Corel Ventura Fórum

There are many PostScript variant in widespread use: PS, EPS, AI, PDF... What are they good for, what's the difference between them, and, first of all, why are there so many of them?

The role of this individual formats used to be much simpler: PostScript was used as an output file, to drive a printer, to create your end product. Nowadays, even these files can be imported into graphics and layout applications. Let's seem them one by one.

PS

PostScript is a standalone programming language, with its declared aim to describe every detail of a document–letters, pictures, images and other graphic elements–in a precise, device and platform independent way. Device independence means that the same PostScript data will display the same way on any printer or other display device (subject to the technical limits like color depth or resolution of the given device). Platform independence means that PostScript as a language is not linked to any architecture (PC or Mac) or operating system (Windows, Unix, MacOS és hasonlók).

The structure and the features of the language were determined by these goals. Its creators, the programmers at Adobe never really intended it to be a graphic interchange format between various programs. However, it turned out quite early during its usage that–although with some limitations–it can be used for this purpose as well. This lead to the inclusion of Encapsulated PostScript into the definition of the language.

To summarize: .PS is the format of the end product send to a printer or filmsetter, often color separated or imposed according to the needs of the printer.

EPS

EPS is not radically different from PostScript. As it can be expected from a graphical interchange format, EPS can only contain the description of a single change, unlike PostScript itself. Also, some PS operations are not allowed inside an EPS file. Its original aim is to be included directly and completely into another PS file. Consequently, it might not contain commands that would interfere with the encapsulating PS file–these commands include operations that would modify the printer environment like media size or other global settings.

The original way of using EPS files was in sync with this concept: the layout program didn't even try to interpret the contents of the EPS (early Venturas only drew an X into the frame containing an EPS file), but copied it verbatim into the PS stream generated when printing–this is why the format was called encapsulated in the first place.

Later on, EPS was improved to include preview images. These are smaller resolution pictures showing the contents of the EPS file, without giving too much attention to finer details, only to facilitate handling the files in layout programs where they can provide some help during positioning, resizing, cropping. Two of such preview formats became widely used: a TIFF image, converted from vector art to bitmap, plus a WMF metafile, retaining the vector art style (though, this second being a Windows specific format, it only was used on this platform). Interestingly, there was a third format called EPSI, more general than TIFF and better suited to the philosophy of PostScript itself, still it never took off the ground, probably because of its significantly larger size requirements.

During the preparation of the PS output, layout programs discard the preview image and only include the real EPS contents (the EPSI preview might just as well be left in the print stream because it is stored as irrelevant comments in PostScript but as neither a printer nor a filmsetter needs this data, there would be little point in storing it inside the PS output). In the case on non-PostScript output devices, only the preview picture will be displayed (unless the application is equipped with a built-in PostScript interpreter than can understand the contents of the EPS and turn it into non-PostScript printing instructions); the result could be quite poor, although WMF–being a resolution independent vector art format–yields better results. Also note that many bitmap picture editors only import the preview and, maybe without even a warning, will overwrite the real EPS data with this low resolution preview when the file is re-saved. Be careful.

To summarize: .EPS is an interchange format for graphics of any complexity, complete typeset material, ads, etc between graphics or layout applications. Although it can be used to transfer grayscale, RGB or CMYK color pictures, too, the specific image formats like TIFF are much more suitable for this purpose.

DCS

The is no other format with as many misunderstandings and false arguments as DCS (Desktop Color Separation). By a superficial look, it is very easy to mix it with EPS: first, it does have an .EPS extension, second, even if you peek into it, it looks very much like EPS. This is a small wonder as they are indeed related, however, the devil is in the details.

This format was created to help out QuarkXPress that was incapable of performing color separation by itself back then. Instead of teaching it how to do it, its programmers found it easier to delegate the task to image editing applications.

As PostScript (and hence, EPS) can of course store images, you can create an EPS that contains nothing but a single bitmap picture, without any text or other vector art. There are many ways to create such EPS file and they will all yield the same printout in the end, although they might not be the same when compared byte to byte. For all these possible notations, one was selected and named DCS. As a result, DCS became just another picture format, like TIFF or anything similar; applications capable of importing DCS files don't contain a complete, universal PS interpreter that could be used to real any other, non-DCS EPS file. They import it as a picture format, one that happens to describe the image using the machinery of PostScript. In other words, DCS is a special, very limited subset of PS/EPS.

In spite of this, formally, it is an EPS file, thus it can be used tha classic way: encapsulating it inside a larger PostScript file, yet it can deliver no more than a image format. If some vector art material (eg. containing text) is read into a picture editor and saved into a DCS file, any surplus that EPS could originally offer over a simple image format will be lost forever.

As a result of having much more appropriate and efficient formats to handle pictures in, DCS is practically obsolete now (although many, especially those working on Macintoshes still seem to stick to it). There is only one area where DCS still can't be avoided today: pictures containing spot colors, eg. duotone, tritone, quadtone etc. images. Although there are other image formats capable of storing such pictures, none of the widespread layout programs can handled them well in all possible output workflows (both separated and composite PS or PDF). Obviously, this is a shortcoming of these applications, not of the image formats themselves.

To summarize: not counting some special cases, using DCS/EPS to store our usual RGB, CMYK and grayscale images in is a complete waste of storage space, providing no advantages whatsoever. Converting vector art contents to DCS/EPS should be avoided at all costst.

AI

AI is the native format of Adobe Illustrator. It builds upon PostScript but is not exactly PS. To understand this, let's make a small detour.

The PS files generated by DTP applications use the programming facilities of PostScript to a considerable extent. Many operations–like using graphics primitives, printing text fragments, selecting fonts–are repeated time and time again; it pays off to specify all these repeating tasks only once and only refer to them later (in traditional programming, this approach is called subroutines). The programs start creating the PS file by downloading a collected description of possible tasks (called prolog in PostScript parlance) and only refer to them later, sparing a considerable amount of time and storage space.

AI relies on something similar but instead of actually storing the prolog in the file, it simply assumes it. Every version of Illustrator has its own virtual prolog that AI file only use, not specify. EPS, as we saw, is only slightly limited when compared to PS (it can make use most of the PostScript language, with a few exceptions). AI, on the other hand, is very restrictive: it can only use what the virtual prolog allows it to do (and there are no programming operations in this command set).

As a consequence, AI files cannot be simply encapsulated verbatim in PS files; the DTP application has to interpret the contents of the file and has to output the appropriate PS operations, corresponding to the original AI instructions, by itself.

To summarize: unlike all other PostScript formats and derivatives, .AI is not a universal, standardized, platform independent format but the native data format of a single application. As such, it is the least suited to transfer graphics, typeset materials. Although all contemporary DTP applications offer to import AI files, the required conversion can cause problems, especially with fonts and texts. If you need to use AI files, always make sure you test their applicability in your own workflow first.

The EPS and AI files created by Adobe Illustrator underwent some major changes as the application evolved. Up to version 6, AI files were written without the prolog while EPS files did contain it, everything else was the same (with the only additional exception that EPS files, as usual, were allowed to have preview images). That made its EPS file a more universally usable version of the AI. Starting with version 7, both AI and EPS contained the prolog, although only EPS could also have a preview image. From version 9, there were sweeping changes: the AI format, while retaining its name and extension, was completely abandoned and a PDF-based format was introduced instead. The EPS files generated by these contemporary versions contains both the PDF and the traditional EPS data stream.

PDF

Portable Document Format is, from many aspects, the most versatile PS based format, with many advantages and disadvantages. Using this format expects attention and experience from the user, with the additional complication that early mistakes cannot always be rectified at a later stage of the publishing process.

On the surface, PDF doesn't show too much similarity to PS files, still, the links are strong. Although the foundations come from PS, there is something in common with AI, too: PDF can only store graphics elements within strict limits. There is no programming logic any more, just the description of the various texts, fonts, images, vector illustrations one by one, As a bonus, it can store a large variety of additional information like table of contents, indexes or hyperlinks.

The greatest advantages are its high internal compression ratio and the fact that its generation parameters can be selected and adjusted very broadly to suit the requirements of the actual usage. Documents to be downloaded from the Internet benefit from the small size achieved by decreasing the picture quality and using efficient compression: PDF files of comparatively small size can still contain the original layout, typefaces, typography, pictures and graphics–unlike HTML pages where we can have more control over contents than appearance.

Desktop publishing doesn't always enjoy the same benefits. Although the format is gaining ever growing acceptance–for instance, more and more service bureaus expect PDF rather than PS files–, things can go wrong. This isn't necessarily the format's fault, rather those of the users. PDF has so many possibilities that it's quite easy to sett one or of two of those erroneously: font format (whether they are included in the PDF or not, and if so, to what extent), image resolution (original or lower), compression level (lossy but efficient JPG, larger but perfect ZIP/LZW), spot colors, separation, composite colors, CMYK or RGB (plus all their combinations, with the added complexity of color tranformations inside), registration and crop marks (that tend to stay accidentally on material intended for reuse in other publications), media size, bleed.

To summarize: originally, .PDF was just as much an output format as .PS, although to a different crowd: PDF documents were primarily aimed at screen display or printout on office printers. Its versatility allows it to take over the role of .EPS as an interchange format between DTP applications: but only for experienced people, in pre-tested environments. But, of course, anyone can become experienced...

However, giving a PDF to a previously unknown partner can easily lead to catastrophic results. We always have to make sure what they indend to do with it. There are stories about service bureaus where the PDF is read into Acrobat, converted to EPS, the EPS read into Quark and finally sent to the filmsetter. Sheesh... Only give them PDF if you know their RIP will understand it directly and natively. if they mention any kind of extra conversion (even if something as seemingly harmful as printing from Acrobat Reader), stick to good ol' PS. In the era of cheap and ubiquitous CD-R and DVD-R discs, size doesn't really matter and if it does, PS–unlike PDF–can be compressed aggressively.

When, Which, How and Why?

When we try to use one of these formats, most of the time, we have to help our DTP application. These formats are quite similar to each other and it is not always easy for the program to tell them apart. If we specify a single, universal PostScript import filter, the application might be misled. If possible, always select the proper import filter manually. The format can usually be recognized if you look into them with any ASCII file viewer (or Notepad, Wordpad if all else fails, but you shouldn't re-save them from those editors). The first lines, starting with a percent character, usually specify the format and the generating program.

Plain PostScript files–final PS output sent to a file instead of a printer or filmsetter (usual extension .PS or .PRN)–can only be read by a few applications like Corel Draw using its PostScript Interpreted filter.

EPS is the most widespread interchange format, so every DTP application offers to import it. However, there is a significant difference whether it only reads and places the EPS file, shows the preview image, sends the EPS verbatim to the printer when printed or it interprets the full contents, converting it into its own internal format. PostScript Interpreted in Corel DRAW, as its name implies, does the second.

.AI files are to be read using the appropriate import filter. Note that earlier versions of Illustrator were quite liberal in their use of extensions, so we might receive one under the disguise of the other. Always remember to make a test print and, if necessary, rename the file.

.PDF files, especially those with fonts and texts inside, don't always co-operate. If you don't need to modify the contents, placing it into the application (that will simply convert it to EPS and send that to the PS print stream at print time) is usually the safest choice.