Bob Newman explains why the camera JPEG can often be better than the raw file
A question that periodically arises is, what are the advantages of using raw files rather than camera- processed JPEGs? The most frequently offered answer is that using a raw file gives better quality, but this is not necessarily true. The in-camera image processor produces its JPEG output from a raw file, even if that raw file never makes it to the memory card. Furthermore, the camera manufacturer may be expected to know the characteristics of the camera better than a photographer, and so its choice of raw-processing procedure should be expected, all else being equal, to produce better results than the photographer can achieve. Generally, those using the manufacturer’s supplied raw-processing tools find that the output is similar to the camera’s JPEG output, in which case the advantage of going through the extra processing step, and using additional storage space to save raw files, is moot.
So, if not quality, what? The answer lies in information. A raw file embeds all the information about the scene that the camera managed to capture with the chosen settings. The output JPEG file does not. The processor will discard information in several ways. First, the JPEG file is lossy compressed to save storage space. The designers of the JPEG format arrange the compression scheme so that it discards information that is likely not to be visible in the final processed image. This discarded information includes three quarters of the colour information (since the eye cannot perceive as much chrominance detail as it can luminance) and any small tone variations that will be masked by overlaying large ones. Information that is unrelated to the selected output tonal range is also discarded.
The best output devices can render, maybe, a10-stop brightness range, while today’s cameras might capture up to 14 stops, so there is a 4-stop excess of information in the raw file, which is excluded by the raw processor. For the information theory minded, each stop equates to a ‘bit’ of information, so the 4-stop excess produces 4 bits of information per pixel more than is needed. So, assuming the designers of the JPEG processing system have done their work well, the JPEG image should have just enough information in it to produce a excellent rendering of the scene, and no less.
What, then, is the use of keeping that extra information? The answer depends on whether the JPEG end result is the one that you, as the photographer, wanted or not. Since it contains just enough information to produce that particular rendering, post- processing or altering the rendering will result in a loss of visible quality (whether that loss is enough to worry about is another matter). It follows, therefore, that the use of the extra information in the raw file is necessary to achieve some different result. This can be of particular importance where there is a large tonal range in the original scene – some of that will need to be discarded and it is very likely that the photographer can make a better choice than can the automatic function of the in-camera JPEG processor.
Bob Newman is currently Professor of Computer Science at the University of Wolverhampton. He has been working with the design and development of high-technology equipment for 35 years and two of his products have won innovation awards. Bob is also a camera nut and a keen amateur photographer