In the times of film based cameras the sensitivity of the film was characteristic for the film type selected by the photographer. This film type was produced to a specified sensitivity. Film processing had limited influence in varying the sensitivity and also the characteristic curve of the film. To get the best results the images were exposed assuming that the average scene has an overall reflection of 18% of the incident light.

To get the exposure right the cameras were able to measure the reflected light at different positions in the image and calculated the exposure based on the algorithms behind the exposure metering. There were and in many cameras still are modes like sport metering, integral metering, center weighted integral metering, or a matrix mode where the camera analyzed the scene and used a scene specific weighting factor for the different positions. The idea behind the metering modes was to find out which part is the best one to use for determination of the 18% gray assumption and therefore the exposure.

The relatively low slope of a color negative film in combination with the printing / copying process onto photographic paper was able to handle inaccurate exposures or intended exposure variations. Slide film however did not have a copying process and a more or less predefined slope and therefore a predefined contrast range it was able to reproduce. That range was typically limited to a 6 to 8 stop contrast variation. For high dynamic range scenes the photographer had to choose by varying the exposure which part of the scene he wanted to be reproduced and visible in the slide. In these scenes either highlights or shadow detail were clipped.


Auto exposure of modern digital cameras
Nowadays cameras are able to capture a contrast range of more than 10 f-stops. This range decreases by increasing the ISO speed setting that equals an increase of the amplification level of the signal. The amplification also affects the noise on the low side of the signal.
Therefore photographers should use the lowest ISO speed possible without blurring the objects due to object motion or handshake (long exposure times) and without getting into a too small depth of field (wide aperture). Based on scene content – scene analysis required - or scene mode the camera is in, the automated evaluation of the right combination of ISO speed, Exposure time, and aperture may be computed. To measure the illumination level the camera may use the imaging sensor or a separate sensor dedicated to this task.


A potential way how a modern Auto Exposure may work
In addition to that the scene lighting and contrast are important to find the right exposure. The tonal enhancement of the shadows increases the amount of visible noise in the image the optimum exposure is when the high lights are exposed to almost the maximum digital output level and the shadows need only a minimum of correction. In order to do that each scene requires an analysis prior to exposure to analyze the contrast and adjust the exposure so that the high lights reach the maximum output. If the contrast level of the scene exceeds the one the camera is able to capture at the required ISO speed level either the camera itself or the user has to choose if and how much the highlights or the shadows get clipped.
Scene contrast levels that are significantly below the camera dynamic range are exposed to the highlights followed by stretching the tonal range in the image processing in the camera to utilize the given output range.
Of course in low key or high key scenes with extremely small contrast levels the stretching of the tonal range should not be used on the images and especially low key scenes should also not be exposed to the maximum output level.
In some cases three-dimensional information may be utilized to expose to the right level. For example the image is a portrait image in front of a bright window. The person can be identified to be in the foreground by the focus measurement and face detection tells the camera that it is a portrait. Therefore it is most likely that the photographer wants the person to be exposed in a way that it is clearly visible and not underexposed. The highlighted background is unimportant and can be clipped if the scene contrast is too high.


A potential logical step by step procedure to adjust the exposure.

  1. The camera selects open aperture and an exposure time at a given or automatically selected ISO speed level for the sensor used to measure the exposure.
  2. It captures an image and analyzes the scene content (faces and other) aided by focus information. This information may be used to divide the image into sections.
  3. Base on the results of step 2 the histogram (max and min values) of the image or sections of the image get analyzed.
  4. If the highlights do not reach the maximum output level it analyzes the maximum value in the picture and increases exposure so that a second image shows an exposure close to the max. output
  5. If the highlights of the first image are at the max. output and the amount of pixels reaching that value is significant the camera takes a second image at a decreased exposure time and goes back to step 3.
  6. Once the highlights have been adjusted that way the dark parts get analyzed if a significant amount of pixels reaches the minimum output level the exposure is increased until the amount of pixels drops to a small number. This information may be derived from images taken for steps 3 to 5. In this case the contrast level of the scene exceeds that of the camera and a decision for the final exposure needs to be made whether highlights or shadows get clipped based on the scene analysis.
  7. If there are no or not many pixels reaching the lowest output value at the correct exposure for the highlights the exposure index determined from the highlights can be used and a potential tonal stretch is applied to the image processing after capture.
  8. The determined exposure index needs to be converted into an appropriate ISO speed, aperture, and exposure time setting depending on the focal length, object motion and required depth of field.

dw