December 12, 2003

Using a Linux desktop for graphics, media production

Author: Chris Gulker

During Gulf War I, I was sitting on the picture desk of a daily newspaper, framegrabbing video, prepping wire photos, and making up street editions with a $50,000 Mac and $100,000 worth of Sony video gear. During Gulf War II I was again framegrabbing for publication, but this time with an $800 Linux box and a $39 video card.
Indeed, Linus Torvalds was still finishing Linux in 1991, even as U.S. troops were marching into Kuwait. Linux has matured tremendously since then, to the point that my 2003 Red Hat 9/Ximian XD2 workstation is some two orders of magnitude cheaper and at least that much faster and producing higher-quality output into the bargain.

Linux desktops have clearly matured and become useful for daily media production, including graphics-intensive media. Tons of Web pages are authored and updated from Linux typically with desktop tools such as Bluefish and GIMP, as well as Perl and PHP-based server-side content management systems.

Linux tools work well in heterogeous systems

Built on open standards, Linux Web tools also tend to work well in heterogenous environments, where they can help glue diverse systems such as Windows- or Mac-based content management to Apache-based Web servers. The blog at gulker.com works in just this way using (formerly) xawtv (now tvtime), GIMP, and Mozilla to place framegrabs, graphics, and text.

I think it's likely that Linux desktops daily produce as many Web pages as, say, the top 100 daily newspapers, and Linux-oriented sites alone -- Slashdot, NewsForge, Linux Journal et al. -- have tens of millions of unique visitors monthly, putting them in a circulation class with major national print publications. Linux may have a small number of desktop users, but those users are highly productive. All the Linux pubs put together likely have less staff than any single paper on the top 100 list.

While Linux and the Web grew up together, the Linux desktop has also become useful for print media. TeX, the typesetting computer language originally developed on a DEC minicomputer in 1980, is alive and well on Linux. In March, India's largest phone company used TeX and a PostgreSQL database to turn out 400,000 copies of the two-volume telephone directory for Thiruvananthapuram, the capital of Kerala province. The project, which had required 50 people and six months' time using conventional print production technology, was accomplished in four months by a smaller team using Linux.

While TeX and the LaTeX app continue to be widely used for developing technical, scientific, and mathematical documentation, Linux has also made gains in areas requiring more user-friendly WYSIWIG applications and print-focused features and utilities.

Color management always a key issue

One area where other platforms, notably the Mac, have an advantage is mature applications for handling CMYK (cyan, magenta, yellow, and black) color, which is the color model used in printing. CMYK image files -- referred to as "color separations" in the trade -- are surprisingly complicated and difficult to master. The GIMP, at least prior to the 2.0 release, only works in RGB color, the model used by the human eye, computer monitors, and Web graphics. While there is at least one GIMP plug-in, Separate, to convert RGB to CMYK, and some workarounds using GIMP's channels feature, nothing on Linux matches Adobe Photoshop and other proprietary apps for CMYK capabilities.

However, a new open source page layout application called Scribus helps solve that problem by taking advantage of the PDF file format. PDF, particularly a variant called PDF/X, has emerged as the standard format for print production. Virtually every magazine, newspaper, and color ad is delivered to its printer as a high-resolution PDF file that contains a mix of vector and raster data.

Scribus gives users a Quark Xpress- or Adobe InDesign-like pasteboard interface for making up publications such as magazines and newsletters. In the process, Scribus can embed RGB bitmapped graphics files in the documents (using an XML-based native format) before exporting them to PDF format. Scribus also takes advantage of the open source littlecms libraries to provide color management -- called CMS in the trade -- a crucial step toward guaranteeing color fidelity throughout the scanning, layout, and output process. CMS systems use open color standards to match the color space of, say, a scanner or digital camera to a monitor or printing press via various color-matching algorithms and lookup tables.

PDF files that contain RGB image files can be converted to CMYK by a RIP -- Raster Image Processor -- a computer that is used to drive the laser plate maker that makes the CMYK plates that will be mounted on the printing press. The most modern digital presses dispense with the platemaker and write directly to printing drums using technologies not unlike those used by office laser printers. Ghostscript, it should be noted, is popular with RIP makers because of its low cost and large feature set.

One advantage of this approach is that the RGB color gamut is, in practice, much larger than the CMYK gamut that is left post-conversion. The same PDF/RGB file may be able to be RIPped for a wide variety of output devices (four-color offset press, digital press, high-end ink jet), whereas each of those output media would require a separate CMYK file for optimum quality. This is particularly important for applications such as advertising, where the same ad layout might be placed in a newspaper or magazine, or on a bus card, billboard, or other large space. The PDF/RGB file can, in theory, fill all three bills with proper attention to issues such as image resolution and color management.

More on page 2 ...

Good vector-based tools available

Scribus also includes vector drawing tools, and allows features such as setting type on a curve. Other open source vector drawing apps include Sketch, Karbon14, part of KOffice, Sodipodi, and OpenOffice.org's Draw, which allow users to create illustrations in vector format. Vector formats use resolution-independent mathematical descriptions of curves, lines gradients, et al., and allow a drawing to be output as a native vector file (like SVG) or a bitmap rasterized for a target resolution like 72 dpi for the Web and 300 or 1200 dpi for print documents. Adobe Illustrator and Macromedia Freehand are two well-known proprietary vector drawing apps.

Down the publishing scale a notch, a huge number of corporate and other publications are created in Microsoft Word -- everything from newsletters to training and HR manuals. StarOffice 7 is not only compatible with MS file formats, but it has built-in PDF and HTML creation, reducing the number of steps required to create a print and screen version of a document that can easily be distributed to both print and online readers.

Improvements in the Linux kernel have also made life easier for those who input images from film and print scanners and digital cameras, thanks to better support for technologies such as USB and FireWire. Open source applications that include SANE (and the SANE plug-in for GIMP) and the commercial VueScan make it easy to capture images from just about any hard-copy medium into a Linux machine. VueScan is a welcome addition because it is a workhorse app that is also widely used on both Mac and Windows. VueScan scans made on Linux, in my experience, reliably match scans made on other OSes with similar equipment, and VueScan supports an almost unbelievable array of SCSI, USB, and FireWire scanners.

Full workstation for under $1,000?

Indeed, it is a marvel that publishers, a notoriously thrifty set who may comprise the only industry that has cut its way to greatness, have not rushed to embrace Linux. You can assemble a very powerful Linux graphics workstation for well under $1,000 (possibly under $500), which would seem very attractive versus an $1,800 Power Mac that also requires some $2,000 worth of proprietary software to function.

When I pinged colleagues for examples of publications using open source apps, all replied that they, too, were amazed that open source wasn't being used more widely on print publication desktops (although Linux is being widely used as a server OS in publishing). The culprit is almost certainly the status quo -- and experience and skill sets -- that are tied to other platforms, primarily the Mac for high-end print graphics work.

The ongoing cost of payroll for skilled print production workers makes the cost of a computer seem small by comparison. However, publishers do regard productivity as key to profitability, and the Mac has long withstood the advance of Windows on the grounds that Mac workers are more productive in the well-developed, easy-to-use Mac environment. Mac OS X uses PDF as its native imaging format, for example.

But can publishers resist open source for long? One daily newspaper publisher rigged a Linotype machine -- a mechanical typesetter that first appeared in 1886 -- with a QWERTY-style keyboard so he wouldn't have to pay both reporters and typesetters. In a business where cheap is king, publishers who are willing to try will likely find that the Linux desktop has matured into an inexpensive and productive tool both for Web, and, increasingly, print publishing.

Chris Gulker, a Silicon Valley-based freelance technology writer, has authored more than 130 articles and columns since 1998. He shares an office with 7 computers that mostly work, an Australian Shepherd, and a small gray cat with an attitude.

Click Here!