Features Profile: Jim DiVitale WPPI Wrap-up Profile: Woody Walters Corel Painter Inside and Out Rf Cookbook Power Selling Profile: Marilyn Sholin Painterly Digitals Profile: Tony Worobiec and Ray Spence Raw Capture: Shooting Digital Originals Columns Insight/ On the Cover Field Test Digital Photography First Exposure Departments Problems & Solutions Calendar Industry News New Products Archives

Rangefinder Magazine
April 2004

Raw Capture: Shooting Digital Originals by Stan Sholik
Choosing the Right File Format: Advantages of Shooting RAW

Photographers working with film understand the value of choosing the right emulsion for the specific requirements of an individual assignment. When they have control over the lighting conditions, know the images will be used for commercial printing, and need to deliver the images quickly with only a quick edit, they know their best option is transparency film. Under unknown or difficult lighting conditions, when the final uses of the images are varied and there is time for film processing, proofing and printing, color negative is a more likely choice.

Photographers shooting digitally are faced with the same shooting situations, but, instead of determining film types, they have to select the right file format for the assignment requirements. For quick shooting, where little needs to be done to the files in post production, and the images have a known use without a lot of potential for reuse, JPEG files are sufficient. They reproduce well enough, and they have the added advantage of being small, therefore taking up less space on media cards than other file formats.

When shooting conditions are outside of the photographer’s control, or the images may have multiple uses, or when the photographer simply wants to have the highest quality image while leaving all options open, then capturing images in RAW format is the best choice.

TIFF is another format that’s an option with some digital cameras. TIFF has been an image file format standard for more than a decade. Until the development of the RAW file formats, TIFF was always the format of choice for serious imaging. But, as will be explained later, it’s no longer the obvious choice it once was.

RAW capture allows you to enhance images as well as correct them without lowering image quality. This is the RAW file as it came out of a FujiFilm Finepix S2 Pro.

Technically, RAW is not an image file format. In fact, it’s not even a single file format. Rather, RAW files are a set of proprietary data files created by higher-end digital cameras, including all professional digital SLRs. RAW files contain the unprocessed image data recorded by the image sensor. Each camera manufacturer creates its own proprietary RAW file format.

Unfortunately, individual proprietary RAW files are incompatible with other manufacturers’ RAW files. Not even the extensions are the same. Some do have the extension “.raw.” Others have different extensions. Nikon’s extension, for example, is “.nef,” while Kodak’s is “.dcr.”

Not only are these proprietary RAW formats unreadable by a competitor’s software, they are often unreadable by earlier versions of the originating manufacturer’s software, even though they have the same file extension. Nikon Capture 3 software, for example, cannot read the NEF files that the Nikon D2H generates.

This image is created from the FujiFilm Finepix S2 RAW camera file by changing the daylight balance to tungsten, applying an exposure correction and changing the tone curve slightly.

Some imaging software from third party sources will read the RAW formats of more than one camera manufacturer. That’s possible in spite of, rather than with the help of, camera manufacturers. In general, digital SLR manufacturers are reluctant to release complete software development kits. That’s why third-party software often lags behind manufacturer software in which specific RAW file formats and updates are supported.

(Editor’s note: An article comparing individual manufacturer’s software with what’s available from third party software developers will appear in a future issue.)

There are numerous advantages of capturing RAW images. One of the greatest is flexibility. Only the actual luminosity data of each sensor element is recorded in the RAW file. Camera settings such as white balance, exposure, tone compensation, saturation, color space and others are not applied to the data, though they are stored with it and applied as defaults after the RAW data is converted.

Not only aren’t camera settings applied to the RAW image, but the data has also not been interpolated. The photodiodes that make up both CCD and CMOS sensors are colorblind, sensing only luminosity. To create color information in most sensor designs, the sensor is covered with a color filter array (CFA). This is a pattern of red, green and blue filters in an arrangement called a Bayer array. Since each photodiode is only recording a single color, the other two colors must be interpolated from neighboring photodiodes covered by the other color filters.

When the image is captured, the analog-to-digital (A/D) converter in the camera converts the analog signal to a 12-bit digital value representing the red, green or blue color value of the picture element (pixel). For RAW images, these values are written to the removable media card, along with the camera settings, for interpolation later on the computer.

	If an image is shot in RAW format, even if the digital settings are set completely wrong as here, it is a simple matter to change the settings in the RAW coversion software to where they should have been set on the camera, then make final minor corrections to the RAW file before saving it as a TIFF or JPEG. In this shot, the Kodak 14n camera settings were “Tungsten” and “Commercial Product Look” with studio strobe lighting.

The CFA interpolation process (also known as demosaicing) takes the sequence of single-channel color pixels and creates a full-color, three-channel RGB image. An image-processing algorithm decides if the “missing” color values are in a smooth area of the image or along an edge, and adaptively determines the best digital code value to use for each missing color value.

The complexity and sophistication of the interpolation algorithm in the software being used to make the conversion is the most important factor in determining the ultimate quality of the final image. In a way, the demosaicing step is analogous to film processing. Third-party RAW conversion software packages offer different options for this “processing.”

Demosaicing provides a full-color, but far from perfect, dark, low-contrast image. The next step involves the application of a “tone curve” (basically a generic camera profile) to the demosaiced image. The tone curve opens up shadow areas and places midtones and highlights in accordance with the dynamic range of the sensor. How well this step is done goes a long way in determining the amount of shadow noise present in the final image and the amount of fine detail that will be present.

Every RAW file converter performs these previous steps automatically before the image is ever seen. Then the software applies the user-selected white balance, saturation, tone compensation, noise removal, color space and sharpening settings, and displays the image on the monitor. To process each of these settings, the software applies separate, sophisticated algorithms. Again, the decisions made by the software designer of each algorithm will have an effect on the quality of the final image.

Here the flexibility of saving images in RAW format is apparent. All of the options selected by the photographer in the camera can be changed in the RAW file conversion software without degrading image quality. If a daylight scene is shot with incandescent color balance by mistake, the photographer can simply select “daylight” balance and the software will go back to the original 12-bit RAW data, tag the file as “daylight balanced” without changing the data, and display the file with the daylight white balance algorithm applied to it.

	This corrected file was made by changing the white balance to “Flash” and the “Look” to “Portrait” in Kodak’s DCS PhotoDesk software. At this point the gray card or other corrections could be used if needed to make final corrections.

Similarly, if the white balance is still not perfect, or if the user wants to adjust the contrast or saturation, the software applications provide tools for this. They will tag the RAW file with the new settings and display the revised image on screen. Only when the RAW file is saved in an industry-standard file format such as JPEG or TIFF are these settings incorporated into the actual 8- or 16-bit file data.

Flexibility is not the only advantage of the RAW file format. For many photographers, image quality is the most important consideration. Converting RAW format images in post-production provides the highest image quality possible from the file.

In the camera, when RAW data is converted to an industry-standard format, the emphasis is on speed, not extracting the last bit of image quality from the data. This is a necessary compromise in order to keep the camera as responsive as possible. The D1X takes less than 0.5 second to write a full-resolution JPEG to a Lexar 40x CompactFlash card.

In comparison, processing a RAW file to the highest-quality JPEG on the computer, while somewhat larger than the camera JPEG, takes just over five seconds on a 3.06GHz Pentium 4. Some of this additional time is devoted to extracting the highest quality image from the data, which is clear when you examine the smoothness of the edge transitions and improved detail in the computer-generated JPEG.

RAW files provide the highest quality possible, and they offer the greatest degree of flexibility, but they do have their drawbacks. There are two main negative factors: storage space and workflow.

RAW files are larger, nearly three times larger, than the highest-quality in-camera JPEGs that can be captured by the D1X. Fewer RAW files fit on the storage card. And, because of their larger size, it takes longer to move the same number of RAW images from the card than it would if they were JPEGs.

But the bigger issue is workflow. The conversion of the RAW data must be done at some point, if not in the camera, then later in the computer. In shooting situations where speed takes precedence over quality, JPEGs have it all over RAW files. And where there is a large quantity of files to process, RAW conversion can be a time-consuming chore, although some of the latest software provides excellent batch-processing options.

As mentioned, TIFF is an option, but no longer the preferred one. In some respects, TIFFs combine the disadvantages of both RAW files and JPEGs and none of the advantages. TIFFs are large, often larger than RAW files. And since they are processed to 8-bit per color files in the camera, suffer from the same inflexibility as in-camera JPEGs. There may be a slight but noticeable improvement in image quality when saving the image as a TIFF rather than a JPEG. Still, that marginal amount of improvement is not enough to recommend shooting TIFFs.

If image quality is the main concern, shooting in RAW mode will guarantee the highest quality image. And it will give the flexibility of having a digital “negative” to work with in post-production, so you can concentrate on the image, not the technical details of digital photography.

Stan Sholik is a contributing writer with NewsWatch Feature Service and a professional photographer handling commercial and advertising assignments in Southern California for 30 years. Amherst Media will publish his latest book, Handbook of Digital SLR Cameras, written with Ron Eggers, in the fall of 2004.