Home > Ted’s Commentaries > Raw Demystified

Raw Demystified


If you’ve spent any time reading about digital cameras in magazines or on Web sites, you’re probably seen some mention of “raw files.” It’s easy to get the impression that these mysterious raw files are needlessly difficult and complicated to use, take up too much space on memory cards, and are suitable only for the most meticulous (or neurotic) professional photographers.

That may have been true a decade ago. But since then, newer software (including Adobe’s Lightroom, Bridge, and Elements) and the declining cost of memory cards have made raw files much more practical and easier to use.

I use raw files for all my digital camera images. That approach works well for me, but I certainly would not recommend it for everyone. My intent here is to clear up the mystery of what raw files actually are, and to describe their advantages and disadvantages. I’ll also discuss when you might want to use raw files, even if you’re happy with the regular JPEG files from your camera.

What is a Raw File?

A raw file contains all the data your camera’s sensor recorded while the shutter was open to take a picture. It’s “raw” because the camera copies that sensor data to a file on the memory card, without any of the “cooking” (processing) that transforms the data into an image.

A raw file is actually more than a simple copy of the sensor data. Along with the sensor data, a raw file includes metadata. That’s information about the camera, lens focal length, exposure, white balance, and various other settings. (JPEG files from a camera also contain metadata.) There’s also a small JPEG “preview” version of the image, which is often what you’re actually seeing when you view a raw file on the camera’s screen. Image editing software also uses this embedded JPEG image for thumbnails and previews. The camera also applies lossless compression to make the file smaller. (“Lossless” means the original data is exactly restored when the file is uncompressed.)

The important thing to remember is that a raw file is not an image file. Rather, it contains minimally-processed data from the camera sensor. You can use the appropriate software— a raw converter— on your computer to turn that data into an image, which you can save in TIFF, JPEG, or whatever image file format you prefer.

To fully understand what this means, and how a raw file is different from a JPEG file, you need a basic idea of how a digital camera sensor works.

Making Sense of Sensors

A digital camera sensor is a silicon chip that contains millions of microscopic photo sites arranged in a grid. Each photo site measures the amount of light falling on that tiny section of the sensor when the shutter is open. When the shutter closes, the computer in the camera queries each photo site and records its measurement as a number. Sensors in cameras that can create raw files use 12 or 14 bits to record those measurements of light intensity. That means each photo site can distinguish and record 4,096 or 16,384 possible tones.

If we were content to take pictures in black and white, that would be the end of the story. Each photo site would directly correspond to a pixel in the image. Taking color pictures is much more complicated. Remember, a photo site is a tiny light meter that “sees” only in black and white. The usual way to make a sensor that can take color pictures is to put red, green, or blue filters over the photo sites in a specific pattern.

Schematic illustration of a Bayer arrayThe Bayer array (a section of which is illustrated here) is the most common arrangement of color filters, used in nearly all camera sensors. It’s named for Bryce Bayer, the Kodak scientist who invented it in 1976. Creating a full-color image from this mosaic of filtered photo sites involves some sophisticated processing. Adjacent photo sites “see” different colors, and there are two green filters for every red or blue one. (Dr. Bayer modeled this pattern after the retina of the human eye, which is most sensitive to green light.) To produce a full-color, full-resolution image from the data in this array— a process called demosaicing— you need to calculate the color of each pixel in the final image from the measurements of least three different photo sites. You also need to fill in the gaps between photo sites that record different colors.

This is where the difference between raw and JPEG becomes apparent:

Making JPEG

To create a JPEG file, the camera first applies the white balance, contrast, sharpening, saturation, “picture style,” and whatever other parameters you’ve selected (or their default settings) to its demosaicing process. Then it reduces the 12 or 14 bits each photo site recorded to 8 bits per color for each pixel, since that’s all the JPEG format allows. That means it reduces the 4,096 or 16,384 tones a sensor can distinguish for each color to 256.

It next applies “lossy” compression. JPEG’s lossy compression greatly reduces the file size by taking advantage of the limitations of human vision. It removes parts of the image your eyes and brain won’t miss. Finally, the camera writes the compressed file to the memory card, and erases the original sensor data to make way for the next picture.

A high-quality JPEG file can be as little as ten percent of the size of the image in an uncompressed format, such as TIFF. But to get that much smaller file size, the process of tone reduction and lossy compression permanently discards most of the data the sensor captured. Still, if you got the exposure, contrast, and white balance right when you took the picture, the image will display and print very nicely. When I’ve experimented with selecting my camera’s option to produce both raw and high-quality JPEG versions of each picture, I’ve found it all but impossible to tell the difference between the JPEG version and the raw file converted to TIFF (an uncompressed format) using the camera’s settings.

But what if you didn’t get the exposure, contrast, or white balance right when you took the picture? You normally won’t miss the color tones and other information that JPEG processing removed from the image. And you can probably make minor adjustments in an image editor without problems. But large changes to the brightness, contrast, or color balance can make the absence of what JPEG discarded very visible. You could end up with cartoonish color (“posterization”), or solid bands in blue sky or other areas that normally have subtle gradations of color. Those adjustments could also exaggerate fuzzy halos around the edges of objects that are often present in JPEG images, but would not otherwise be noticeable.

Raw Advantages

Unlike a “lossy” 8-bit-per-color JPEG file created in the camera, a raw file retains all the data the sensor captured. A raw converter on your computer can translate the 12-bit or 14-bit measurements from the photo sites into pixels with 12 or 14 bits per color, and save the resulting image in a 16-bit-per-color TIFF file. That gives you much more data and many more color tones, letting you alter the exposure and color balance without producing posterization, banding, or visible artifacts. The raw converter can also extract detail from bright and dark parts of the image that might have been permanently “clipped” to solid white or solid black in a JPEG file from the camera.

(When you’re done with the 16-bit image you created from a raw file, and made all the necessary adjustments in your image editor, you can always save the finished image as a JPEG file. The “lossy” compression won’t be a problem, as long as you’re sure you won’t be making any more adjustments other than cropping or resizing.)

With raw files, you need not be concerned with setting the camera’s white balance. The camera setting doesn’t matter, since what you choose in the raw converter is what determines the color balance. You can set the white balance to “auto,” and never touch it again. That’s one less thing to bother with while you’re taking pictures.

The camera will record the white balance it chooses in the raw file’s metadata; and the raw converter will use that setting (along with contrast, vividness, sharpness, and other parameters) to display an initial version of the converted image. But unlike JPEG files, where those camera settings are “baked in” to the image, you can change any or all of those settings to make the picture look the way you want.

If you learn about color management and calibrate your monitor, you can also use a “wide-gamut” color space like Adobe RGB or ProPhoto. Then your pictures can include certain vivid colors that can’t be recorded in the standard sRGB color space used for a camera’s JPEG files. Don’t worry if that doesn’t mean anything to you. Color management and color spaces are far more complex topics than raw files. The point is that raw files give you the possibility of a greater range of vivid colors, if and when you’re ready to explore it. 2007 version of Aliakmon Abstract 2012 version of Aliakmon Abstract

Raw files also let you take advantage of improvements in raw conversion technology. New software might have a better demosaicing algorithm that extracts more detail or reproduces color more accurately. It might provide better noise reduction, or improved ability to adjust parts of the image. You can “re-develop” your old raw files to take advantage of the improved processing.

I took this picture of an oil tanker’s colorful hull in 2007. For the first version (left) I used the raw converter included in Photoshop CS2, which tended to turn intense red tones orange. Adobe has since improved the Photoshop raw converter’s ability to render vivid red accurately. For the second version (right) I used Photoshop CS5 to “re-develop” the raw file in 2012. (I also decided that cropping would improve the composition.) You can see a larger version of the updated picture here.

You might think of a raw file as something like an undeveloped film negative that the raw converter “develops” to create an image. (Adobe’s Lightroom uses this metaphor.) You can control this “development” process— and do it as many different ways as you want— to make a final “negative” that corrects any flaws, looks the way you want it, and is ready for “printing” in an image editor. By that analogy, a JPEG file is a color slide or a Polaroid® print that you pretty much have to accept just the way it pops out of the camera.

Raw Disadvantages

Raw conversion adds an extra step to your workflow— the sequence of steps you take to get pictures ready for whatever you want to do with them. That’s probably not a problem for photographers who carefully select images for adjustment in Photoshop or a similar program. And current versions of Adobe’s Lightroom, Bridge, and Elements make working with collections of raw files nearly as easy and convenient as JPEG. But that extra step might be too much trouble if you just want “good enough” pictures straight from the camera for printing, or for sharing downsized versions on Instagram or Facebook.

Raw files are also significantly larger than JPEG files. A raw file from a 12-megapixel camera averages around 14 megabytes, but can reach as much as 17 megabytes if you’re using a high ISO setting (a higher ISO image has more noise, which makes lossless compression less efficient). The corresponding JPEG file would be 4 to 6 megabytes.

The increasing capacity and declining cost of memory cards have reduced the relevance of the old arguments about the “inefficiency” of raw files. But the large files can still cause problems when you’re shooting fast action in continuous bursts. Larger files take longer to write. If your camera doesn’t disable continuous shooting in raw mode, it becomes significantly slower. Using memory cards that can accept data as fast as your camera can write it can help. But even a photographer who otherwise shoots only raw will switch to JPEG when his daughter is about to score a soccer goal.

Another disadvantage isn’t often mentioned in “raw vs. JPEG” discussions. “Raw” isn’t a single standard like JPEG. Each camera manufacturer has its own proprietary format for raw files, which may even be specific to a camera model. The business model for digital camera manufacturers relies on rapid obsolescence, with new models replacing old ones every year (or less). When a camera becomes obsolete, its raw files become obsolete as well.

It’s quite possible that a manufacturer will stop supporting an obsolete camera in its raw converter, and also that other developers of raw conversion software will stop supporting it. You can keep the software that came with the camera, but there’s no way to know if it will run on the computer or operating system you’ll have in ten or twenty years.

TIFF and JPEG files are probably “archival” for the indefinite future. They’re in standard formats, and billions of those files already exist. It is therefore sensible to convert the raw files of images you want to keep, and then save them in those formats. Conversely, the formats of raw files are proprietary, undocumented, and tied to the cameras that create them. You can count on them becoming rapidly obsolete, which means the continued ability to read them is unknown. (Adobe’s DNG is an attempt to address this problem with a standard for raw files, along with free software to convert proprietary raw files to the DNG format. But its future is no more certain than the camera-specific formats it purports to replace.)

Raw in a JPEG World

What you should take away from all this technical discussion is that raw files are most useful when you intend to adjust or manipulate your digital photos in Photoshop or a similar program. The extra information in raw files allows a lot more latitude for adjustment of exposure and color balance, and may give you more color and detail in the picture.

That said, if you’re like most digital camera users you probably don’t want to spend much time in front of a computer monitor tweaking your pictures. And you don’t need the ultimate image quality for large prints. JPEG files straight for the camera are usually fine for the prints in your scrapbook, or for e-mailing pictures of the kids to Grandma and sharing selfies on Facebook.

But there still may be times to consider switching your camera to “raw.” Those situations include:

Can You Even Use Raw Files?

Now that you know what raw files are and when you might want to use them, one important question remains: Can your camera even create them? Most cameras can’t, but check the manual for yours to be sure.

If you’re looking for a new camera and want the option of raw files, a good rule of thumb is that nearly all cameras with interchangeable lenses support raw files. This category includes DSLRs (a “digital single lens reflex”) and the newer “mirrorless” cameras. Conversely, very few cameras with lenses you can’t remove can write raw files. But there are exceptions.

Generally, the more expensive a camera is, the more likely it is to create raw files. A very few high-end point-and-shoot cameras (such as Canon’s PowerShot G series) have that capability. They’re marketed to “enthusiasts” who are most likely to want raw files. Some high-end cellphones can also produce raw files. There are also apps that can add raw capability to some other phones that don’t have it built in.

About Raw Converters

Any camera capable of producing raw files will almost always include the manufacturer’s raw converter in the box. But you’ll probably want one or more “third-party” raw converters, developed by someone other than the camera manufacturer. Camera manufacturers are usually quite good at designing cameras, scanners, and other hardware. But their software to support that hardware is too often lacking, as it’s not their “core competency.” Other software developers can often do better.

Canon’s raw converter is Digital Photo Professional (DPP). It works well enough, but many Canon camera users (including me) intensely dislike its user interface and workflow. It’s included on the CD that comes with raw-capable Canon cameras. You can download an update to the latest version free of charge from Canon’s Web site.

Carson mansion with ACR 4.4.4

Adobe Camera Raw 4.4.1 produced noticeable red and green stripe artifacts when rendering this raw file. This cropped section of the image is 75% of full size. At normal magnification, the stripes give edges an unnatural jagged appearance. I couldn’t see the stripes when I later tried converting the raw file with Photoshop CS5’s ACR 6.7.

Carson Mansion with DPP 3.4

Canon Digital Photo Professional 3.4 rendered the raw file with very faint striping, visible only at high magnification.

Maui sunset with DPP 3.4

Adobe Camera Raw 3 produced ugly artifacts in the clipped highlights around the clouds at the top of this picture of a Maui sunset. Canon’s DPP rendered it without the artifacts.

Canon, Olympus, Samsung, and Sony have always provided full-featured raw converters with their cameras. For a long time, Nikon’s main raw converter was an extra-cost accessory (and a profit center). Their raw-capable cameras came with Nikon View NX, crippled entry-level software that provided very limited control of its processing. If you wanted official Nikon software to make full use of raw files, you’d need to spend $180 for Capture NX 2. But Nikon now provides the free (but less powerful) Capture NX-D that replaced Capture NX 2 to anyone who ones one of their cameras.

Rather than developing their own raw converter, Panasonic provides a customized version of Silkypix, a product of Ichikawa Soft Laboratory. I haven’t tried Silkypix, but nearly all the reviews I’ve seen criticize it as needlessly difficult and unintuitive to use.

Adobe Camera Raw (ACR) may be the most frequently-used raw converter. That’s because it’s included (and highly integrated) with Adobe’s Photoshop, Lightroom, and Bridge. It supports over 300 different camera models, and Adobe is frequently adding new ones. Photoshop Elements also uses ACR, but with a simplified interface that omits certain advanced features. If you use any of those Adobe products, you’ll probably find that ACR fits more conveniently into your workflow than the camera manufacturer’s raw converter. Opening a raw file in Bridge or the built-in file organizer automatically starts ACR.

Corel’s Paint Shop Pro also has a built-in raw converter, but it’s one of that program’s weak links. The way it’s integrated into the image editor doesn’t allow full control over the processing of raw files. Corel also seems incapable of providing timely updates for new cameras between their major releases.

Some other third-party raw converters are Corel’s AfterShot Pro (formerly Bibble); DxO PhotoLab; and free open-source Raw Therapee. Free open-source dcraw is meant for Linux command-line wizards, but is available for Windows and various other platforms; it’s also the basis of numerous other raw converters.

Whichever raw converter you prefer, I recommend collecting as many raw converters as you have disk space (and money) for. That’s because developing the necessary algorithms for demosaicing and processing raw files is as much an art as a science. That art and science has matured to the point where you should get very similar output no matter which software you use. But because each raw converter uses different algorithms, you will inevitably find that some images will yield better results with a raw converter other than your preferred one.

I use Adobe Camera Raw for nearly all my digital camera images. But my software collection also includes Canon’s DPP, Raw Therapee, and the discontinued DxO Optics Pro. Before I switched to Windows 7, I also used the discontinued Pixmantec RawShooter Essentials, my preferred converter before I got Photoshop. VueScan, which I use for film scanning, incorporates dcraw code, so it can also process camera raw files.

The top two pictures at left illustrate how a collection of raw converters can be very useful. They’re a cropped section of a picture of the Carson Mansion, a famous Victorian house in Eureka, California. This image has some unusual configuration of lines that fall slightly askew on the sensor mosaic, and possibly some fine details at the limit of the camera sensor’s resolution. Adobe Camera Raw 4.4.1 (with Photoshop CS3) rendered it with noticeable red and green striped artifacts on the horizontal carved moldings (top left). Although the colored stripes themselves are visible only at pixel-peeping magnification (these sections are 75% of full size), they give edges a strange jagged appearance in normal-sized views or prints. I haven’t noticed this problem on any other raw files, including other pictures of the Carson Mansion.

I used Canon’s DPP 3.4 for the final version of this picture (middle left). Although DPP’s rendering also shows some very faint striping, it’s noticeable only on very close inspection at full magnification. Without the red and green artifacts, it isn’t a problem. But DPP lacks ACR’s adjustments for highlights, shadows, noise reduction, chromatic aberration, and de-fringing; so I had to spend more time in Photoshop adjusting the final image.

I didn’t see the stripes when I tried the file in RawShooter Essentials and Raw Therapee as a test after I finished working on the picture; so I have to assume that the artifacts are specific to ACR’s demosaicing algorithm. I later tried this image in ACR 6.7, the final update of ACR 6 for Photoshop CS5, and DxO Optics Pro 8.5, which DxO gave away as a free promotion at the end of 2014 (the current version was 9 at the time). Both converted the image without visible stripes. The ACR version looks slightly sharper than DPP’s, although I disabled sharpening in both raw converters and used the same setting in Focus Magic for “capture sharpening.”

A picture of a sunset at Kuau, Maui (bottom left) is another case where DPP came to the rescue. The Hawaiian sunset scene exceeded the dynamic range of the camera sensor. The clipped highlights at the top of the picture produced ugly artifacts in ACR version 3 no matter how I adjusted it. (ACR 6.7 produced the same results when I tried it.) RawShooter Essentials wasn’t much better. But Canon’s DPP rendered the clipped highlights without the artifacts, rescuing this rather nice picture from the recycle bin.

I wouldn’t use DPP regularly. But I’m glad to have it around for those rare occasions when ACR doesn’t give me what I want. Whatever failings camera manufacturers’ software might have, the proprietary knowledge available to their developers may help it handle images with difficult “boundary conditions” better than third-party software, whose developers rely on painstakingly reverse-engineering a limited number of raw files.

Commentary Contents

tedsimages.com Virtual Light Table Home Page