Graphic File Types Overview

Graphic Image File format. Uses a CLUT (color lookup table) to define the colors as though they were individual color chips, and only supports up to 256 colors per image. Although it can simulate continuous-tone colors by dithering, that’s generally best left to the JPEG or PNG formats. GIF87 was the original Web graphic file format, way back in 1993. The current version, GIF89a, supports 1-bit (jagged-edge) transparency, comments, and simple animation.

GIF is your best Web option for images with flat, solid colors — the sort of images you normally create in vector drawing programs such as Illustrator. Of course we’d love to put our Illustrator images on the Web in their original vector form, but for the most part that’s not possible (but see SVG lower down in this table). So we export our vectors to raster formats like GIF or PNG for the Web.

GIF is rarely a good choice for non-Web use.

Technically GIF and its LZW compression algorithm are “lossless,” but since it supports indexed color only (8-bit or less), you often have to permanently throw away image data prior to (or in the process of) exporting your master file as a GIF. Don’t be confused by Photoshop’s Save for Web options for GIF, where you’ll see a “lossy” checkbox: that just rearranges the pixel patterns slightly, prior to exporting the GIF, to enhance the compression. GIF being a “lossless” format means that, unlike with JPEGs (a lossy format), you could possibly open a GIF repeatedly, edit it, and re-save it back out again without necessarily degrading it. That’s not ideal, unless you really know what you’re doing. It’s generally best to go back to the master file — often a PSD — for editing, and then re-export the GIF.

Joint Photographic Experts Group. Actually a family of file formats; usually refers to JFIF JPEG.

For Web, digital camera storage, and stock photo dissemination. (By the way, this latter use is counter-intuitive, since it’s a lossy format and stock photos are often intended for high-end prepress work; but in this kind of use the files are so huge, and the compression applied so minimal, that there’s no significant data loss.) But for everyday use — which generally means in Web development — the lossiness is a major issue. The tradeoff is between file size and quantization artifacts.

Typically JPEGs are 24-bit RGB files. There are many variants and choices to be made in exporting JPEGs. Many people save JPEGs from Photoshop’s built-in “Save for Web” dialog, but there are lots of other programs — including Web-specific graphics applications like Fireworks — that can save JPEGs. Each has its own vagaries. Be aware that the JPEG “quality” scales (0–10, 0–100, or whatever) in these programs have no universal meaning and are not standard across programs; they’re only meaningful for comparisons within the same program.

Also be careful not to edit JPEG files if at all possible; go back to the original master file (PSD, TIFF, or whatever) for editing and then re-export a new JPEG with your changes. JPEG compression always involves data loss and degradation; editing JPEGs is just corrupting the corruption — it’s degrading!

Portable Network Graphics. Relatively recent substitute for GIFs (and some JPEGs) online. Many technical advantages, such as…

• Lossless compression (which means you could use it as an editable format, although you probably shouldn’t in most cases).
• Multi-bit transparency map (alpha channel), even for photo-type images.
• Metadata for color management (gamma and ICC color profile), although this is something of a tease since most browsers don’t support those things.
• Can hold either RGB data (like a JPEG) or indexed-color data (like a GIF) — but not CMYK, since that’s designed for the Web, not for print.

Unfortunately, there are still a fair number of older browsers in use that don’t support PNG’s alpha-channel transparency; and most browsers don’t support color management at all, or have it turned off by default. This makes the use of these features by Web creators problematical.

By the way, the Web-graphics creation program Fireworks uses a kind of PNG as a native file format; but it may contain vector and animation data, which browsers can’t read from a PNG. Web creators should be careful to distinguish between that master-file format and the program’s Web-export “flat PNG” format.

PNG’s use is growing slowly over time, especially as newer browsers come into play. It’s already a reasonable replacement for many GIFs (but not for animated GIFs); but for photo-type images, JPEG will usually be more efficient.

Scalable Vector Graphics. Wouldn't you love to put flat-color or other simple graphics (like logos and diagrams) online in compact, scalable vector form? The W3C has approved this Adobe-sponsored XML derivative. Not really usable yet in most real-world projects, because browser support is still spotty — but improving. The only Web vector format widely used at present is to embed images in Flash, but that’s not really a substitute for SVG (you wouldn’t create a Flash file just to hold a non-animated scalable logo, for example).

PostScript is a page-description language that some programs can generate and some printers (the expensive kind) can print from. A .ps is a simple text file that results when you tell a program to send its PostScript instructions to a file on your hard drive instead of to a printer; it’s therefore called a “print to disk file.” (It’s also sometimes called a “pure PostScript file” or a “PostScript dump.”)

There are basically three things you can do with a .ps file: send it to a printer (which should then print the original page, not the PostScript text); convert it to PDF via Acrobat Distiller; or — if you’re a PostScript programmer — edit it directly in a text editor. (I’m not a PostScript programmer but, being rather geeky for a creative type, I’ve been able to make some simple, useful changes within .ps files on occasion. As with any programming code, you have to be careful — one tiny mistake and the whole thing may not work.)

Encapsulated PostScript. EPS is essentially a PostScript file in an “envelope.” It usually — but not always — includes a rasterized preview in TIFF or PICT, plus some metadata. EPS was originally the native format of Illustrator, back in the primordial days of PostScript.

Like virtually all vector file formats, EPS can hold raster data too; and it’s the standard way to import photos with clipping paths (which is to say, raster data surrounded by a vector) into QuarkXPress and some other programs. EPS files are mostly exported from a “creation” program like Illustrator or Photoshop, and then placed in a layout program like Quark or InDesign... that is to say, EPS files usually aren’t edited directly. But if you need to, modern apps can sometimes open them... provided (1) the EPS isn’t too old (the format has evolved) and (2) you’re feeling lucky. Sometimes the data is editable; sometimes, even if you can crack open the file, it turns out to be in a form where editing is impractical. Moral: always save the master files used to create your export files!

Pros: can contain clipping path, true font data, various kinds of metadata; widely accepted.

Cons: previews optional & nonstandard; sometimes contains insufficient data but you wouldn’t know because the preview can be misleading; file format has evolved so is nonstandard; may only print properly to PostScript printers; can be flaky at times.

The Adobe Illustrator program’s proprietary format originally was EPS. Illustrator’s current .ai format is essentially an extension of that original EPS format, broadened to accomodate Illustrator’s newer capabilities. A few other programs may support .ai files, but it’s a moving target as Illustrator evolves. Most other programs stick to the original, more generic form (EPS).

Portable Document Format, also known as “Adobe Acrobat format.” Not really a “graphic file format,” since it’s designed to contain entire pages including graphics, type, vector shapes, and overall layout; but I include it here because it can, in fact, be used purely as a graphic file format (to contain one or more images). For example, at one time the Macintosh used this format to store screenshot images.

The PDF format attempts to capture or “freeze” the appearance that a document will have when printed to a PostScript printer, and make the document scalable (capable of being enlarged cleanly), even without the components of the document (images, fonts, etc.) being available as separate files. (Traditionally, page layout programs require that the components other than text be available and linked into the main file.) PDF files are often slightly editable with Acrobat Pro in case you don’t have access to the original master file and/or authoring application.

PDF tends to be more stable and universal than other forms of PostScript, but be careful: just because a PDF looks good (and is cleanly zoomable) on your computer screen doesn’t necessarily mean it can print out cleanly at high resolution, especially on a printing press. It depends on how the file was created.

Adobe has been slowly moving towards PDF as a universal file format (especially within its own product line), but it’s not clear whether it will ever replace most proprietary and generic graphic file formats.

You might like to visit Adobe’s PDF Tech Center online.
Still to come: discussion of prepress variants of PDF with more specific specs:
PDF/X-1a
PDF/X-3

PSD stands for “Photoshop document.” It’s an application-specific proprietary format, but because of Photoshop’s dominant position in the pixel-editing world, PSD has become something of a quasi-standard. A number of other programs, even some that don’t come from Adobe, support PSD as an additional file format — but usually as read-only or for import/export purposes, not as their true native file format.

In addition to pixel data, the PSD format can contain a broad array of other data concerning such things as layers, color modes, color-management profiles, and even real type and other kinds of vector data. However, not all programs (other than Photoshop itself) support all these data types within the PSD format; they may either ignore such data or (in some cases) actually throw it away. So you have to be careful if using PSD as an inter-program transfer format, which is not really its intended purpose.

The current versions of the two major print-oriented layout programs, QuarkXPress and Adobe InDesign, can both import PSD files natively. In some cases you could now actually go to press without exporting your master PSDs to TIFF or other more traditional prepress formats. Web browsers cannot read PSDs, however.

Tagged Image File Format. Raster only. Most widely used format for photos in prepress world (although EPS is also used for this purpose). Cross-platform.

Pros: most reliable, widespread format for raster data. Can hold almost any resolution, color scheme, etc. I like this format because it’s uncomplicated and (unlike EPS) rarely causes trouble.

Cons: can’t hold vector data; large, and sometimes slow to print; Mac and PC have slightly different flavors (although most Mac apps support both); many nonstandard “improved” versions floating around.

There are many variations of TIFF — in fact, it’s really a whole family of file formats — and you can’t assume that a program that “supports TIFF” will understand all of them. Variations include compressed (lossless LZW, or other methods) vs. uncompressed; RGB vs. CMYK; 24 bit vs. 48 bit; and a variety of TIFF-based alternative file formats, such as TIFF IT (widely used by ad agencies). TIFF is really designed for use by professional graphic designers. Although the format itself is relatively trouble-free, the numerous variations are likely to confuse non-professionals trying to match individual TIFFs to the programs and processes that support them.

It’s hard to talk about “Raw” because it’s not really the name of a file format. Originally the term was used within Photoshop for doing a desperation import of a mystery file that contains no metadata whatsoever, not even the most basic facts such as file format, size, color mode, etc. You give Photoshop your guesses about these things and then it tries to open the file (and generally fails).

A newer (and basically unrelated) meaning of “raw” — more formally called Camera Raw — has become very common with the rise of digital photography. Some midrange, and all high-end, digital cameras have the ability to save images with no lossy compression (as opposed to the JPEGs that are more common as a digital camera format). Essentially, they’re saving the full information that their sensors capture. Camera Raw isn’t really a file format in the conventional sense because each brand of camera structures the data differently, and expects you to read the data (or transfer it to your computer), and perhaps edit it as well, using their own software. I suspect the lock-in is intentional, and of course it’s bad for consumers.

Adobe has been trying to overcome this lock-in problem in two ways: via plug-ins for Photoshop that can import some of these files directly; and by promoting a new file format called DNG (Digital Negative) as a universal lossless format for digital cameras. Only some of the newer digital-camera manufacturers have supported DNG; so far, the traditional camera makers have stuck with their proprietary formats.

BMP is very common in the Windows world, but is not cross-platform. It’s not supported by professional prepress processes or by Web browsers, and so is only appropriate for internal use within the local Windows environment.