg foo.ps
You can also use the -r switch to specify the imaging resolution and thus the output file's size in pixels. The default resolution is normally 72×72dpi.
The JPEG devices support several special parameters to control the JPEG "quality setting" (DCT quantization level).
- -dJPEGQ=N (integer from 0 to 100, default 75)
- Set the quality level N according to the widely used IJG quality scale, which balances the extent of compression against the fidelity of the image when reconstituted. Lower values drop more information from the image to achieve higher compression, and therefore have lower quality when reconstituted.
- -dQFactor=M (float from 0.0 to 1.0)
- Adobe's QFactor quality scale, which you may use in place of JPEGQ above. The QFactor scale is used by PostScript's DCTEncode filter but is nearly unheard-of elsewhere.
At this writing the default JPEG quality level of 75 is equivalent to -dQFactor=0.5, but the JPEG default might change in the future. The JPEG drivers could be extended to support additional JPEG compression options, such as the other DCTEncode filter parameters, but so far they haven't been.
There are two unrelated sets of TIFF drivers. There are two color TIFF drivers that produce uncompressed output:
- tiff12nc
- Produces 12-bit RGB output (4 bits per component).
- tiff24nc
- Produces 24-bit RGB output (8 bits per component).
The remaining TIFF drivers all produce black-and-white output with different compression modes:
- tiffcrle
- G3 fax encoding with no EOLs
- tiffg3
- G3 fax encoding with EOLs
- tiffg32d
- 2-D G3 fax encoding
- tiffg4
- G4 fax encoding
- tifflzw
- LZW-compatible (tag = 5) compression
- tiffpack
- PackBits (tag = 32773) compression
The black-and-white TIFF drivers support creation of files that are comprised of more than a single strip. Multi-strip files reduce the memory requirement on the reader, since readers need only store and process one strip at a time. These drivers provide two parameters beyond the standard set:
- -dMaxStripSize=N (non-negative integer; default = 0)
- Set the maximum (uncompressed) size of a strip.
- -dAdjustWidth=true|false (Boolean; default = true)
- If true, then if the requested page width is close to either A4 (1728 columns) or B4 (2048 columns), set the page width to A4 or B4 respectively.
The TIFF 6.0 specification, Section 7, page 27, recommends that the size of each strip be about 8 Kbytes. As an example, to generate the 'tiger' in tiffg4 format that has the recommended strip size, use:
gs -sDEVICE=tiffg4 -sOutputFile=tiger.tif -dMaxStripSize=8192 examples/tiger.ps
If the value of the MaxStripSize parameter is smaller than a single image row, then no error will be generated, and the TIFF file will be generated correctly using one row per strip. Note that smaller strip sizes increase the size of the file by increasing the size of the StripOffsets and StripByteCounts tables, and by reducing the effectiveness of the compression which must start over for each strip.
If the value of MaxStripSize is 0 (the default), then the entire image will be a single strip.
HP provides official drivers for many of their Deskjet printer models. In order to use these drivers, you will need the HP Inkjet Server as well as Ghostscript, available from http://hpinkjet.sourceforge.net. This version of Ghostscript includes the patch from version 0.97 of the hpijs software. If you are installing hpijs from an RPM, you will only need the hpijs RPM, not the ghostscript-hpijs one, as the code needed to work with hpijs is already included.
As of the 0.97 version, hpijs supports the following printer models:
You will need to add the following line to your makefile:
DEVICE_DEVS2=$(DD)DJ630.dev $(DD)DJ6xx.dev $(DD)DJ6xxP.dev $(DD)DJ8xx.dev $(DD)DJ9xx.dev $(DD)DJ9xxVIP.dev $(DD)AP21xx.dev
Please see http://hpinkjet.sourceforge.net for more information about this driver. Thanks to the folks at HP, especially David Suffield for making this driver available and working to integrate it with Ghostscript.
This section, written by Uli Wortmann <uliw@erdw.ethz.ch>, deals with the DeskJet 670, 690, 850, 855, 870, 890, 1100, and 1600.
The source module gdevcd8.c contains four generic drivers:
cdj670 HP DeskJet 670 and 690 cdj850 HP DeskJet 850, 855, 870, and 1100 cdj890 HP DeskJet 890 cdj1600 HP DeskJet 1600
Credits: Much of the driver is based on ideas derived from the cdj550 driver of George Cameron. The support for the hp670, hp690, hp890 and hp1600 was added by Martin Gerbershagen.
11.11.96 Version 1.0 25.08.97 Version 1.2 Resolved all but one of the known bugs, introduced a couple of perfomance improvements. Complete new color-transfer-function handling (see gamma). 01.06.98 Version 1.3 Due to the most welcome contribution of Martin Gerbershagen (ger@ulm.temic.de), support for the hp670, hp690 and hp890 and hp1600 has been added. Martin has also resolved all known bugs. Problems:Dark colors are still pale.
The hp690 is supported through the hp670 device, the hp855, hp870 and the hp1100 through the hp850 device. The driver needs no longer special switches to be invoked except -sDEVICE=cdj850, -sDEVICE=CDJ890, -sDEVICE=CDJ670, or -sDevice=CDJ1600. The following switches are supported.
-dPapertype= 0 plain paper [default] 1 bond paper 2 special paper 3 glossy film 4 transparency film Currently the lookup tables are unsuited for printing on special paper or transparencies. For these please revert to the gamma functions. -dQuality= -1 draft 0 normal [default] 1 presentation -dRetStatus= 0 C-RET off 1 C-RET on [default] -dMasterGamma= 3.0 [default = 1.0]
Note: To take advantage of the calibrated color-transfer functions, be sure not to have any gamma statements left! If you need to (i.e., for overhead transparencies), you still can use the gamma functions, but they will override the built-in calibration. To use gamma in the traditional way, set MasterGamma to any value greater than 1.0 and less than 10.0. To adjust individual gamma values, you have to additionally set MasterGamma to a value greater than 1.0 and less than 10.0. With the next release, gamma functions will be dropped.
When using the driver, be aware that printing at 600dpi involves processing large amounts of data (> 188MB !). Therefore the driver is not what you would expect to be a fast driver ;-) This is no problem when printing a full-sized color page (because printing itself is slow), but it's really annoying if you print only text pages. Maybe I can optimize the code for text-only pages in a later release. Right now, it is recommended to use the highest possible optimisation level your compiler offers. For the time being, use the cdj550 device with -sBitsPerPixel=3 for fast proof prints. If you simply want to print 600dpi BW data, use the cdj550 device with -sBitsPerPixel=8 (or 1).
Since the printer itself is slow, it may help to set the process priority of the gs process to "regular" or even less. On a 486/100MHz this is still sufficient to maintain a continuous data flow. Note to OS/2 users: simply put the gs window into the background or minimize it. Also make sure that print01.sys is invoked without the /irq switch (great speed improvement under Warp4).
The printer default settings compensate for dot-gain by a calibrated color-transfer function. If this appears to be too light for your business graphs, or for overhead transparencies, feel free to set -dMasterGamma=1.7. Furthermore, you may tweak the gamma values independently by setting -dGammaValC, -dGammaValM, -dGammaValY or -dGammaValK (if not set, the values default to MasterGamma). This will only work when -dMasterGamma is set to a value greater than 1.0.
Further information, bugs, tips etc, can be found at my website. To learn more about gamma, see ftp://ftp.igd.fhg.de/pub/doc/colour/GammaFAQ.pdf.
Depending on how you transfer the files, under UNIX you may need to remove the CRs of the CR-LF sequence used for end-of-line on DOS-based (MS Windows-based) systems. You can do this in unpacking the files with unzip -a hp850.zip.
To compile with gs5.x or later, simply add to your makefile
DEVICE_DEVS4=cdj850.dev cdj670.dev cdj890.dev cdj1600.dev
Have fun!
Uli <uliw@erdw.ethz.ch>
http://www.erdw.ethz.ch/~bonk/
This section, written by Matthew Gelhaus <hp880@gelhaus.net>, deals with the DeskJet 812, 815, 832, 880, 882, 895, and 970.
This is a modified version of the HP8xx driver written by Uli Wortmann. More information and download are available at http://www.gelhaus.net/hp880c/.
The source module gdevcd8.c contains one generic driver:
cdj880 HP DeskJet 812, 815, 832, 880, 882, 895, and 970
Credits: This driver is based on the cdj850 driver by Uli Wortmann, and shares the same internal structure, although the PCL3+ interpretation has changed.
15.03.99 Version 1.3 Initial version, based on Version 1.3 of Uli Wortmann's driver. 26.02.00 Version 1.4beta Greatly improved color handling & dithering, but not yet complete enough to use for text.
All printers are supported through the cdj880 device. Invoke with -sDEVICE=cdj880. The following switches are supported.
-dPapertype= 0 plain paper [default] 1 bond paper 2 special paper 3 glossy film 4 transparency film Currently the lookup tables are unsuited for printing on special paper or transparencies. For these please revert to the gamma functions. -dQuality= -1 draft 0 normal [default] 1 presentation -dMasterGamma= 3.0 [default = 1.0]
The printer default settings compensate for dot-gain by a pre-defined color-transfer function. If this appears to be too light for your business graphs, or for overhead transparencies, feel free to set -dMasterGamma=1.7. Furthermore, you may tweak the gamma values independently by setting -dGammaValC, -dGammaValM, -dGammaValY or -dGammaValK (if not set, the values default to MasterGamma). This will only work when -dMasterGamma is set to a value greater than 1.0.
Further information, bugs, tips etc, can be found at my website.
To compile with gs6.x or later, simply add to your makefile
DEVICE_DEVS4=$(DD)cdj880.dev
Matthew Gelhaus <mailto:hp880@gelhaus.net>
http://www.gelhaus.net/hp880c/
This section, written by George Cameron, deals with the DeskJet 500C, DeskJet 550C, PaintJet, PaintJet XL, PaintJet XL300, the DEC LJ250 operating in PaintJet-compatible mode.
The source module gdevcdj.c contains six generic drivers:
cdj500 HP DeskJet 500C and 540C cdj550 HP DeskJet 550C, 560C, 660C, 660Cse pjxl300 HP PaintJet XL300, DeskJet 1200C, and CopyJet pjtest HP PaintJet pjxltest HP PaintJet XL declj250 DEC LJ250
All these drivers have 8-bit (monochrome), 16-bit and 24-bit (colour) and for the DJ 550C, 32-bit (colour, CMYK mode) options in addition to standard colour and mono drivers. It is also possible to set various printer-specific parameters from the command line, for example
gs -sDEVICE=cDeskJet -dBitsPerPixel=16 -dDepletion=1 -dShingling=2 tiger.ps
Note: the old names cDeskJet, cdjcolor and cdjmono drivers have been retained; however, their functionality duplicates that available using the drivers above (and cDeskJet is identical to cdj500). That is, we can use
gs -sDEVICE=cdj500 -dBitsPerPixel=24 for cdjcolor, and gs -sDEVICE=cdj500 -dBitsPerPixel=1 for cdjmono
If the preprocessor symbol A4 is defined, the default paper size is ISO A4; otherwise it is U.S. letter size (see about paper sizes in the usage documentation). You can specify other paper sizes on the command line, including A3 for the PaintJet XL and PaintJet XL300, as also explained in the usage documentation.
The DeskJet's maximum printing width is 2400 dots, or 8 inches (20.32cm). The printer manuals say that the maximum recommended printing height on the page is 10.3 inches (26.16cm), but since this is obviously not true for A4 paper, and I have been unable to detect any problems in printing longer page lengths, this would seem to be a rather artificial restriction.
All DeskJets have 0.5 inches (1.27cm) of unprintable bottom margin, due to the mechanical arrangement used to grab the paper. Side margins are approximately 0.25 inches (0.64cm) for U.S. letter paper, and 0.15 inches (0.38cm) for A4.
Several printer "properties" have been implemented for these printers. Those available so far are all integer quantities, and thus may be specified, for instance, like
gs -dBitsPerPixel=32 -dShingling=1 ...
which sets the BitsPerPixel parameter to 32 and the Shingling parameter to 1.
If the preprocessor symbol BITSPERPIXEL is defined as an integer (see below for the range of allowable values), that number defines the default bits per pixel (bit depth) for the generic drivers. If the symbol is undefined, the default is 24 bits per pixel. It is, of course, still possible to specify the value from the command line as described below. Note also that the cDeskJet, cdjcolor and cdjmono drivers are unaffected by setting this symbol, as their default settings are predefined to be 1, 3 and 24 respectively.
All of the drivers in gdevcdj.c accept a command line option to set the BitsPerPixel property. This gives considerable flexibility in choosing various tradeoffs among speed, quality, colour, etc. The valid numbers are:
- 1
- A standard Ghostscript monochrome driver, using black ink (by installing the separate mono cartridge in the case of the DeskJet 500C, or automatically for the other printers).
- 3
- A standard Ghostscript colour driver, using internal dithering. This is fast to compute and to print, but the clustered dithering can lose some detail and colour fidelity.
- 8
- An "error-diffusion" monochrome driver which uses Floyd-Steinberg dithering to print greyscale images. The patterns are much more randomised than with the normal clustered dithering, but the data files can be much larger and somewhat slower to print.
- 16
- A "cheaper" version of the 24-bit driver, which generates Floyd-Steinberg colour dithered output using the minimum memory (this may be helpful when using Ghostscript has not been compiled using a 16-bit build environment). The quality can be almost as good as the 24-bit version.
- 24
- A high-quality colour driver using Floyd-Steinberg dithering for maximum detail and colour range. However, it is very memory-intensive, and thus can be slow to compute. It tends to produce rather larger raw data files, so they can also take longer to print.
- 32
- Only for the DeskJet 550C, which uses the black cartridge and the colour cartridge simultaneously (that is, CMYK printing). This printer can both be faster and give higher quality than the DeskJet 500C, because of the true black ink. (Note that the 24-bit mode also permits CMYK printing on this printer, and uses less memory. Any differences between 24-bit and 32-bit should be small.)
Name Type
BlackCorrect int Colour correction to give better blacks when using the DJ500C in colour mode. For example, the default of 4 reduces the cyan component to 4/5. Range accepted: 0 - 9 (0 = none). Shingling int Interlaced, multi-pass printing: 0 = none, 1 = 50%, 2 = 25%, 2 is best and slowest. Depletion int "Intelligent" dot-removal: 0 = none, 1 = 25%, 2 = 50%, 1 best for graphics? Use 0 for transparencies.
Name Type
PrintQuality int Mechanical print quality: -1 = fast, 0 = normal, 1 = presentation. Fast mode reduces ink usage and uses single-pass operation for some media types. Presentation uses more ink and the maximum number of passes, giving slowest printing for highest quality RenderType int
- 0
- driver does dithering
- 1
- snap to primaries
- 2
- snap black to white, others to black
- 3
- ordered dither
- 4
- error diffusion
- 5
- monochrome ordered dither
- 6
- monochrome error diffusion
- 7
- cluster ordered dither
- 8
- monochrome cluster ordered dither
- 9
- user-defined dither (not supported)
- 10
- monochrome user-defined dither ns.
The PaintJet (non-XL) has no additional properties.
One consequence of using Floyd-Steinberg dithering rather than Ghostscript's default clustered ordered dither is that it is much more obvious that the ink dots are rather larger on the page than their nominal 1/180-inch or 1/300-inch size (clustering the dots tends to minimise this effect). Thus it is often the case that the printed result is rather too dark. A simple empirical correction for this may be achieved by preceding the actual PostScript file to be printed by a short file which effectively sets the gamma for the device, such as
gs ... gamma.ps colorpic.ps -c quit
where gamma.ps is
%!
/.fixtransfer {
currentcolortransfer 4 {
mark exch
dup type dup /arraytype eq exch /packedarraytype eq or
1 index xcheck and { /exec load } if
0.333 /exp load
] cvx 4 1 roll
} repeat setcolortransfer
} bind odef
.fixtransfer
/setpagedevice { setpagedevice .fixtransfer } bind odef
This does the gamma correction after whatever correction the device might be doing already. To do the correction before the current correction,
%!
/.fixtransfer {
currentcolortransfer 4 {
mark 0.333 /exp load 4 -1 roll
dup type dup /arraytype eq exch /packedarraytype eq or
1 index xcheck and { /exec load } if
] cvx 4 1 roll
} repeat setcolortransfer
} bind odef
.fixtransfer
/setpagedevice { setpagedevice .fixtransfer } bind odef
This example sets the gamma for R, G, and B to 3, which seems to work reasonably well in practice.
This feature is available on HP's more recent inkjet printers, including the DeskJet 520 (mono), 540 (mono or colour) and 560C (mono and colour). The colour and monochrome drivers for the HP DeskJet 550c are (probably) the best you will get for use with Ghostscript, for the following reasons.
These printers do not offer true 600×300dpi resolution. Those that print in colour are strictly 300×300dpi in colour mode, while in mono mode there is a pseudo 600×300dpi mode with the restriction that you can't print two adjacent dots. In effect what you have is 600dpi dot positioning, but on average you don't get more dots per line. This provides the possibility, for instance, to have sharper character outlines, because you can place dots on the edges nearer to their ideal positions. This is why it is worth doing.
However, HP will not support user-level programming of this resolution-enhanced mode, one reason being that (I understand) all the dot spacing has to be done by the driver, and if you get it wrong, you can actually damage the print head.
To summarise, you may lose a smidgin of (potential) text clarity using the 550c drivers (cdj550, cdjcolor, cdjmono etc.), but other than that, they are the ones for the job.
For all the printers above, the choice of paper is critically important to the final results. The printer manuals suggest type of paper, but in general, smoother, less fibrous types give better results. In particular, the special ink-jet paper can make a big difference: colours are brighter, but most importantly, there is almost no colour bleed, even with adjacent areas of very heavy inking. Similarly the special coated transparencies also work well (and ordinary transparencies do not work at all!).
The Unix procedure unix-lpr.sh provides one example of setting up a multi-option colour PostScript lpr queue on Unix systems, and includes the ability to choose a range of different colour options and printer accounting and error logging.
Caveat emptor! It is not always easy for me to test all of these drivers, as the only colour printer I have here is the DeskJet 500C. I rely on others to test drivers for the additional machines and report their findings back to me.
This section was contributed by the author of the uniprint configuration files for the Canon BJC-8200, Stephan C. Buchert <scb@stelab.nagoya-u.ac.jp>. These files also handle the Japanese Canon F850 printer.
Warning: Usage of this program is neither supported nor endorsed by the Canon corporation. Please see the Ghostscript license regarding warranty.
The Canon Bubble Jet printer BJC-8200 is designed for printing digital photos and halftone images. Software drivers for Windows 95-2000 and Mac are usually included and can be downloaded from the Canon web sites like http://consumer.usa.canon.com/bjc/bjc8200/ for the US market. If these drivers cannot be used for some reason, then at present Ghostscript is probably the alternative giving the best results.
The BJC-8200 has features not found among the specs of earlier bubble jet models (except the even more advanced BJC-8500) and is advertised to offer:
Access to features 5 and 6 requires use of the original Canon drivers for the foreseeable future. This README is about getting the printer features 1-3 working with Ghostscript. No (re)compilation of Ghostscript is normally required.
Ghostscript comes with a relatively highly configurable driver, called uniprint, for printers which understand raster images in various propriety formats. Most options for this driver are usually organized into files having the suffix ".upp." Ghostscript versions >= 5.10 (or even earlier) include such uniprint control files for the Canon BJC-610. They work also well for some other Canon Bubble Jet models, for example for my BJC-35vII. But when using them for a BJC-8200 the result is unsatisfactory.
After some experimenting with the options for uniprint I have obtained quite satisfactory prints with my printer(*). This distribution includes six new uniprint control files:
They are included in Ghostscript >=6.21. For older versions you can put them anywhere in the Ghostscript search path (type "gs -h" to see the path), but should perhaps add the files to the directory with the other *.upp files. This is "/usr/share/ghostscript/gs6.01/lib" in my RedHat 6.1 Linux box with Aladdin Ghostscript 6.01.
Here is an explanation of my file name convention: the prefix "bj8" should perhaps be used for the Canon BJC-8200 and compatible (like the Japanese F850 and perhaps the non-Japanese BJC-8500) models. The next two letters indicate the print media:
The numbers at positions 6 and 7 indicate the resolution
The last letter stands for a quality factor that effects also the print speed (presumably related to the number of passes that the printhead makes).
Printing a postcard size (~10x15 cm^2) image at 1200x1200 DpI^2 takes about 3 minutes. The output of Ghostscript is then typically 4-5 MByte. The bootleneck seems to be the transfer of the raster image in run-length encoded Canon format to the printer (via the parallel port on my system) or the printer's speed, not Ghostscript or the uniprint renderer.
Here is information useful for changing or adding uniprint control files for the BJC-8200:
In "-dupBeginPageCommand=..." use the line
1b28 64 0400 04b0 04b0
for 1200x1200 resolution, and
1b28 64 0400 0258 0258
for 600x600. The "-r" option in the control file must of course match this line. Other resolutions might work as well, but I didn't try.
Crucial are the numbers in the lines like
1b28 63 0300 3005 04
^ ^
Plain Paper 0 4 Highest quality
OHP transparency 2 .
T-shirt transfer 3 .
Glossy Photo Film 5 .
High Gloss Paper 6 0 Lowest quality
Photo Paper Pro 9
Presently uniprint can use the black (K), cyan (C), magenta (M), and yellow (Y) colors in the BJC-8200. The unused colors are photo (or light) cyan (c) and magenta (m). Also the Canon driver seems to use only CMYK, for example when printing on Photo Paper Pro in "Camera" or "SuperPhoto" mode. These modes supposedly produce prints of the best quality that the Canon driver can offer. Other modes of Canon driver do use up to all six color cartridges (CMYKcm). Therefore expanding uniprint's capabilities for six colors would be interesting, but it may not increase the output quality of 6-color printers such as the BJC-8200 drastically.
More control files for uniprint could be added in order to offer more versatility for controlling the BJC-8200 within a Ghostscript installation. The number of possible combinations for media type, resolution and print quality factor is very large, many combinations would not make much sense, many might be used here and there, but relatively rarely. The user would have to remember a name for each combination that is used.
A better way would be to let the user patch optionally a user owned or system wide uniprint control file before each print via some print tool. This is similar to the approach taken by Canon with their driver for Windows. Similarly a uniprint tool could also incorporate other functions such as printing test and demo pages and the low ink warning once the protocol for this is known. Clearly it would be difficult to code such a uniprint tool for all the platforms where Ghostscript is running.
In order to install a BJC-8200 printer on a RedHat Linux system with RedHat's printtool, you need also to insert with a text editor the contents of the file bj8.rpd into the RedHat printer database /usr/lib/rhs/rhs-printfilters/printerdb. Insert it most appropriately after the section
StartEntry: U_CanonBJC610
.
.
.
EndEntry
< --- insert here "bj8.rpd" from this distribution:
< --- StartEntry: U_CanonBJC8200
.
.
.
E-mail address: <scb@stelab.nagoya-u.ac.jp>
Footnotes:
(*) Actually I have a F850, not a BJC-8200. That model is sold for the Japanese market only. The specs and also the external look are the same as those of the BJC-8200 models for the American and European markets. I expect that the raster image mode which is used exclusively by Ghostscript is entirely compatible for both models.
Stephan C. Buchert
This section was contributed by the author of the drivers, Yves Arrouye (<yves.arrouye@usa.net>, but please do not send questions to him: he no longer maintains these drivers.) The drivers handle Canon BJC-600, BJC-4xxx, BJC-70, Stylewriter 2x00, and BJC-800 printers.
The BJC-600 driver was written in the first place by Yoshio Kuniyoshi and later modified by Yves Arrouye. We tried to make it evolve synchronously, though Yoshio cannot be reached since a long time ago. The drivers are based on code for the HP printers by George Cameron (in fact, they are in the same file!), so he's the first person to thank.
The 2.00 version of the drivers was a complete rewrite of the driver (arguments, optimization, colour handling, in short: everything!) by Yves Arrouye. That release was also the first one to be able to use the full width of an A3 paper size. PostScript Printer Description (PPD) files for the drivers were released with version 2.15. They are incomplete, but they can be used to drive the printers' main features.
Configure the drivers by modifying the default values in the file gdevbjc.h or on the compilation line. If you don't do that, the drivers use reasonable defaults that make them work "as expected". All default values shown here are defined in that file.
By default, the drivers use the same algorithm as Ghostscript to convert CMYK colors to RGB. If you prefer to use Adobe formulas, define USE_ADOBE_CMYK_RGB when compiling. (See the top of the file gdevcdj.c to see the difference between the two.)
The drivers center the imageable area horizontally but not vertically, so that what can be printed does use the most of the output media. If you define BJC_DEFAULT_CENTEREDAREA when compiling, then the top and bottom margins will be the same, resulting in a (smaller) vertically centered imageable area also.
If you define USE_RECOMMENDED_MARGINS, then the top and bottom margins will be the same (that is, BJC_DEFAULT_CENTEREDAREA will be defined for you) and the margins will be the 12.4mm recommended by Canon. Since margins are complicated (because one must rely on the mechanical precision of the printer), the drivers do something about the bottom margin: by default the bottom margin is 9.54mm for the BJC-600 driver and 7mm for the BJC-800. If you define USE_TIGHT_MARGINS, then the bottom margin is 7mm for both drivers (but I never managed to get my own BJC-600 to print a line on this low bound, hence the larger default). Regardless of the presence of this definition, USE_FIXED_MARGINS will not allow the BJC-800 to use the lower 7mm bottom margin, so if you have a problem with the bottom margin on a BJC-800, just define that (without defining USE_TIGHT_MARGINS, of course).
A quick way to be sure the margins you selected is to print a file whose contents are:
%!
clippath stroke showpage
If the margins are okay, you will get a rectangle visibly surrounding the printable area. If they're not correct, one or more of the sides will be either incomplete or completely unprinted.
Make sure the bjc600 or bjc800 devices are in DEVICE_DEVS in the makefile; that is, look in the makefile for your platform and add them if necessary -- they may already be there. As of Ghostscript 5.10, for instance, one makefile has
DEVICE_DEVS6=bj10e.dev bj200.dev bjc600.dev bjc800.dev
There are two drivers here. The "bjc600" one supports the BJC-600 and BJC-4xxx (maybe the BJC-70 as well) and the "bjc800" one supports the BJC-800 series. Remarks here that apply to both drivers use the name "bjc".
Note: "options", "properties", and "parameters" designate the same thing: device parameters that you can change.
Giving an option an incorrect value causes an error. Unless stated otherwise, this error will be a rangecheckerror. Options may be set from the Ghostscript command line (using the -d and -s switches or other predetermined switches if they have an effect on the driver) or using the PostScript Level 2 setpagedevice operator if Ghostscript has been compiled with the level2 or level3 device (which it should ;-)). There are no special-purpose operators such as one was able to find in Level 1 printers.
The bjc uses 24 bits per pixel by default (unless you change the value of BJC_BITSPERPIXEL), corresponding to CMYK printing. Supported modes are 1 bpp and 4 bpp (gray levels), 8 bpp, 16 bpp, 24 bpp and 32 bpp (colours). Colours are preferably stored in the CMYK model (which means, for example, that with 16 bpp there are only 16 different shades of each color) but it is possible to store them as RGB color for some depths. Some modes do Floyd-Steinberg dithering and some don't, but use the default Ghostscript halftoning (in fact, when halftoning is used, dithering takes also place but because of the low point density it is usually not efficient, and thus invisible).
Descriptions of printing modes by bpp and Colors
bpp Colors Mode
32 4 CMYK colour printing, Floyd-Steinberg dithering 24 4 The same. (But each primary colour is stored on 6 bits instead of 8.) 24 3 RGB colour printing, Floyd-Steinberg dithering. This mode does not use the black cartridge (that's why it exists, for when you don't want to use it ;-)). Each primary colour is stored in 8 bits as in the 32/4 mode, but black generation and under-color removal are done on the driver side and not by Ghostscript, so you have no control over it. (This mode is no longer supported in this driver.) 16 4 CMYK colour printing, halftoned by Ghostscript. F-S dithering is still visible here (but the halftone patterns are visible too!). 8 4 The same. (But each primary colour is stored in 2 bits instead of 4.) 8 3 RGB colour printing. This mode is not intended for use. What I mean is that it should be used only if you want to use custom halftone screens and the halftoning is broken using the 8/4 mode (some versions of Ghostscript have this problem). 8 1 Gray-level printing, Floyd-Steinberg dithering 1 1 Gray-level printing halftoned by Ghostscript
These modes are selected using the BitsPerPixel and Colors integer options (either from the command line or in a PostScript program using setpagedevice). See below.
A note about darkness of what is printed: Canon printers do print dark, really. And the Floyd-Steinberg dithering may eventually darken your image too. So you may need to apply gamma correction by calling Ghostscript as in
gs -sDEVICE=bjc600 gamma.ps myfile.ps
where gamma.ps changes the gamma correction (here to 3 for all colors); 0.45 gives me good results, but your mileage may vary. The bigger the value the lighter the output:
{ 0.45 exp } dup dup currenttransfer setcolortransfer
The drivers support printing at 90dpi, 180dpi and 360dpi. Horizontal and vertical resolutions must be the same or a limitcheck error will happen. A rangecheck will happen too if the resolution is not 90 ×2^N. If the driver is compiled with -DBJC_STRICT a rangecheck also happens if the resolution is not one of those supported. This is not the case, as we expect that there may be a 720dpi bjc some day.
Here are the various options supported by the bjc drivers, along with their types, supported values, effects, and usage:
Note that when this is set for the first time, the Colors property is automatically adjusted unless it is also specified. The table here shows the corresponding color models and the rendering method visible: "GS" for Ghostscript halftoning and "F-S" for Floyd-Steinberg dithering. When both are present it means that the dithering of halftones is visible. Default choices are indicated by asterisk "*".
Valid Colors values for
allowed BitsPerPixel values
bpp Colors Color model Dithering
32 4 CMYK F-S 24 4 * CMYK F-S 3 RGB F-S 16 4 CMYK GS, F-S 8 4 * CMYK GS 3 RGB GS 1 K (CMYK) F-S 1 1 * K (CMYK) GS
Also note that automagical change of one parameter depending on the other one does not work in a setpagedevice call. This means that if you want to change BitsPerPixel to a value whose valid Colors values do not include the actual Colors value, you must change Colors too.
Note that setting this property does limit the choices of BitsPerPixel. As for the previous property, its first setting may induce a setting of the "other value" (BitsPerPixel here). The table here indicates valid combinations with "V", default values with asterisk "*".
Valid BitsPerPixel values
for allowed Colors values
BitsPerPixel OK values
Colors Type 32 24 16 8 1
4 CMYK V * V V 3 RGB * V 1 K V *
Also note that automagical change of one parameter depending on the other one does not work in a setpagedevice call. This means that if you want to change Colors to a value whose valid BitsPerPixel values don't include the actual BitsPerPixel value, you must change BitsPerPixel too.
| Value | bjc600 | bjc800 | ||||
|---|---|---|---|---|---|---|
| Low | X | Has the effect of making only two printing passes instead of four, so should be twice the speed; known as "CN" (Color Normal) mode | ||||
| Draft | X | X | Unlights the "HQ" light on a BJC-600 | |||
| Normal | X | X | Default for both drivers; lights the "HQ" light on a BJC-600 | |||
| High | X | X | Means 200% black and 100% CMY; lights the "Bk+" light on a BJC-600 | |||
Note that the MediaType and ThickMedia options will be replaced by the use of the device InputAttributes and OutputAttributes as soon as possible. Please note too that the print mode may be reset at the start of printing, not at the end. This is the expected behaviour. If you need to reset the printer to its default state, simply print a file that does just a showpage.
Here is other information published by the driver that you will find in the deviceinfo dictionary.
The BJC printers have top and bottom hardware margins of 3mm and 7.1mm respectively (Canon says 7mm, but this is unusable because of the rounding of paper sizes to PostScript points). The left margin is 3.4mm for A4 and smaller paper sizes, 6.4mm for U.S. paper sizes, envelopes and cards. It is 4.0mm for A3 paper on the BJC-800.
The maximum printing width of a BJC-600 printer is 203mm. The maximum printing width of a BJC-800 printer is 289mm on A3 paper, 203mm on U.S. letter and ISO A4 paper.
The files CBJC600.PPD and CBJC800.PPD (whose long names are, respectively, Canon_BubbleJetColor_600.ppd and Canon_BubbleJetColor_800.ppd) are PPD files to drive the features of the bjc600 and bjc800 drivers. They can be used, for example, on NextStep systems (presumably on OpenStep systems too) and on Unix systems with Adobe's TranScript and pslpr (not tested). The files are not complete at the moment. Please note that NextStep's printing interface does not correctly enforce constraints specified in these files (in UIConstraints descriptions): you must force yourself to use valid combinations of options.
By default the PPD files are set for U.S. letter size paper, and they use a normalized transfer function. If you choose to use A4 printing by default, you must replace "Letter" with "A4" in these (noncontiguous) lines:
[...]
*DefaultPageSize: Letter
[...]
*DefaultRegion: Letter
[...]
*DefaultImageableArea: Letter
[...]
Some versions of Ghostscript have problems with normalized colors, which makes them add magenta in gray levels. If you experience this problem, in the PPD file replace the line
*DefaultTransfer: Normalized
with the alternate line
*DefaultTransfer: Null
The "thick media" option is implemented by choosing a value of 120 or 80 (for thick and thin media respectively) for the MediaWeight feature of the drivers. If you ever change the threshold for thick media in the driver code, you may need to change the values in the PPD files too.
All customization should be done using the "*Include:" feature of PPD files so that your local changes will be retained if you update the PPD files.
Yves Arrouye no longer maintains this driver, and will not answer questions about it. If you are posting a question about it in a public form, please be as descriptive as possible, and please send information that can be used to reproduce the problem. Don't forget to say which driver you use, and in what version. Version information can be found in the source code of the driver or by issuing the following command in a shell:
echo "currentpagedevice /VersionString get ==" | gs -q -sDEVICE=bjc600 -
I am particularly grateful to Yoshio Kuniyoshi <yoshio@nak.math.keio.ac.jp> without whom I'd never make these drivers, and also to L. Peter Deutsch, who answered all my (often silly) questions about Ghostscript's driver interface.
Thanks also to the people who volunteered to beta-test the v2.x BJC drivers: David Gaudine <david@donald.concordia.ca>, Robert M. Kenney <rmk@unh.edu>, James McPherson <someone@erols.com> and Ian Thurlbeck <ian@stams.strath.ac.uk> (listed alphabetically) were particularly helpful by discovering bugs and helping find out exact paper margins on printers I don't have access to.
And many thanks to Klaus-Gunther Hess <ghess@elmos.de> for looking at the dithering code and devising a good CMYK dithering algorithm for the Epson Stylus Color, which I then adapted to the code of these drivers.
This section was contributed by Gunther Hess <ghess@elmos.de>, who also wrote uniprint, a later set of drivers. You should probably see the section on uniprint for whether it might be better for your uses than this driver.
This driver is selected with "-sDEVICE=stcolor", producing output for an Epson Stylus Color at 360dpi resolution by default. But it can do much more with this printer, and with significantly better quality, than with the default mode; and it can also produce code for monochrome versions of the printer. This can be achieved via either command-line options or Ghostscript input. For convenience a PostScript file is supplied for use as an initial input file. Try the following command:
gs -sDEVICE=stcolor -r{Xdpi}x{Ydpi} stcolor.ps {YourFile.ps}
where {Xdpi} is one of 180, 360, or 720 and {Ydpi} is one of 90, 180, 360, or 720. The result should be significantly better. You may use stcolor.ps with other devices too, but I do not recommend this, since it does nothing then. stcolor.ps should be available with binary distributions and should reside in the same directory as other Ghostscript initialization files or in the same directory as the files to be printed. Thus if Ghostscript is part of your printer-spooler, you can insert
(stcolor.ps) findlibfile { pop run } if pop
in files you want to use the improved algorithms. You may want to adapt stcolor.ps file to your specific needs. The methods and options for this are described here, but this description is restricted to Ghostscript options, while their manipulation at the PostScript level is documented in the material on the relationship of Ghostscript and PostScript and in stcolor.ps.
Now to explain the options (as written on my UNIX system). The order is somehow related to their use during the printing process:
gscmyk fast color output, CMYK process color model (default) gsmono fast monochrome output gsrgb fast color output, RGB process color model fsmono Floyd-Steinberg, monochrome fsrgb Floyd-Steinberg, RGB process color model (almost identical to the cdj550/bjc algorithm) fsx4 Floyd-Steinberg, CMYK process color model (shares code with fsmono and fsrgb, but is algorithmically really bad) fscmyk Floyd-Steinberg, CMYK process color model and proper modifications for CMYK hscmyk modified Floyd-Steinberg with CMYK model ("hs" stands for "hess" not for "high speed", but the major difference from fscmyk is speed) fs2 algorithm by Steven Singer (RGB) should be identical to escp2cfs2.
ESC/P2 allows any resolutions to be valid in theory, but only -r360x360 (the default) and -r720x720 (not on STC-IIs ? and st800) are known to work with most printers.
Valid option combinations
Stylus I & Pro-Series only
Resolution escp_Band Weave usable escp_Band &
number of passes
180x90 15 noWeave 180x180 1 , 8, 24 noWeave, Microweave 15/2 SoftWeave 180x360 15/4 SoftWeave 180x720 15/8 SoftWeave 360x90 15 noWeave 360x180 1, 8, 24 noWeave, Microweave 15/2 SoftWeave 360x360 1, 8, 24 noWeave, Microweave 15/4 SoftWeave 360x720 15/8 SoftWeave 720x90 15 noWeave 720x180 15/2 SoftWeave 720x360 15/4 SoftWeave 720x720 1 noWeave, Microweave 15/8 SoftWeave
Beware: there are only few validity checks for parameters. A good example is escp_Band: if you set this, the driver uses your value even if the value is not supported by the printer. You asked for it and you got it!
Quite a bunch of parameters. Hopefully you never need any of them, besides feeding stcolor.ps to Ghostscript in front of your input.
After answering some questions over fifty times I prepared a FAQ. Here is version 1.3 of the FAQ, as of stcolor version 1.20 (for Ghostscript 3.50).
Yes, this driver supports the A3-size printer: merely set the required pagesize and margins. A simple way to do this is to specify the command-line switch "-sPAPERSIZE=a3" or include the procedure call "a3" in the PostScript prolog section. To optimize the printable area or set the proper margins, see the next paragraph.
I refuse to add code to stcolor that tries to guess the proper margins or page size, because I found that such guessing is usually wrong and needs correction in either the source or the parameters. You can modify stcolor.ps to do that, however. After the line
mark % prepare stack for "putdeviceprops"
insert these lines, which define page size and margins in points:
/.HWMargins [9.0 39.96 12.6 9.0] % Left, bottom, right, top (1/72") /PageSize [597.6 842.4] % Paper, including margins (1/72") /Margins [ % neg. Offset to Left/Top in Pixels 4 index 0 get STCold /HWResolution get 0 get mul 72 div neg 5 index 3 get STCold /HWResolution get 1 get mul 72 div neg ]
Feel free to change the values of .HWMargins and PageSize to match your needs; the values given are the defaults when the driver is compiled with "-DA4". This option or its omission may cause trouble: the Stylus Color can print up to exactly 8 inches (2880 pixels) at 360dpi. The remaining paper is the margin, where the left margin varies only slightly with the paper size, while the right margin is significantly increased for wider paper, such as U.S. letter size.
Note that if you are using an ISO paper size with a version of stcolor after 1.20 and compiled without "-DA4", then the default margin is too large, and you need to add the proper ".HWMargins" to the command line or to stcolor.ps.
First the good news: the driver can print on the Stylus Color II. Now the bad news:
To make things work, you MUST disable the driver's internal weaving (Softweave), in one of these two ways:
gs -dMicroweave ...
gs -dnoWeave -descp_Band=1 ...
Version 1.90, current as of Ghostscript 5.10, fixes this bug by new default behaviour. I experienced significantly increased printing speed with the second variant on the old Stylus Color, when printing mostly monochrome data.
The next section is a contribution from Jason Patterson <jason@reflections.com.au> who evaluated a previous version (1.17). Ghostscript was invoked as follows:
- gs
- -sDEVICE=stcolor -r720x720 -sDithering=... -sOutputFile=escp.out stcolor.ps whatsoever.ps
where "..." is the name of the desired algorithm. stcolor.ps was omitted for the gs-algorithms (gsmono, gsrgb and gscmyk), for which it is useless and would not allow the selection of "gscmyk".
Here are data about the EPSON Stylus Color driver's different dithering methods, based on a little experiment using four good quality scanned images of quite varied nature, to begin with, a summary of the results of the four experiments. Sanity note: the results here are from only four images and a total of 24 printouts (eight on 720dpi paper, sixteen on plain paper). Your results will almost certainly vary, and your standards might not be the same as mine, so use these results only as a guide, not as a formal evaluation.
Quality of output by method
gsmono Pretty much what you'd expect from a mono ordered pattern. Looks like what a lot of mono laser printers produce. fsmono Excellent for monochrome. gscmyk Not very good, but expected from an ordered pattern. gsrgb A little better than gscmyk. More consistent looking. fs2 Good, but not quite as good as fsrgb. Gets the brightness wrong: too light at 720dpi, too dark at 360dpi. fsrgb Very good, but a little too dark and has a slight blue tint. hscmyk Excellent. Slightly better than fsrgb and fs2. Better than fscmyk on some images, almost the same on most. fscmyk Best. Very, very slightly better than hscmyk. On some images nearly as good as the EPSON demos done with the MS Windows driver.
Overall visual quality (1-10), best to worst
Monochrome fsmono ****************** gsmono ********** 0 1 2 3 4 5 6 7 8 9 10
Colour fscmyk ******************* hscmyk ******************* fsrgb ****************** fs2 ***************** gsrgb ********** gscmyk ********* 0 1 2 3 4 5 6 7 8 9 10
In the initial version of the driver distributed with Ghostscript 3.33, the parameter "SpotSize" was the only way to manipulate the colors at the driver level. According to the parameters enumerated above, this has changed significantly with version 1.16 and above as a result an ongoing discussion about dithering algorithms and "false color" on the Epson Stylus Color. This initiated the transformation of the stcolor driver into a framework for different dithering algorithms, providing a generalized interface to the internal Ghostscript color models and the other data structures related to Ghostscript drivers.
The main thing such a framework should be able to do is to deliver the values the dithering algorithm needs; and since this directly influences the optical image impression, this transformation should be adjustable without the need for recompilation and relinking. In general the process can be described as follows:
ColorAdjustMatrix Coding Transfer | Ghostscript color | => | Ghostscript raster | => | Dithering data | | | | 1/2/4/8/16/32-bit | | | | 1/3/4 16-bit values | | 1/3/4 values | | (arbitrary type) |
Due to the limitations on raster storage, information is lost in the first transformation step, except for the 16-bit monochrome mode. So any color adjustment should take place before this step and this is where the optional ColorAdjustMatrix works.
The first transformation step, called "coding", is controlled by the ?coding arrays. The decoding process expands the range of values expontentially to a larger range than that provided by the initial Ghostscript color model, and is therefore a reasonable place to make device- or algorithm-specific adjustments. This is where the ?transfer arrays are used. Array access might be not the fastest method, but its generality is superior, so this step is always based upon internally algorithm-specific array access. If 8 bits are stored per color component and if the algorithm uses bytes too, the second transformation is included within the first, which saves significant computation time when printing the data.
The driver supports different values for ProcessColorModel, which raises the need for different color adjustments. Here "CAM" stands for "ColorAdjustMatrix".
if ((r == g) && (g == b)) K' = 1.0 - R; else K' = 1.0 - CAM[0] * R + CAM[1] * G + CAM[2] * B;
According to the documentation on drivers, the latter (the "else" clause) should never happen.
if((r == g) && (g == b)) R' = B' = G' = R; else R' = CAM[0]*R + CAM[1]*G + CAM[2]*B; G' = CAM[3]*R + CAM[4]*G + CAM[5]*B; B' = CAM[6]*R + CAM[7]*G + CAM[8]*B;
The printer always uses four inks, so a special treatment of black is provided. Algorithms may take special action if R, G, and B are all equal.
if((c == m) && (m == y))
K' = max(C,K);
C' = M' = Y' = 0;
else
K = min(C,M,Y);
if((K > 0) && ColorAdjustMatrix_present) { => UCR
C -= K;
M -= K;
Y -= K;
}
C' = CAM[ 0]*C + CAM[ 1]*M + CAM[ 2]*Y + CAM[ 3]*K;
M' = CAM[ 4]*C + CAM[ 5]*M + CAM[ 6]*Y + CAM[ 7]*K;
Y' = CAM[ 8]*C + CAM[ 9]*M + CAM[10]*Y + CAM[11]*K;
K' = CAM[12]*C + CAM[13]*M + CAM[14]*Y + CAM[15]*K;
Again we have a special black treatment. "max(C,K)" was introduced because of a slight misbehaviour of Ghostscript, which delivers black under certain circumstances as (1,1,1,0). Normally, when no special black separation and undercolor removal procedures are defined at the PostScript level, either (C,M,Y,0) or (0,0,0,K) values are mapped. This would make the extended ColorAdjustMatrix quite tedious, and so during mapping, black separation is done for (C,M,Y,0) requests; and if there is a ColorAdjustMatrix, undercolor removal is used too. In other words the default matrix is:
1 0 0 1 0 1 0 1 0 0 1 1 0 0 0 1
and it is applied to CMYK values with separated and removed black. Raising the CMY coefficients while lowering the K coefficients reduces black and intensifies color. But be careful, because even small deviations from the default cause drastic changes.
If no ColorAdjustMatrix is set, the matrix computations are skipped. Thus the transformation reduces to range inversion in monochrome mode and black separation in CMYK mode.
These two (groups of) parameters are arrays of floating-point numbers in the range 0.0 to 1.0. They control the truncation to the desired number of bits stored in raster memory (BitsPerPixel) and the ink density. The "truncation" may become a nonlinear function if any of the ?coding arrays is set. Assume the following Ghostscript invocation:
gs -sDEVICE=stcolor -sDithering=fscmyk -dBitsPerPixel=16 \
-dKcoding='{ 0.0 0.09 0.9 1.0 }' \
-dMcoding='{ 0.0 0.09 0.9 1.0 }' \
-dKtransfer='{ 0.0 0.09 0.9 1.0 }' \
-dYtransfer='{ 0.0 0.09 0.9 1.0 }'
We may have either or both of ?coding and ?transfer, giving four possible combinations. (These four combinations appear in the given example.) The resulting mapping appears in the following tables, where except for the internal Indices (4 components × 4 bits = 16 BitsPerPixel), all values are normalized to the range 0 to 1. The actual range is 0 to 65535 for the Ghostscript color and 0 to 16777215 for the ink values delivered to the fscmyk algorithm. Sorry for the bunch of numbers following, but you may try this example in conjunction with stcinfo.ps, which should give you a graphical printout of the following numbers when you issue a showpage command.
Cyan
Magenta
CI/15 gs_color_values CI Ink gs_color_values CI Ink
0.000 0.000 - 0.062 0 0.000 -0.123 - 0.123 0 0.000 0.067 0.063 - 0.125 1 0.067 0.123 - 0.299 1 0.247 0.133 0.125 - 0.187 2 0.133 0.299 - 0.365 2 0.351 0.200 0.188 - 0.250 3 0.200 0.365 - 0.392 3 0.379 0.267 0.250 - 0.312 4 0.267 0.392 - 0.420 4 0.406 0.333 0.313 - 0.375 5 0.333 0.420 - 0.447 5 0.433 0.400 0.375 - 0.437 6 0.400 0.447 - 0.475 6 0.461 0.467 0.438 - 0.500 7 0.467 0.475 - 0.502 7 0.488 0.533 0.500 - 0.562 8 0.533 0.502 - 0.529 8 0.516 0.600 0.563 - 0.625 9 0.600 0.529 - 0.557 9 0.543 0.667 0.625 - 0.687 10 0.667 0.557 - 0.584 10 0.571 0.733 0.688 - 0.750 11 0.733 0.584 - 0.612 11 0.598 0.800 0.750 - 0.812 12 0.800 0.612 - 0.639 12 0.626 0.867 0.813 - 0.875 13 0.867 0.639 - 0.715 13 0.653 0.933 0.875 - 0.937 14 0.933 0.715 - 0.889 14 0.778 1.000 0.938 - 1.000 15 1.000 0.889 - 1.111 15 1.000
The difference between cyan and magenta is the presence of a coding array. The coding process must map a range of color values to each of the sixteen component indices. If no coding array is given, this is accomplished by dividing by 4096, equivalent to a right shift by 12 bits. The final ink density resides in the given interval and moves from the left to the right side from 0 to 15. For magenta there is a coding array and the ink value matches the center of the intervals. But the distribution of the mapped intervals follows the given coding array and is nonlinear in the linear color space of Ghostscript.
Now let us take a look at the case with transfer arrays:
Yellow
Black
CI/15 gs_color_values CI Ink gs_color_values CI Ink
0.000 0.000 - 0.062 0 0.000 -0.123 - 0.123 0 0.000 0.067 0.063 - 0.125 1 0.018 0.123 - 0.299 1 0.067 0.133 0.125 - 0.187 2 0.036 0.299 - 0.365 2 0.133 0.200 0.188 - 0.250 3 0.054 0.365 - 0.392 3 0.200 0.267 0.250 - 0.312 4 0.072 0.392 - 0.420 4 0.267 0.333 0.313 - 0.375 5 0.090 0.420 - 0.447 5 0.333 0.400 0.375 - 0.437 6 0.252 0.447 - 0.475 6 0.400 0.467 0.438 - 0.500 7 0.414 0.475 - 0.502 7 0.467 0.533 0.500 - 0.562 8 0.576 0.502 - 0.529 8 0.533 0.600 0.563 - 0.625 9 0.738 0.529 - 0.557 9 0.600 0.667 0.625 - 0.687 10 0.900 0.557 - 0.584 10 0.667 0.733 0.688 - 0.750 11 0.920 0.584 - 0.612 11 0.733 0.800 0.750 - 0.812 12 0.940 0.612 - 0.639 12 0.800 0.867 0.813 - 0.875 13 0.960 0.639 - 0.715 13 0.867 0.933 0.875 - 0.937 14 0.980 0.715 - 0.889 14 0.933 1.000 0.938 - 1.000 15 1.000 0.889 - 1.111 15 1.000
Yellow uses a transfer array. There is no linear correspondence between the color and the ink values: this correspondence is defined through the given array. In other words, the transfer arrays define a nonlinear ink characteristic, which is exactly the same functionality that PostScript's "(color)transfer" function provides.
While for yellow the intervals match the intervals used with cyan, for black the intervals match the magenta intervals. But watch the correspondence between the CI/15 values and the ink density for black: this is a linear distribution in the ink domain.
Not a bad idea, I think. Consider the fs2 algorithm: it uses values in the range 0 to 255. If any transfer array were alone, some of the 256 possible values would never be used and others would be used for adjacent intervals several times. Establishing an identical coding array solves this problem, so the full potential of the algorithm is used.
Another useful feature of the coding arrays is that they are internally normalized to the range 0-1. In 720x720dpi mode the transfer arrays in stcolor.ps limit the dot density to about 50%, so these arrays end at 0.5 (and begin at 0.5 for RGB). Because of automatic normalization, these arrays can also be used as coding arrays. But of course in the fs2 case mentioned above, values from 0 to 127 will never be delivered to the algorithm, while values 128-255 are delivered for adjacent intervals.
To clarify the intended use of the three parameters (parameter groups), keep this in mind:
- ColorAdjustMatrix is never used when transferring gray values. This restricts it to what the name says: adjustment of colors, that is, correction for miscolored ink. Do not use it for saturation or brightness control.
- ?transfer arrays control the values delivered to the driver, which in turn controls the ink quantity. Use these arrays to control saturation and brightness. In general these arrays are identical for all inks. If they differ they provide a simpler scheme for color correction, which is not necessarily faster than the ColorAdjustMatrix.
- ?coding arrays control the color value intervals mapped to the internal color indices.
The Epson Stylus Color has a head assembly that contains two physically identifiable heads, one for black and one for cyan, magenta, and yellow (CMY). This makes four "logical" heads, one for each color component. Each of these four heads has several jets at some vertical (Y) distance from one another, so several horizontal lines can be printed of a given color during one pass of the heads. From experience I think there are fifteen jets per color, spaced at 1/90in.
So the question arises of how to print at a Y resolution of 360dpi with 90dpi jets. Simply by division one gets 360dpi/90dpi = 4, which tells us that 4 passes of the head assembly are needed to achieve a Y resolution of 360dpi.
Weaving is the method of how the fifteen jets are used to print adjacent horizontal rows separated here by 1/360 inch:
Print-head jets used with and without weaving
Weaving
noWeave
Pass 1 2 3 4 1 2 3 4 Row 0 jet 0 -- -- -- jet 0 -- -- -- 1 -- jet 1 -- -- -- jet 0 -- -- 2 -- -- jet 2 -- -- -- jet 0 -- 3 -- -- -- jet 3 -- -- -- jet 0 4 jet 1 -- -- -- jet 1 -- -- -- 5 -- jet 2 -- -- -- jet 1 -- -- 6 -- -- jet 3 -- -- -- jet 1 -- ...
Now let's assume that the dot diameter is different for each individual jet, but the average among the jets matches the desired resolution. With weaving, adjacent rows are printed by different jets, thus some averaging takes place. Without weaving, adjacent rows are printed by the same jet and this makes the dot diameter deviations visible as 1/90in stripes on the paper.
The parameters "Unidirectional", "Microweave", "noWeave", "OutputCode", "Model" and the given resolution control the data generated for the printer.
Simply toggles the unidirectional mode of the printer. Setting "Unidirectional" definitely slows printing speed, but may improve the quality. I use this for printing transparencies, where fast head movement could smear the ink.
The first are two booleans, which implies that four combinations are possible. Actually only three exist (if you don't count for deltarow): Softweave, Microweave, and noWeave. The first and second are functionally identical, the difference being whether the driver or the printer does the job.
In the default Softweave mode the driver sends the data properly arranged to the printer, while in Microweave mode, it is the printer that arranges the data. But in general the host processor is much faster than the printer's processor, and thus it is faster for the host do the job. In addition to that, for 720dpi eight passes are required, and the amount of buffer space needed to buffer the data for the passes is far beyond the printer's memory. Softweave requires an odd value of "escp_Band"; the Stylus Color provides fifteen for that.
"OutputCode" controls the encoding used. In the basic modes, the choice consists of "plain" and "runlength". The computation of runlength-encoded data does not take much time, less than the data tranfer to the printer; thus this is the recommended mode, and of course the default. With the Stylus Color, Epson introduced some new encoding principles, namely "tiff" and "deltarow". While the first was omitted from this driver for lack of apparent advantages, "deltarow" is available as an option. "Softweave" cannot be used with this encoding, so if OutputCode=deltarow is set, Microweave becomes the default. Maybe that the size of the ESC/P2 code becomes smaller, but I have never observed faster printing speed. Things tend to become slower with deltarow compared to Softweave.
Some ESC/P2 printers such as the Stylus 800 do not offer Microweave or the commands required to do Softweave. Setting Model just changes the defaults and omits some parts of the initialization sequence which are not compatible with the given printer model. Currently only "st800" is supported besides the default stcolor.
This section gives an overview of performance in terms of processing and printing times, from tests run after version 1.13. Printing was done offline (simply copying a processed file to the printer) to measure real printing speed without regard to speed of processing on the host, since at high resolutions, processing time is the same order of magnitude and thus may become the limiting factor.
I ran several files though Ghostscript and recorded the size of the resulting print code, the processing time, and the printing time, at least for some of the files, always using these options:
gs -sDEVICE=stcolor -sPAPERSIZE=a4 stcolor.ps - < file.ps
(Actually "-sPAPERSIZE=a4" is in my gs_init.ps since I'm a germ.)
"deltarow" is the new encoding principle ("ESC . 3 10 10 1") with Microweave on. It is activated with "-sOutputCode=deltarow".
"Softweave" actually means that nothing else was used: it is the default, and implies that odd v=40/h=10/m=15 mode ("ESC . 1 40 10 15").
"Microweave" means "-dMicroweave", equivalent to "ESC . 1 10 10 1", with full skip optimization and microweave activated.
Finally I wanted to see the plain Kathy Ireland, and used "-sOutputCode=plain", which just replaces runlength encoding (RLE) by no encoding, thus using "ESC . 0 40 10 15". [So sorry ;-) Kathy was still dressed in blue in front of the blue sea on a blue air cushion -- nice to see but hard to dither.]
So here are the results.
File sizes and printing speeds with various weaving methods
golfer.ps colorcir.ps drawing.ps brief.ps
deltarow 572751/48.180u 643374/41.690u 90142/46.180u/1:50 178563/49.350u/2:22 Softweave 559593/46.810u 669966/44.960u 296168/48.160u/1:30 269808/43.320u/1:55 Microweave 590999/56.060u 754276/42.890u 338885/47.060u/1:50 282314/44.690u/2:22
Kathy Ireland
kathy.ps
deltarow 3975334/111.940u/5:35 Softweave 3897112/101.940u/3:10 Microweave 4062829/100.990u/3:15 plain/soft 5072255/104.390u/3:05
It may be that I've not chosen the optimal deltarow code, but even if it saves at lot of bytes, printing-speed is not increased.
At least the printer prefers plain Kathy. In other words, sending 1 Megabyte or 20% more data has no impact on printing speed. drawing.ps is an exception to this rule: plain prints slower than RLE.
"Unclever" coding -- especially with deltarow -- can significantly slow down printing. But even if very significant advantages in the size of the code are achieved, "deltarow" is not competitive. colorcir.ps shows savings with deltarow, but printing is a mess.
Full page halftone images printed, unless otherwise noted.
Printing time related to other options
dpi Print mode Size KB Time Comments
180x180 mono -/uni 358 1:15 -/bi 358 0:45 micro/bi 205 0:45 Not weaving soft/bi 179 1:25 color -/bi 641 2:45 soft/bi 556 1:32 360x360 mono -/uni 269 0:50 Monochrome text -/bi 269 0:35 Monochrome text micro/bi 269 2:25 Monochrome text soft/uni 250 3:15 Monochrome text soft/bi 250 1:55 Monochrome text color -/bi 346 1:00 Sparse-color page, visible displacements micro/bi 346 1:50 Sparse-color page, looks buggy -- printer? soft/bi 294 1:30 Sparse-color page, O.K. -/bi 2218 2:45 Visible stripes micro/bi 5171 3:17 soft/bi 3675 3:05 360x720 mono soft/bi 2761 5:40 color soft/bi 7789 6:15 Just a small difference! 720x360 color soft/bi 7182 5:40 720x720 color micro/bi 14748 30:26 Actually beyond printer's capabilities soft/bi 14407 11:08
This driver was copied from gdevcdj.c (Ghostscript 3.12), which was contributed by George Cameron, Koert Zeilstra, and Eckhard Rueggeberg. Some of the ESC/P2 code was drawn from Richard Brown's gdevescp.c. The POSIX interrupt code (compilation option -DSTC_SIGNAL) is from Frederic Loyer. Several improvements are based on discussions with Brian Converse, Bill Davidson, Gero Guenther, Jason Patterson, ? Rueschstroer, and Steven Singer.
While I wish to thank everyone mentioned above, they are by no means responsible for bugs in the stcolor driver -- just for the features.
Gunther Hessuniprint is a unified parametric driver by Gunther Hess <ghess@elmos.de> for several kinds of printers and devices, including
- any Epson Stylus Color, Stylus, or Stylus Pro
- HP PCL/RTL
- Canon BubbleJet Color 610
- NEC P2X
- Sun raster file format
This driver is intended to become a unified printer driver. If you consider it ugly, please send me your suggestions for improvements. The driver will be updated with them. Thus the full explanation of the driver's name is:
Ugly- -> Updated- -> Unified Printer Driver
But you probably want to know something about the functionality. At the time of this writing uniprint drives:
It can be configured for various other printers without recompilation and offers uncompressed (ugly) Sun rasterfiles as another format, but this format is intended for testing purposes rather than real use. The usage of this driver is quite simple. The typical command line looks like this:
gs @{MODEL}.upp -sOutputFile={printable file} MyFile.ps -c quit
For example, from my Linux box:
gs @stc.upp -sOutputFile=/dev/lp1 tiger.ps -c quit
Unified Printer Parameter files distributed with Ghostscript
Canon BJC 610 (color, rendered) bjc610a0.upp 360×360dpi plain paper, high speed bjc610a1.upp 360×360dpi plain paper bjc610a2.upp 360×360dpi coated paper bjc610a3.upp 360×360dpi transparency film bjc610a4.upp 360×360dpi back print film bjc610a5.upp 360×360dpi fabric sheet bjc610a6.upp 360×360dpi glossy paper bjc610a7.upp 360×360dpi high gloss film bjc610a8.upp 360×360dpi high resolution paper bjc610b1.upp 720×720dpi plain paper bjc610b2.upp 720×720dpi coated paper bjc610b3.upp 720×720dpi transparency film bjc610b4.upp 720×720dpi back print film bjc610b6.upp 720×720dpi glossy paper bjc610b7.upp 720×720dpi high-gloss paper bjc610b8.upp 720×720dpi high resolution paper HP Ink-Printers cdj550.upp 300×300dpi 32-bit CMYK cdj690.upp 300×300dpi Normal mode cdj690ec.upp 300×300dpi Economy mode dnj750c.upp 300×300dpi Color -- also good for 450C dnj750m.upp 600×600dpi Monochrome NEC P2X necp2x.upp 360×360dpi 8-bit (Floyd-Steinberg) Any Epson Stylus Color stcany.upp 360×360dpi 4-bit, PostScript halftoning stcany_h.upp 720×720dpi 4-bit, PostScript halftoning Original Epson Stylus and Stylus Pro Color stc.upp 360×360dpi 32-bit CMYK, 15-pin stc_l.upp 360×360dpi 4-bit, PostScript halftoning, weaved noWeave stc_h.upp 720×720dpi 32-bit CMYK, 15-pin Weave Epson Stylus Color II stc2.upp 360×360dpi 32-bit CMYK, 20-pin, Epson Stylus Color II(s) stc2_h.upp 720×720dpi 32-bit CMYK, 20-pin, Epson Stylus Color II stc2s_h.upp 720×720dpi 32-bit CMYK, 20-pin, Epson Stylus Color IIs Epson Stylus Color 200 stc200.upp 360×720dpi Plain Paper Epson Stylus Color 300 stc300.upp 360×360dpi 32-bit CMYK, plain paper stc300bl.upp 180×180dpi black only, plain paper stc300bm.upp 360×360dpi black only, plain paper Epson Stylus Color 500 (good transfer curves for plain paper) stc500p.upp 360×360dpi 32-bit CMYK, noWeave, plain paper stc500ph.upp 720×720dpi 32-bit CMYK, noWeave, plain paper Epson Stylus Color 600, 32/90-inch weaving stc600pl.upp 360×360dpi 32-bit CMYK, 32-pin, plain paper stc600p.upp 720×720dpi 32-bit CMYK, 32-pin, plain paper stc600ih.upp 1440×720dpi 32-bit CMYK, 30-pin, inkjet paper Epson Stylus Color 640 stc640p.upp 720×720dpi plain paper? st640p.upp 720×720dpi CMYK, plain paper st640pg.upp 720×720dpi grayscale, plain paper st640pl.upp 360×360dpi CMYK, plain paper st640plg.upp 360×360dpi grayscale, plain paper st640ih.upp 1440×720dpi CMYK, inkjet paper st640ihg.upp 1440×720dpi grayscale, inkjet paper Epson Stylus Color 800, 64/180-inch weaving stc800pl.upp 360×360dpi 32-bit CMYK, 64-pin, plain paper stc800p.upp 720×720dpi 32-bit CMYK, 64-pin, plain paper stc800ih.upp 1440×720dpi 32-bit CMYK, 62-pin, inkjet paper stc1520.upp 1440×720dpi 32-bit CMYK, 62-pin, inkjet paper Sun raster file ras1.upp 1-bit monochrome (Ghostscript) ras3.upp 3-bit RGB (Ghostscript) ras4.upp 4-bit CMYK (Ghostscript) ras8m.upp 8-bit grayscale (Floyd-Steinberg) ras24.upp 24-bit RGB (Floyd-Steinberg) ras32.upp 32-bit CMYK (CMYK-Floyd-Steinberg)
Thanks to Danilo Beuche, Guido Classen, Mark Goldberg and Hans-Heinrich Viehmann for providing the files for the stc200, hp690, stc500 and the stc640. Thanks to Michael Lossin <losse@germanymail.com> for the newer st640 parameter sets.
Please note the following:
- Changing the resolution with Ghostscript's -r switch is usually not possible.
- For Epson Stylus Color models not listed above, the two stc500 variants are likely to work in addition to stcany, but their gamma correction might be wrong.
The coding of uniprint was triggered by the requirements of the various Stylus Color models and some personal needs for HP and NEC drivers. Thus the Epson models are well represented among the distributed parameter files. When this driver entered the beta test phase, three other drivers appreared on the scene that could be at least partially integrated into uniprint: cdj850 by Uli Wortmann (available at http://www.erdw.ethz.ch/~bonk/hp850/hp850.html), hpdj by Martin Lottermoser, and bjc610 by Helmut Riegler.
Uli addresses features of the more recent DeskJet models that will not be available in uniprint soon. Martin taught me a lesson on HP-PCL3 headers that will be available in uniprint soon. Helmut in turn followed an almost similar idea, but targetted primarily for printing on Canon printers from the pbmplus library. Starting with version 1.68 of uniprint, BJC support is available. Work on the hpdj integration will start after the update of my website.
uniprint is actually an update of stcolor, but much more versatile than its predecessor; stcolor, in its turn, started as a clone of the color DeskJet family of drivers (cdj*). Finally, cdj* can be considered an addition of features to the simpler monochrome drivers of Ghostscript. This addition of features is useful to get an idea of the functionality of uniprint:
- Monochrome to advanced color (cdj*):
- This adds color mapping and rendering functions to the driver. Error diffusion is especially important for the quality of printing.
- HP color to Epson Color (stcolor)
- The Epson Stylus Color offered two features simultaneously: it could produce 720×720dpi output and it could soak the paper. In other words, it required more color management features inside the driver. This is still the major conceptual difference in the data generation for HP and Epson printers.
- Weaving techniques (stcolor)
- Besides the internal color management, the Stylus Color did not provide enough buffer space to operate the printer fast at 720×720dpi. The use of weaving could yield triple the print speed. Weaving, also called interleaving, is present in some monochrome drivers too. The new thing in stcolor was the combination with error diffusion. Unfortunately the weaving was somehow hard-coded, as the problems with the newer members of the Stylus Color family of printers demonstrated.
- Generalized output format and weaving (uniprint)
- The features mentioned above yield about 90% of stcolor's source code; only 10% is related to the formatting of the output. The idea to make the output format switchable came up soon after completing stcolor, but its final design was triggered by the (personal) necessity to drive a NEC P2X and a Designjet 750c.
Thus uniprint accumulates almost any features that can be found among the other printer drivers, which clearly has some disadvantage in processing speed -- true in particular of version 1.75, since it was targetted for functionality, and several speed-gaining features were (knowingly) omitted.
To summarize and to introduce the terms used in the description of the parameters, the features of uniprint that can be parameterized are:
- color mapping,
- color rendering (error diffusion or Floyd-Steinberg),
- output format, including
- weaving.
Here is one of the distributed parameter files (stc_l.upp) with some added comments. Also see the section that describes all uniprint's parameters in brief.
-supModel="Epson Stylus Color I (and PRO Series), 360x360DpI, noWeave"
-sDEVICE=uniprint -- Select the driver
-dNOPAUSE -- Useful with printers
-dSAFER -- Provides some security
-dupColorModel=/DeviceCMYK -- Selects the color mapping
-dupRendering=/ErrorDiffusion -- Selects the color rendering
-dupOutputFormat=/EscP2 -- Selects the output format
-r360x360 -- Adjusts the resolution
-dupMargins="{ 9.0 39.96 9.0 9.0}" -- Establishes (L/B/R/T margins in points)
-dupComponentBits="{1 1 1 1}" -- Map: bits per component (default: 8)
-dupWeaveYPasses=4 -- Weave: Y-passes (default: 1)
-dupOutputPins=15 -- Format/weave: scans per Command
-dupBeginPageCommand="< -- Goes to the printer
1b40 1b40 -- ESC '@' ESC '@' -> dual reset
1b2847 0100 01 -- ESC '(' 'G' 1 0 1 -> graphics
1b2869 0100 00 -- ESC '(' 'i' 1 0 1 -> no HW weave
1b2855 0100 0A -- ESC '(' 'U' 1 0 10 -> 360dpi
1b5500 -- ESC 'U' 0 -> bidir print
1b2843 0200 0000 -- ESC '(' 'C' 2 0 xx -> page length
1b2863 0400 0000 0000 -- ESC '(' 'c' 4 0 xxxx -> margins
>" -- as it is, unless:
-dupAdjustPageLengthCommand -- Adjust page length in BOP requested
-dupAdjustTopMarginCommand -- Adjust top margin in BOP
-dupAdjustBottomMarginCommand -- Adjust bottom margin in BOP
-dupEndPageCommand="(\033@\014)" -- Last (but one) data to the printer
-dupAbortCommand="(\033@\15\12\12\12\12 Printout-Aborted\15\014)"
That's short, and if one removes upWeaveYPasses and upOutputPins it becomes shorter, almost stcany.upp. This miniature size is because I am most familiar with ESC/P2, and was able to add defaults for the omitted parameters. Now a few notes about the parameters used in this example:
/SunRaster creates Sun raster files and requires no other parameters /Epson is used for the elderly ESC/P format (used by many printers) /EscP2 is used by more recent Epson printers (no X weaving supported) /EscP2XY supports X-Weaving, used with 1440dpi printers and in stc2s_h /Pcl HP PCL/RTL-style output formatter without weaving
/DeviceGray /DeviceRGBW /DeviceRGB /DeviceCMYK /DeviceCMYKgenerate
0 White White Red Black Black 1 -- Red Green Cyan Cyan 2 -- Green Blue Magenta Magenta 3 -- Blue -- Yellow Yellow This order may not be suitable for some printers, so another parameter "upOutputComponentOrder", also an array of integers, selects the output order using the numbers on the left.
One group of very important parameters not used in the example above deserves to be mentioned here: the transfer arrays, named "up{color}Transfer", where {color} is one of the names in the table above. These are arrays of floats in the range 0.0 - 1.0 representing the color transfer functions. They are used during mapping and rendering. In the simplest case, these arrays ensure an equidistant distribution of the stored values within the device space (which means a nonlinear mapping from Ghostscript's point of view). If the given array does not cover the entire range from 0 to 1, which applies for the Stylus Color family at high resolution for some media, only the relevant part gets mapped to raster memory (meaning that is's fully utilized) and the rendering takes care of the "overhang" (in this case the post-diffusion of 1-bit components makes sense).
Finally an important note on the transfer arrays: for monochrome devices the stored component is White, which is the way PostScript defines these devices, but most printers require Black. Thus one has to provide a falling "upWhiteTransfer" for such printers.
If no other weave parameters are given, uniprint computes several defaults which together do no weaving. The /Epson and /EscP2XY formats take care of "upWeaveXPasses" too.
Normally uniprint does not change the "upBeginPageCommand", nor does it provide a default. However, if the above boolean values are set, the corresponding values are changed (provided that the code of the formatters supports this change and the commands to be adjusted are included in the BOP string).
For the ESC/P(2) formats all commands represent binary data, while for the PCL/RTL formatter some of them are formats for fprintf. These strings must explicitly have a trailing "\0'.
I should write more, but the only recommendation is to take a look at the various parameter files. Here are a few more hints.
This table gives a brief explanation of every parameter known to uniprint, listing them in alphabetical order. "[ ]" denotes that a parameter is an array, and "(RO)" that it is read-only.
All uniprint parameters
Parameter Type Use
upAbortCommand String End of page and file on interrupt upAdjustBottomMarginCommand Bool Manipulate bottom margin in upBeginPageCommand upAdjustMediaSizeCommand Bool Manipulate Mediasize [intended] upAdjustPageLengthCommand Bool Manipulate page length in upBeginPageCommand upAdjustPageWidthCommand Bool Manipulate page width in upBeginPageCommand upAdjustResolutionCommand Bool Manipulate resolution upAdjustTopMarginCommand Bool Manipulate top margin in upBeginPageCommand upBeginJobCommand String Begin each output file upBeginPageCommand String Begin each page upBlackTransfer Float[ ] Black transfer (CMYK only!) upBlueTransfer Float[ ] Blue transfer upColorInfo Int[ ] struct gx_device_color_info upColorModel Name Select color mapping upColorModelInitialized Bool (RO) Color mapping OK upComponentBits Int[ ] Bits stored per component upComponentShift Int[ ] Positioning within gx_color_index upCyanTransfer Float[ ] Cyan transfer upEndJobCommand String End each file unless upAbortCommand upEndPageCommand String End each page unless upAbortCommand upErrorDetected Bool (RO) Severe (VM) error, not fully operational upFSFixedDirection Bool Inhbits direction toggling in rendering upFSProcessWhiteSpace Bool Causes white-space rendering upFSReverseDirection Bool Run rendering in reverse (if fixed) upFSZeroInit Bool Non-random rendering initialization upFormatXabsolute Bool Write absolute X coordinates upFormatYabsolute Bool Write absolute Y coordinates upGreenTransfer Float[ ] Green transfer upMagentaTransfer Float[ ] Magenta transfer upMargins Float[ ] L/B/R/T margins in points upModel String Comment string, holds some info upOutputAborted Bool (RO) Caught an interrupt upOutputBuffers Int Number of rendering buffers (2^N) upOutputComponentOrder Int[ ] Order of components when printing upOutputComponents Int Number of written components, not fully operational upOutputFormat Name Select output format upOutputFormatInitialized Bool (RO) Format data OK upOutputHeight Int Output height in pixels upOutputPins Int Number of pins / nozzles per command upOutputWidth Int Output width in pixels upOutputXOffset Int Offset in pixels, if upFormatXabsolute upOutputXStep Int Divisor or multiplier for X coords upOutputYOffset Int Offset in pixels, if upFormatYabsolute upOutputYStep Int Divisor or multiplier for Y coords upRasterBufferInitialized Bool (RO) GS buffer OK upRedTransfer Float[ ] Red transfer upRendering Name Select rendering algorithm upRenderingInitialized Bool (RO) Rendering parameters OK upSelectComponentCommands String[ ] Establish color (output order!) upSetLineFeedCommand String Adjust linefeed (Epson only) upVersion String (RO) Source code version upWeaveFinalPins Int[ ] Number of bottom pins on EOP passes upWeaveFinalScan Int Begin EOP passes (Y-coord) upWeaveFinalXStarts Int[ ] X-pass indices for EOP passes upWeaveFinalYFeeds Int[ ] Y increments for EOP passes upWeaveInitialPins Int[ ] Number of top pins on BOP passes upWeaveInitialScan Int End BOP passes (Y coord) upWeaveInitialXStarts Int[ ] X-pass indices for BOP passes upWeaveInitialYFeeds int[ ] Y increments for BOP passes upWeavePasses Int XPasses × YPasses upWeaveXPasses Int Number of X passes upWeaveXStarts Int[ ] X-pass indices for normal passes upWeaveYFeeds Int[ ] Y increments for normal passes upWeaveYOffset Int Number of blank or incomplete scans at BOP upWeaveYPasses Int Number of X passes upWhiteTransfer Float[ ] White transfer (monochrome devices!) upWriteComponentCommands String[ ] Commands to write each component upWroteData Bool (RO) Something (BeginJob) written to output upXMoveCommand String X positioning command upXStepCommand String Single step to the right upYFlip Bool Flips output along the Y axis upYMoveCommand String Y positioning command upYStepCommand String Single step down upYellowTransfer Float[ ] Yellow transfer
I should mention all of the people who were involved in stcolor's evolution, but I've decided to start from scratch here for uniprint:
Gunther Hess
- John P. Beale
- for testing the stc600 modes
- Bill Davidson
- who triggered some weaving research and tested stc2s_h
- L. Peter Deutsch
- who triggered ease of configuration
- Mark Goldberg
- who prepared the stc500 transfers
- Scott F. Johnston and Scott J. Kramer
- for testing the stc800 modes
- Martin Lottermoser
- for his great commented H-P DeskJet driver
- Helmut Riegler
- for the BJC extension
- Hans-Gerd Straeter
- for some measured transfer curves and more
- Uli Wortmann
- for discussions and his cdj850 driver
- My family
- for tolerating my printer-driver hacking
I wrote this because the documentation was very brief and I really struggled with it for a while, but it is very simple once you understand what is going on.
This only describes how to work out the Y parameters, I haven't looked at the X parameters yet.
You need to know how many nozzles the printer has and the spacing between them. Usually you can find this out from the printer manual, or the printer supplier, but you may have to dissect a couple of printer output files produced with the driver supplied with the printer. There is a utility called escp2ras* that will help with that. Sometimes the term pin is used instead of nozzle but they mean the same thing.
The number of nozzles will be the value assigned to the upOutputPins parameter.
Actually you don't have to print with all the pins available but for the purpose of demonstration I'll assume that we are using them all.
* escp2ras is available from Gunther Hess' page at http://www-md.e-technik.uni-rostock.de/ma/gunther/gs/index.html
The number of passes required is going to depend on the required resolution and the nozzle spacing.
passes = resolution * nozzle spacingThis will be the value assigned to the upWeaveYPasses parameter.
For example if the desired resolution is 360 dpi and the nozzles are spaced at 1/90in then 360 * 1/90 = 4 passes are required. For 720 dpi 8 passes would be required. The printer would, of course, have to be capable of moving the paper in increments of either 360 or 720 dpi too.
You need to work out how much to feed the paper so that when the paper has moved by one head length in however many passes you have then each row space on the paper has been passed over by at least one nozzle. There will be one feed value for each pass and the feed values must comply with the following rules:
sum of feeds = passes * nozzles
feed%passes != 0 (feed is not exactly divisible by passes)
sum of (nozzles - feed) = 0
For example if passes=4 and nozzles=15, then sum of feeds=60. The feed values could be 1,1,1,57 or 15,15,15,15 or 14,15,18,13.
These values will be assigned to the upWeaveYFeeds parameter.
You would need to experiment to see what combination looks best on the printer.
I found it convenient to draw several lines of nozzles and then move them around to see how the different combinations would fill the paper. A computer drawing tool makes this easier than pencil and paper (I used Dia, a GNOME app). The number of nozzles would probably be be a good place to start.
Remember that if the number of passes is more than 1 then the feed increment will be less than the nozzle spacing and passes × feed increment size must equal the physical distance between each nozzle.
These values will be assigned to the upWeaveInitialPins parameter and are the numbers of nozzles to operate in each of the initial passes at the top of a page. The nozzles that the values refer to are the topmost nozzles on the head, nearest the top margin. If the image doesn't start at the top margin then uniprint doesn't use these feeds.
I don't know a mathematical relation for this except that at least one of the values must be the number of nozzles, but I'm sure that there must be one. I used a graphical method, the description that follows refers to the ascii diagram in below.
Draw a line of nozzles for each pass arranged as they would be using the normal Y feed increment determined in step 3. In the diagram below this would be passes 5-8.
Draw a line of nozzles that would print just before the first normal pass. The feed increment for this pass will be close to and most likely 1 or 2 units less than the feed increment of the last normal pass. In the example below this line is pass 4 and the feed increment is 13 whereas the normal feed increment is 15.
Draw each pass before that with a small feed increment so that if all of the nozzles appearing above the first nozzle of the first normal pass operate then all of the spaces will be filled. This feed increment is usually 1 except in cases where some jiggery pokery is going on to make the printer print at an apparent higher resolution than the nozzle diameter.
Now select the nozzles that will operate in each of theses initial passes so that the paper is filled. In each pass the nozzles must be adjacent to each other and at least one of the passes will have all the nozzles operating. I suspect that for each combination of normal Y feed increments there will only be one set of valid beginning of page increments.
-dupWeaveYPasses=4
-dupOutputPins=15
-dupWeaveYFeeds="{15 15 15 15}"
-dupWeaveInitialYFeeds="{1 1 1 13}"
-dupWeaveInitialPins="{ 4 15 11 7}"
x=nozzle operates, o=nozzle not used in this pass
1 2 3 4 5 6 7 8 - pass no
0 x
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
0 x
1 x
2 x
3 x
4 x
5 x
6 o x
7 x
8 x
9 x
0 o x
1 x
2 x
3 x
4 o x
5 x
6 x
7 x
8 o x
9 x
0 x
1 o x
2 o x
3 x
4 x
5 o x
6 o x
7 x
8 x
9 o x
0 o x
1 x
2 x
3 o x
4 o x
5 x
6 o x
7 o x
8 o x
9 x
0 o x
1 o x
2 o x
3 x
4 o x
5 o x
6 o x
7 x
8 o x
9 o x
0 x
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
0 x
1 x
2 x
3 x
4 x
5 x
6
7 x
8 x
9 x
0
1 x
2 x
3 x
4
5 x
6 x
7 x
8
9 x
0 x
1
2
3 x
4 x
5
6
7 x
8 x
9
0
1 x
2 x
3
4
5 x
6
7
8
9 x
0
1
2
3 x
4
5
6
7 x
These parameters would also work:
-dupWeaveYPasses=4
-dupOutputPins=15
-dupWeaveYFeeds="{14 15 18 13}"
-dupWeaveInitialYFeeds="{1 1 1 13}"
-dupWeaveInitialPins="{ 4 11 7 15}"
The Epson Stylus Color 300 uses a different command set to other Epson
Stylus Color printers that use the ESC/P2 language. As far as I can
tell its commands are a subset of ESC/P2. In ESC/P2 the colour to be
printed is selected by a 'set colour' command and then the data sent
is only printed in that colour until the colour is changed with another
'set colour' command. The Stylus Color 300 lacks this functionality.
The data sent to the printer maps directly to the ink nozzles and colour
of an output scan line in the printed output is determined by the position
of the scan line within the data. This means that the driver must know
how the nozzles are arranged and must format the output accordingly.
The extension adds a format that I have called EscNozzleMap and adds some
additional parameters to uniprint.
/EscNozzleMap produces output for the Epson Stylus Color 300
uniprint parameters for the EscNozzleMap formatParameter Type Use upNozzleMapRowsPerPass Int output rows to generate for each pass of the head upNozzleMapPatternRepeat Int no. of rows that correspond to the repeat pattern of the nozzles upNozzleMapRowMask Int[] mask indicating the colour of the nozzles upNozzleMapMaskScanOffset Int[] mask indicating the physical position of the nozzles
A value of 0 means that the nozzle is not used and the row in the output data will be padded with zeros.
colour mask value K 1 C 2 M 3 Y 4 no data 0
| etc ... | |||||||||||||||||||
-dupWeaveYPasses=6
-dupOutputPins=11
-dupWeaveYFeeds="{ 11 11 11 11 11 11 }"
-dupWeaveInitialYFeeds="{ 1 1 1 1 1 7 }"
-dupWeaveInitialPins="{ 2 11 9 7 5 3 }"
-dupNozzleMapRowsPerPass=64
-dupNozzleMapPatternRepeat=6
-dupNozzleMapRowMask="{ 2 4 1 3 0 0 }"
-dupNozzleMapMaskScanOffset="{ 0 1 2 3 0 0 }"
In the example there are 64 rows of data required for each head pass. Each row must be completely filled with data for each pass so if certain nozzles do not print in the pass then the rows for those nozzles will be padded with zeroes.
The row mask translates to "C Y K M 0 0" so in the output data rows 0,7,13,... will contain data for cyan, rows 1,8,14,... will contain data for yellow, etc. Rows 4,10,16,... and 5, 11,15,... will always be padded with zeroes. The upNozzleMapPatternRepeat parameter defines the length of the mask.
The row mask is repeated for each group of upNozzleMapPatternRepeat rows in the output data. In this case there are 64 rows so there will be 10 groups of "C Y K M 0 0" followed by "C Y K M" which is equivalent to 11 output pins.
The upNozzleMaskScanOffset array indicates how the data from the scan buffer is mapped to the output data. The data is presented to the formatter as a buffer of four colour scanlines. The index of the scanline being printed, lets call it y, always corresponds, in this example, to the physical position of the cyan nozzle but since the nozzles are not on the same horizontal line then the other colours for the current pass must come from other scanlines in the scan buffer. The example is { 0 1 2 3 0 0 }, this means that when printing a 4 colour image the magenta data would come from scanline y+3, the black from scanline y+2, etc. It would have been possible in this case to use the array index instead of the upNozzleMaskScanOffset parameter however the parameter is necessary to be able to use the full capability of the printer in black only mode.
-dupMargins="{ 9.0 39.96 9.0 9.0}"
-dupWeaveYPasses=1
-dupOutputPins=31
-dupNozzleMapRowsPerPass=64
-dupNozzleMapPatternRepeat=6
-dupNozzleMapRowMask="{ 0 0 1 0 1 1}"
-dupNozzleMapMaskScanOffset="{ 0 0 0 0 1 2 }"
In this example there is no weaving.
The ESC300 has black nozzles evenly physically arranged as K K K but the data must be sent to the printer as 00K0KK. This is handled by the upNozzleMapRowMask and upNozzleMaskScanOffset arrays. The upNozzleMapRowMask array is { 0 0 1 0 1 1} which translates to { 0 0 K 0 K K } so rows 0, 1 and 3 will always contain zeros and the other rows will contain data.
The upNozzleMaskScanOffset array in this case is { 0 0 0 0 1 2 } so if the data for the 1st nozzle comes from row y in the scan buffer then the data for the 2nd and 3rd nozzles will come from rows y+1 and y+2.
-dupWeaveYPasses=2
-dupOutputPins=31
-dupWeaveYFeeds="{31 31}"
-dupWeaveInitialYFeeds="{1 31}"
-dupWeaveInitialPins="{16 31}"
-dupNozzleMapRowsPerPass=64
-dupNozzleMapPatternRepeat=6
-dupNozzleMapRowMask="{ 0 0 1 0 1 1}"
-dupNozzleMapMaskScanOffset="{ 0 0 0 0 2 4 }"
In this example 2 weave passes are required to achieve the desired resolution.
The upNozzleMaskScanOffset array in this case is { 0 0 0 0 2 4 } because there are two weave passes so if the data for the first nozzle comes from row y in the scan buffer then the data for the 2nd and 3rd nozzles must come from rows y+(1*2) and y+(2*2).
Glenn Ramsey
This section was contributed by Martin Schulte.
With a SPARCprinter you always buy software that enables you to do PostScript printing on it. A page image is composed on the host, which sends a bitmap to the SPARCprinter through a special SBUS video interface. So the need for a Ghostscript interface to the SPARCPrinter seems low, but on the other hand, Sun's software prints some PostScript drawings incorrectly: some pages contain a thin vertical line of rubbish, and on some Mathematica drawings the text at the axes isn't rotated. Ghostscript, however, gives the correct results. Moreover, replacing proprietary software should never be a bad idea.
The problem is that there has yet been no effort to make the SPARCPrinter driver behave like a BSD output filter. I made my tests using the script shown here.
Add sparc.dev to DEVICE_DEVS and compile Ghostscript as described in the documentation on how to build Ghostscript. Afterwards you can use the following script as an example for printing after modifying it with the right pathnames -- including for {GSPATH} the full pathname of the Ghostscript executable:
outcmd1='/vol/local/lib/troff2/psxlate -r'
outcmd2='{GSPATH} -sDEVICE=sparc -sOUTPUTFILE=/dev/lpvi0 -'
if [ $# -eq 0 ]
then
$outcmd1 | $outcmd2
else
cat $* | $outcmd1 | $outcmd2
fi
Since /dev/lpi can be opened only for exclusive use, if another job has it open (engine_ctl_sparc or another Ghostscript are the most likely candidates), Ghostscript stops with "Error: /invalidfileaccess in --.outputpage--"
In case of common printer problems like being out of paper, a warning describing the reason is printed to stdout. The driver tries access again each five seconds. Due to a problem with the device driver (in the kernel) the reason for printer failure isn't always reported correctly to the program. This is the case, for instance, if you open the top cover (error E5 on the printer's display). Look at the display on the printer itself if a "Printer problem with unknown reason" is reported. Fatal errors cause the print job to be terminated.
This section was contributed by Mark Wedel <master@cats.ucsc.edu>.
The Apple Dot Matrix Printer (DMP) was a parallel predecessor to the Imagewriter printer. As far as I know, Imagewriter commands are a superset of the Dot Matrix printer's, so the driver should generate output that can be printed on Imagewriters.
To print images, the driver sets the printer for unidirectional printing and 15 characters per inch (cpi), or 120dpi. It sets the line feed to 1/9 inch. When finished, it sets the printer to bidirectional printing, 1/8-inch line feeds, and 12 cpi. There appears to be no way to reset the printer to initial values.
This code does not set for 8-bit characters (which is required). It also assumes that carriage return-newline is needed, and not just carriage return. These are all switch settings on the DMP, and I have configured them for 8-bit data and carriage return exclusively. Ensure that the Unix printer daemon handles 8-bit (binary) data properly; in my SunOS 4.1.1 printcap file the string "ms=pass8,-opost" works fine for this.
Finally, you can search devdemp.c for "Init" and "Reset" to find the strings that initialize the printer and reset things when finished, and change them to meet your needs.
Copyright © 1996, 2000 Aladdin Enterprises. All rights reserved.
This file is part of AFPL Ghostscript. See the Aladdin Free Public License (the "License") for full details of the terms of using, copying, modifying, and redistributing AFPL Ghostscript.
Ghostscript version 7.03, 20 October 2001