This section provides definitions for all the RAW-based measurements presented within the Image Quality Database: ISO sensitivity (speed), noise (as related to standard deviation and Signal-to-Noise Ratio, dynamic range, and tonal range), color sensitivity, and controlled print comparison. A brief overview of DxO Labs’ test protocols as implemented within our DxO Analyzer image quality evaluation solution precedes the discussion about the importance of measuring in RAW.

ISO Sensitivity

Sensitivity is the ability of a sensor to provide a defined response for a given level of lighting. This information is used by photographers to determine the nominal exposure conditions. If the actual ISO sensitivity of a digital camera’s sensor is lower than the sensitivity set by the user, the image is underexposed; if the sensitivity is greater, the image is overexposed. To be easily understood by photographers, the ISO sensitivity of digital cameras has been defined such that it is similar to the ISO sensitivity of photographic film cameras, thus lower sensitivities require longer exposure for the same luminance to produce the same result. However, just as very sensitive films are known to be very grainy, parallels can be drawn for digital cameras, since high sensitivities are related to high gain and noise amplification.

The myth of high ISO

While it is a common practice for camera vendors to emphasize high ISO settings on their cameras, it must be said that high ISO does not mean good image quality. Any serious photographer knows that the lowest ISO should be used to shoot a scene with a longer exposure time. Only when the conditions do not allow it (as in photojournalism, low light conditions or sport photography) should lower exposure time and high ISO be used (typically to limit motion blur). With film photography, changing the ISO meant having to change the film (which was very annoying). The intrinsic sensitivity of a digital camera is determined by the silicon structure of the sensor itself and cannot be changed, but the ISO of the camera can be artificially increased to arbitrary values by applying a gain to the signal. The price to pay is a proportionate increase of noise and eventually, a decrease of SNR for a given output value (see Explaining Noise Characteristics). The only trick is that the gain is applied before analog/digital conversion so as to avoid quantization effects. So the fact that a camera attains ISO 10000 is not a warranty of image quality; the noise level at this ISO has to be reported as well.

Definition

ISO sensitivity (also known as ISO speed) is a numerical value calculated from the exposure provided at the focal plane of a digital camera to produce specific camera output signal characteristics.

ISO Standard 12232 defines two ways to measure ISO sensitivity. The first relates sensitivity to the exposure necessary to saturate the camera. The second, seldom used, compares the relative exposures to obtain different signal-to-noise ratios. This paper focuses on the more common saturation-based method.

The saturation focal plane exposure is defined as the exposure (illumination multiplied by exposure time in lux.s) necessary to reach sensor saturation. ISO sensitivity is then defined by



When a focal plane exposure measurement is not possible, as for a camera with non-removable optics, it is possible to compute the focal plane exposure as


where
L is the scene luminance (cd/m²),
t is the exposure time (s),
A is the lens aperture (f-Number),
and
,

with T as the transmission factor of the lens,
v as the vignetting factor,
and θ as the angle of the image point from the optical axis.

ISO 12232 considers a transmission factor T = 9/10, an angle θ = 10°, and a vignetting factor v = 98/100, which leads to q = 65/100.

DxOMark measures ISO sensitivity at the image center and thus θ = 0° and v = 1, considering the same transmission factor T = 9/10, which leads to q = 71/100
ISO sensitivity is then defined by

,
with Lsat being the minimum luminance necessary to reach sensor saturation.

As tests show, the ISO settings reported by camera manufacturers can differ significantly from measured ISO in RAW. This difference stems from design choices, in particular the choice to keep some “headroom” to avoid saturation in the higher exposures to make it possible to recover from blown highlights.