Camera performance is now measured as much in terms of Dynamic Range as Resolution. Phil Rhodes explains this essential aspect of camera technology
For more than a decade, dynamic range been one of the most important, even controversial, measures of camera performance. Principally because of unflattering comparisons with film, video camera manufacturers have worked hard to match the results of traditional techniques. Those difficulties are now on the edge of being overcome, with fourteen-stop electronic cameras such as the FS700 now available even to relatively modest productions, although there's still a lot to talk about with regard to both the absolute performance of modern cameras and the still-inevitable comparisons with a century of photochemical filmmaking.
Dynamic range was not routinely discussed over lunch until people started speculating about electronic acquisition replacing film, and this speculation began long, long before electronic acquisition was anywhere near what we'd now call ready to do so. Even then, the principal stumbling block was seen as resolution, not dynamic range, although early tube cameras did have fairly poor range compared to modern semiconductor sensors. When, in the late 80s or early 90s, high definition cameras began to appear which could compete with the resolution of film, it was still obvious that something was missing.
In its strictest sense, dynamic range is the ratio between two quantities: the smallest quantity which a system is capable of detecting, and the largest. Lots of things have a dynamic range, and the term is frequently used to refer to the range of loudness detectable by audio devices as well as the range of brightness detectable by video cameras. In terms of an electronic camera, the largest detectable quantity (or white point) is easy to understand, because the point at which each photodetector on the sensor reaches full capacity is well-defined and consistent. The smallest detectable quantity (or black point), however, is somewhat subject to interpretation, because electronic imaging sensors never output exactly zero-value pixels. Noise, from thermal and other sources, makes it inevitable that sensor output will never quite reach true black, and deciding on the notional black point is to some extent subjective. For this reason, the absolute dynamic range of any electronic camera is, to within perhaps one or one and a half stops, controlled by how much noise one is willing to tolerate, and therefore a matter of opinion.
The actual numbers involved in these calculations can be large. Because each stop represents a doubling of the amount of light striking the sensor, a 14-stop sensor must be able to handle two to the power fourteen or 16,384 times more light reflecting from a completely white subject than a completely black one. This implies very good electronics, capable of handling desperately small and fragile signals without introducing 1/16384th of the maximum signal level as noise. Improvements in the electronics used to offload photodetector values from the sensor itself have made for improvements in both sensitivity and dynamic range by making the shadow detail quiet enough to be usable as picture information.
Early photographic materials actually had fairly poor dynamic range – the first multiple-exposure HDR techniques, as we discussed once before on this site, were developed to mitigate this limitation. Eventually, the relationship between grain size and sensitivity was established, and it became possible to manufacture film stocks with both higher sensitivity and greater dynamic range, with various grain characteristics. These techniques have been applied to sensors, too; some have used a combination of large and small photosites to create two simultaneous exposures on a single device, although this approach has not found widespread favour, possibly because a single large photosite is still better.
Back in the early days, though, few people were thinking about dynamic range in terms of the stop counts we use today. The response of film to light is highly nonlinear in any case, since the density of the negative does not precisely double with double the exposure, but the operative point here is that this is an artistic decision as much as it is a technical one. Various film stocks produce results that various people like under various conditions. Until December last year, Kodak made a colour reversal motion picture film, Ektachrome 100D 5285, which probably didn't have the dynamic range of the best modern video cameras – although it did have the nice highlight handling of everyone's favourite analogue photochemical acquisition format. The contrasty, saturated look of this reversal stock meant that it was sufficiently popular to be spared, at least initially, from the cull of reversal materials that took place when Kodak stopped manufacturing complex E6-process slide film.
And that's the thing. While it's good to understand the theory, photography is not about the numbers. Video cameras are sometimes criticised for lacking dynamic range, and that's true to an extent; the problem, though, is how they run out of that range, and the manner in which the highlights clip or the shadows go noisy, and that's an almost entirely subjective thing. This might all be based on our learned responses to the way movies are culturally expected to look, but that's life. We're shooting pictures to please an audience, and regardless of the equipment we use or the capabilities it has, it's up to us to make it look right regardless.