Pixiq

Simon Stafford goes behind the scenes of Nikon’s proprietary 3D Color Matrix metering II to help you make the most of this sophisticated TTL metering system.

Nikon’s Matrix metering takes its name from the way the system divides the frame area into segments (the “matrix”); it compares the light detected by each segment and assesses the difference between them to establish a pattern (map) of brightness across the scene. The brightness pattern is then compared against a proprietary database created by Nikon of image brightness patterns derived from actual photographs of a huge range of different subjects, scenes, and lighting conditions; in current and recent Nikon D-SLR camera models the database comprises over 30,000 brightness patterns. This comparison process helps the system to fine-tune its assessment of the nature of the subject / scene that is being photographed and set exposure accordingly. 

However, this represents only one of a number of other processes that take place when Matrix metering sets an exposure. From the introduction of D-type Nikkor lenses back during the early 1990’s and more recently with G-type Nikkors, the camera receives information from the lens concerning the focus distance (this is not a precise measurement but an approximation based on a series of zoned focus distances) and the focal length. The camera uses this information to determine how far away the subject is from the camera and what sort of picture is being taken. For example, if you use a telephoto lens focused on a subject close to the camera, the brightness pattern recognition process will discount significantly brighter areas in the upper half of the frame area, based on the assumption that it is probably a sky and therefore less important compared with the subject at the reported focus distance; however, shoot the same scene under the same light with a wide-angle lens focused at a far distance, and the camera will assess that you are probably taking a landscape picture and the brightness in the upper part of the frame is a sky area, so it will place more emphasis on setting exposure for this region. 

The introduction of the D3 and D300 during late 2007 brought about a significant modification to the Matrix metering system, as used in previous Nikon SLR cameras, by the addition of a diffraction grating, known as the Diffractive Optical Element (DOE), between the viewfinder prism and the lens that directs light onto the RGB Matrix metering sensor. 

Prior to the introduction of the DOE the light passing out of the camera lens is reflected by the main reflex mirror into the glass pentaprism located in the viewfinder head of the camera, from there it moves on to the objective lens immediately in front of the viewfinder eyepiece. A small prism positioned between the pentaprism and viewfinder objective lens refracts a little light up to a small lens in the roof of the prism head, which focuses the subject onto the surface of the RGB Matrix metering sensor. The RGB sensor has an array of alternating columns; each column comprises red, green, or blue sensitive pixels. Consequently the red, green, and blue images of the subject detected by the sensor are not in precise registration with each other, which limits its ability to resolve the subject accurately.

The DOE is a special filter that has a textured structure on its surface measured in nanometers (1 nanometer = 1 millionth of a millimetre), which diffracts (bends) light by a proportional amount according to its wavelength (colour). The amount of diffraction is controlled to an extremely fine degree, so that red, green and blue wavelengths are offset by a distance equal to the spacing of the columns of pixels on the RGB sensor. As a result the efficiency of the pixels is increased improving the signal from the sensor, and the resolution of the subject is much improved, enabling the system to detect the brightness pattern and distribution of colors within the frame area with far greater precision. 

The net result is a significant improvement in the accuracy of the automated control of exposure and white balance setting, plus enhancement of the autofocus system, as the Matrix metering sensor is able to track the subject’s position within the frame area by mapping its brightness and color patterns, a function that is at the core of the Dynamic-area AF 3D-tracking feature of Nikon D-SLR cameras, such as the D3-series, D700, and D300-series, while the autofocus system reciprocates by passing information to the Matrix metering system. Nikon refer to this close integration of the Matrix metering, autofocus, and automatic white balance, as the Scene Recognition System; it is a feature of all Nikon D-SLR cameras from the D3 and D300 onwards, including the recently announced D3100 and D7000.

To summarize the functioning of Nikon’s 3D Color Matrix metering II system in these camera models, as discussed so far, it draws on five key sets of data:

• The overall brightness of the scene
• The difference in brightness between each individual segment of the RGB metering sensor
• The distribution of colour(s) in the frame area 
• Focus distance and focal length information from D, or G-type Nikkor lenses
• Information from the AF system as to which AF points are active and how they are interacting 

It may come as a surprise but in the integrated world of Nikon’s Scene Recognition System the camera’s AF-area mode influences how Matrix metering determines exposure settings.

• Single-point AF: Matrix metering places great significance on what is under the selected AF point in terms of both brightness and colour, biasing exposure accordingly. If whatever is under the AF point is significantly lighter than a mid tone (18% grey) the exposure level is decreased, while if it is significantly darker than a mid tone the exposure is increased, which can lead to the image being under-, or overexposed respectively. This aspect of metering and AF interaction is probably the single most common reason behind inconsistent exposure levels when shooting the same general scene in constant illumination but with different parts of the scene at the point of focus.
• Dynamic-area AF: Matrix metering adopts a more generalized approach by looking at what is under the pattern of sensors (for example on the D3-series, D700, and D300-series models that is either 9, 21, or 51 AF points), again to determine if the exposure needs to be biased for this area of the frame
• Auto-area AF: Matrix metering assesses the entire frame area and if it detects a significant level of brightness under any AF point, the system will bias exposure, reducing it in an effort to prevent overexposure of highlight regions.

By way of an illustration consider what happens when the camera is pointed at a subject positioned against a bright blue sky and relatively darker tones beneath it. Using single-point AF-area mode the camera determines where in the frame the subject is positioned and approximately how far it is from the camera by knowing which AF point is reporting focus and the information it receives from the lens about the set focus distance. Matrix metering looks at the brightness and colour patterns and will decide that the large area of bright blue is probably a sky in the background, so it places less importance on this compared with the subject that is in focus. If Matrix metering detects a flesh tone under the AF point it will assume the subject is a person and fine-tune exposure settings accordingly. The key point with Matrix metering is the way it reacts to the difference in brightness between regions within the frame area, rather than the actual brightness value it measures for each region.  

In a situation where the difference in brightness across the frame area is small, in other words the scene contrast is low, Matrix metering is very reliable and accurate. However, if the difference in brightness across the scene is large, as it will be in many typical daylight scenes lit by bright sunlight, it is likely to be beyond the measurable brightness range of Matrix metering and the ability of the camera’s sensor to record detail in the brightest regions (highlights) and darkest regions (shadows) in a single exposure. In this situation Matrix metering will generally try to hold highlight detail by reducing the exposure level at the expense of the shadows, which will be underexposed, because once a highlight is overexposed on a digital camera there is nothing that can be done subsequently to recover it.

So far I have dealt with four of the five key data sets used by Matrix metering, the fifth is overall scene brightness. In a situation where the incident light illuminates the scene at “normal” levels (i.e. it is neither too bright, nor too dim) Matrix metering functions predictably, as described above; however, in low light, or very bright conditions some subtle changes in the way Matrix metering operates can and often do occur.

In low light (dark) scenes Matrix metering will frequently take into account the central area of the frame, as defined by the central 12mm (FX-format) and 8mm (DX-format) circle marked on the camera’s focusing screen, placing greater emphasis on this region, so it is important to keep the AF point(s) over the part of the scene from which you want the exposure level to be set, especially if the subject is off centre (i.e. outside of the central circle).

In very bright (light) scenes with low contrast Matrix metering will generally bias the exposure toward the mid and lower tones on the assumption that the dynamic range of the camera will enable it to retain highlight as well, while in high contrast situations it will work in the opposite way biasing exposure to retain highlight detail at the potential expense of underexposing the middle and shadow tones.

Finally, there is one other camera function that can affect our perception of exposure accuracy. I often hear complaints about how, in the opinion of the user, Nikon D-SLR cameras from the D3 and D300 onward produce consistently “overexposed” pictures; however, a little investigation usually reveals that the exposure is in fact “correct”, since the highlights show plenty of texture and tone, with the histogram indicating that a full range of tones has been recorded. 

The root of the issue sits in the Nikon Picture Control system, in particular the ‘Brightness’ (gamma) setting. Its default value for each of the applicable Nikon Picture Controls is zero, although the value of a zero setting is not the same in each Picture Control. The ‘Brightness’ control affects rendition, principally, of mid-tone values. At default settings this generally places them at a higher tonal value, irrespective of the overall exposure level, making them appear visually brighter. Therefore, it is important that exposure is considered separately from the tone rendition applied by a Picture Control, but to understand that the two are linked.