Automatic exposure fusion

Question

First, I am not an expert in photography, but a programmer. I am currently programming a camera for Exposure fusion, mixing maybe 3 images with different shutter speed.

I am looking for an optimal solution for automating this part. My base idea is that 3 photos should be enough and that one of the photos should use a shutter speed from auto mode. Then based on if the auto value is one the lower or higher end of shutter speeds I am choosing different approaches.

For example, if auto value is on the higher end I choose two lower values on the other two photos. The idea behind this is that I am wasting one of the 3 available photos if I take another photo in the "higher end".

The question is if this is the correct approach and if there is any recommendation based on this?

As an example we could say that we have shutter speeds of 1-10. What algorithm should we use to set shutter speed on the other two photos for each auto value of 1 to 10.

Keep in mind that this will be an automatic calculation and no user will control this.

Answer 1

There is no lower or higher end of shutter speeds in the sense you seem to be using. Ie., there are technical limits for every camera, eg. 1/8000 to 30 secs., but if you reach those, you'll have to adjust another exposure variable anyway as you'll be shooting extremely dark/light subjects.

The "auto" exposure will always be the middle of any exposure bracket you want to make, as the camera determines auto exposure to produce a scene of a middle gray average. To capture the full range of a high dynamic range scene, you need a slower and a faster exposure (if you've determined you want three exposures).

Remember, every doubling or halving of shutter speed gives you 1 stop additional dynamic range, to extend your range by, eg. 4 stops you'll take a picture with 1/4 and one with 4 times the middle exposure time.

Answer 2

You are describing what we call in photo jargon “bracketing”. We choose our best guess at the correct exposure and then make a series of exposures. The idea is, we bracket the spot-on with one higher and one lower as to exposing energy. This can be accomplished by increasing / decreasing the working diameter of the cameras aperture. This can also be accomplished by increasing / decreasing the shutter speed. It is also possible to accomplish a bracket using both shutter speed and aperture.

The first question to be considered is the increment of change. Traditionally the photo industry has uses the magnitude of one “f-stop” as the principle increment. This is a doubling or halving of the exposing energy. Thus 1 f-stop is a 2X incremental change. To accomplish a three exposure series using shutter speed, the factor is 2x the base time – base time – 0.5x the base time. If the 2X increment is too course, you use a 1/2 stop increment which translates to a 150% of base time – base time – 50% of base time. If this increment is too course, than the 1/3 stop increment is next.

If you choose the use to use an aperture adjustment as opposed to shutter speed, we adjust the aperture in terms of an f-stop which is a ratio. This this focal length of the lens divided by the working aperture diameter. This method is used because the ratio is dimensionless. In other words, any lens functioning at a given f-number yields the same image brightness regardless of the dimensions of the lens. So a camera set to f/8 exposes a scene exactly the same as another camera set to f/8, even if the two cameras are different as to size or lens. To make a bracket based on a 2X change via f/number, the factor to use as the multiplier or divisors is the square root of 2 = 1.414. Thus: Base f/number x 0.707 (invers of sq. root 2) – base f/number – base f/number x 1.414).

To adjust the f/numbers in 1/2 stop increments the factor is the 4 root of 2 = 1.189 and its inverse 0.84. To adjust the f/numbers In 1/3 stop increments the factor is the 6 root of 2 = 1.122 and its inverse 0.89.

Answer 3

The purpose you have mentioned in the comment is wahat Canon label HDR backlight control. In the user manual for EOS 700D/Rebel T5i on page 74 there is:

When shooting a scene having both bright and dark areas, use the HDR* backlight mode. When you take one picture in this mode, three continuous shots are taken at different exposures. This results in one image, with wide tonal range, that has minimized the blocked-up shadows caused by backlighting.
* HDR stands for High Dynamic Range.

To be honest I do not know how the camera accomplishes the different exposures but I think it alters apperture or ISO. Altering shutter speed will lead to merging images with different motion blur. Altering appertures leads to different depths of focus and altering ISO leads to different noise levels, though. It, definitely, uses algorithm that tries to minimize all the effects I have mentioned.