Adobe has convinced computer users that PDFs are universally portable. However, what you see in a PDF is how it will look when printed. This is especially true with respect to printing on high-speed commercial printers. Viewing a PDF is not the same as viewing a printed proof of your book.

PDF stands for "portable document format." It was Adobe's idea for using their postscript printer language to make document files that would be independent of computer, operating system and also independent of the application that created the document in the first place. The idea is that Word users can share PDF files with WordPerfect users. It is also that a PDF created on a Mac will look the same if was viewed on a PC.

PDF is a way of packaging postscript to make it more easily viewable. In recent years, the PDF specification has been expanded to include all kinds of visual and multimedia elements. In addition, the complexity of graphical elements allowed within the PDF specification has grown.

Some document creation programs, particularly sophisticated image editing programs like Photoshop and publishing programs like InDesign, are capable of making highly complex PDF files that have multiple overlapping image and font elements. These elements can have varying amounts of "transparency," allowing partial viewing through them to reveal overlapped elements below. Eventually all of the elements must be flattened into a bit map image that is printed. The bit map is made up of a two dimensional array of pixels, one per spot. Each pixel has a color that depends on the order and transparencies of the overlapping elements at that location in the original image.

For complex PDFs, the elements in different layers of the image are "put down" on the virtual page in sequence. In the PDF viewer (like Adobe's Reader) the image is assembled by going back over the page, again and again, to add layer upon layer to the image. After doing that assembly, the resulting image that displays on your screen is a simple bit map. Again, each pixel of that assembled  image has one set of color codes. At this point the viewed image can be printed by reading back all those pixels and printing them pixel by pixel. This is a so-called "rasterized image."

When you print a PDF from a desktop computer the image is usually assembled in the viewer and then printed as a bitmap. When a PDF is printed in a commercial setting there is no viewer creating the bitmap. Instead there is "RIP" software.

High-speed printers do all of the creation of a bit-mapped print image in software before sending the data to the printer. The software that does this is called a "RIP." RIP stands for "raster image processing." For a high speed printer, The RIP software cannot afford the time to do complex repainting of a virtual image until a page is created and it can move on to the next page. Commercial RIP software assumes that the PDF is not complex. The latest bells and whistles of PDF specification are not included in part because it is expensive to keep revising and in part because it would slow down the processing.

This is a well-know problem with PDFs

This problem is well-known, if not exactly well-publicized. The commercial print industry has be dealing with PDF incompatibilities for years. They finally decided that it was necessary to create a subset of the PDF specification that would have compatibility with commercial RIP software. This grew into several sets of specifications and updated specifications. You can learn about these by googling the Internet for terms like PDF/X, PDF/X-1, PDF/X-3 and so on. These specifications are limited subsets of the full PDF specification. Applications that can create PDFs in these subset specifications have a better chance at easily making documents that print well on commercial printers.

Font embedding

How does all this relate to fonts and font embedding? Adobe's PDF specification allows a large set of "built-in" fonts to be used without having the fonts embedded into the PDF. The commercial RIP software has no such set of built-in fonts. All fonts must be embedded in your PDF for it to print properly.

Unfortunately, the process of font embedding is also somewhat complicated. There are different ways to embed fonts and not all of these ways are equally supported by commercial RIP software. In addition, some elements in the font embedding process can be non-standard and still allow the PDF to view reasonably but not to print properly. There are probably many ways this happens. The fundamental reality is that some versions of the widely used "ghostscript" program produce PDFs that are poorly supported by commercial RIP software.