What "Exposure Compensation" does

What does "Exposure Compensation" do?

If your camera-produced JPEG are coming out darker or lighter than you like, then you can use "Exposure Compensation" to make the JPEGs lighter or darker.

If you select one stop of positive exposure compensation, the resulting JPEGs should be one stop lighter. Similarly, a stop of negative exposure compensation should make the resulting JPEG one stop darker.

Image lightness is a result of the relationship between exposure (light reaching the sensor) and the ISO setting. The camera's metering system strives to find an exposure that will give a reasonable result in the context of the selected ISO. Depending on the selected mode, the camera may try to alter aperture, shutter speed, or ISO to get a good match between the exposure and ISO.

Conceptually, Exposure Compensation applies a bias to the camera's metering system.

When you select a positive exposure compensation, you bias the meter to think there is a little less light. This will cause the camera to try for a higher exposure and/or a higher ISO.

The name "Exposure Compensation" is a holdover from the days of film. With film, the camera could not adjust the film speed, and therefore the only option available to the camera was to try to raise the exposure.

With digital cameras, it really should be called something like "Lightness Compensation," as it may alter ISO instead of altering exposure.

There is some shorthand in that paragraph. By a JPEG that is one stop lighter, you mean a JPEG that is as light as one that received one stop more exposure, assuming no auto ISO. The pixels in that JPEG file may or may not be one stop brighter.

When you are shooting raw, lightness is not a property of the pixels in the file the camera produces.

i just picked up an olympus trip 35 film camera made between 1967 and 1984 they sold 10 million of them the largest selling camera in the world and it has only 2 shutter speeds 1/40 and 1/200 sec :-) when you adjust exposure comp you adjust the ASA :-)which would in turn adjust the aperture i figure :-)

isnt the raw file is produced after iso/analog gain value is applied from the pixel voltage.

Some of the time, yes. There are also changes in conversion gain, which is different from analog gain, and digital gain. And some cameras at some ISO settings write metadata for the raw converter instead of applying gain to the raw values.

But that's beside my point. The data in raw files is not lightness. The raw converter produces images that have that property, and you can't determine the lightness of a pixel by looking at the raw value associated with that pixel.

Correct. Neither lightness nor ISO is defined for raw files.

"ISO" provides a context for interpreting the raw data. When you alter the raw processing parameters to make the image lighter or darker, you are effectively changing the ISO.

However, if you don't deviate from the processing parameters specified in the raw file, then positive exposure compensation should result in a lighter JPEG. This will typically be the result of an increased exposure, and/or an increased ISO value stored in the raw file.

You are correct in that if you deviate from the processing parameters stored in the raw file, all bets are off. For instance with the ISO set to 100, you could take two shots, one with a normally metered exposure, and one a stop higher. If you manually process the raw files, you can arrange it so the JPEG produced from the file with the higher exposure is actually darker (even though the camera was set to the same ISO for both of them).

Correct. If the camera is set to a fixed ISO, one stop of exposure compensation should result in a one stop change in exposure. That will change the lightness of the camera produced JPEG.

That's an implementation detail that can vary with different models of camera.

With some cameras, the ISO setting is only a field in the raw file. The data from the sensor is the same for ISO 100 or ISO 800. The only difference is the context for interpreting that data.

Some cameras do scale the data (with analog and/or digital circuits), however this sort of implementation detail is something a beginner photographer (and may professional photographers) don't need to worry about.

Right, but how the camera does that can drastically affect headroom. Compare the GFX 50S to the GFX 100S, for example.

I rather liked Maarten Heilbron's explanation earlier this month - www.youtube.com/watch?v=Jr6E63LL2kw

looked like an april fools day utube to me :-) lost me when you can lock iso and still set exposure comp in other modes. ;-)

That is being very pedantic, Jim!
If I look at two pixels in the raw file (and consider just one colour channel), suppose one has the value 0.5 (on the scale 0 to1) and the other has the value 0.25, then the first pixel recorded about twice as much light as the second. If that is not lightness, then what would you call it?

  I prefer to use "brightness" for the subject (or scene) and "lightness" for the image (or photograph).

Don

Of course. However, the two pixels will have a lightness ratio of 2:1 in the "raw" image. It is perfectly reasonable to imagine the raw data as representing an image. We don't have display devices that are capable of displaying raw data directly, but one can still imagine such a device.

When referring to a given Exposure, what is captured by a pixel in the raw data is a representation of a certain amount of energy, usually referred to as 'intensity' in imaging (see e.g. Cornsweet). Intensity is linear with Exposure all else being equal: double the Exposure, the intensity (in DNs) doubles in the raw data. This is true of radiometric as well as photometric Exposure. It is independent of non-linear effects introduced by the imaging pipeline or the Human Visual System.

However if Exposure is doubled, humans do not perceive a doubling of the light stimulus because of the non linear response of the HVS. For this other non-linear quantity we normally talk about perceived 'brightness'.

The imaging pipeline that produces a rendered image from raw data intensity is often non-linear, therefore 'intensity' no longer strictly applies to jpeg data. However, in image processing the value of a pixel, whether raw or processed, is still often referred to as 'intensity' (E.g. Gonzalez and Woods).

Jack

I disagree with this statement.

Correct. Neither lightness nor ISO is defined for raw files.

the iso setting or lets call it a simple potentiometer/amplifier is directly after the photo diode. the raw file is a stored digital file, and only the expanded iso settings are applied to the raw image which in-turn produces a less clean file. you can correct me if im wrong but i dont think im wrong.