The Flatiron Building


ISO and exposure compensation – what’s the relationship?

If I shoot a digital SLR – let’s say a Nikon D200 or a Nikon D40 – at 800 ISO and set +1.0 exposure compensation, is that the same as shooting at 400 ISO?

I wanted a theoretical answer to that question because it seemed to me that increasing the sensitivity of the sensor by increasing the ISO is not the same as setting the sensitivity at a certain ISO and then pumping more light onto the sensor by increasing exposure.

But perhaps that is not so; and I say that because I do not know how the hardware or the software inside cameras work. I only know what works.

First of all, why is exposure compensation necessary?
The answer to that is that the metering system in most SLR cameras (film as well as digital) measures light reflected off the subject, and the light passing through the lens is measured inside the camera.

According to Nikon, their Matrix Metering System can set optimal exposure for virtually every situation.

However, they caution that the internal metering system is designed to provide optimal exposure for a subject with standard reflectivity. The ‘standard’ is an 18% gray, which is halfway between black and white in reflectivity terms.

But when the reflectivity of the subect is very different, the metering system may not be up to dealing with it.

Nikon, and all manufacturers and experts, recommend increasing exposure compensation for scenes that are likely to be underexposed using the metering system alone.

These are scenes where the subject is white or yellow or where sky fills a lot of the scene. Conversely, decreasing exposure compensation is suggested for scenes with a lot of deep green, or for dark subjects, or where a good part of the subject is in shadow.

The bride and groom present a real problem for wedding photographers – she’s wearing a white dress and he is wearing a dark suit.

And the fact that cameras can deal with these shots, shows that film and digital are able to handle the problem with proper technique.

Back to the problem
Well I haven’t got an answer to that theoretical question of the relationship between exposure compensation and ISO in digital cameras, but I have done some tests using the Nikon D40 and the D200. The really interesting thing for me is that the two cameras did not behave the same way.

How I did the test
I did not shoot in Aperture Priority or Exposure Priority or any of the programs where the camera decides the appropriate speed or aperture side of the exposure equation.

Rather, I shot in Manual mode and adjusted exposure using the bar that appears in the LCD. Once the indicator on the bar showed that exposure was correct, I shot the frame.

So, in detail, I shot at 400 ISO, then changed the ISO to 800, adjusted the exposure in the bar to zero and shot again.

Finally, I left the ISO at 800 and changed the exposure compensation from zero to +1.0 but did not adjust the exposure in the bar.

I did this for each camera in turn. The day was sunny and cloudless, so the overall light levels remained more or less the same throughout the test.

Exposure Bar
exposure bar

What I expected is that with +1.0 exposure compensation the shot would be overexposed. After all, it had received twice as much light as the meter suggested it should receive to expose the shot correctly. And so it was with the D40.

I opened the three frames together in Camera Raw 4.2 and looked at the exposure. The shot taken at 800 ISO with +1.0 exposure compensation was lighter than the other two. One of the neat things one can do with Adobe Camera Raw 4.2 is to hold down the ALT button on the keyboard (I use a Mac) and then press the Exposure slider in the program.

The frame blacks out and only shows those highlights (if any) that are blown outside the maximum exposure range.

If the highlights are not blown, the scene will be black. (Of course, a shot could be underexposed, but there is another slider to deal with that.)

With the D40, all the shots had some areas of blown highlights but the shot with +1.0 overexposure showed substantially more blown highlights than the other two shots. I needed to pull exposure back with that shot to get the blown highlights to disappear.

What I actually did was to pull the exposure back precisely one stop, which seemed about right.

With the D200 it was a different story. When I looked at all three shots they all showed very nearly the same areas of blown highlights, which were in the white window frame that occupied one side of the frame. No shot showed a greater area than the other two.

So in terms of comparing one shot to the others, they were all exposed correctly.

I do not understand how the camera could expose correctly vis a vis the other settings when I had put in the exposure compensation.

I shot in Manual mode so there is no possibility that the camera adjusted the exposure, yet the shot with +1.0 exposure compensation looked the same as the others.

I spent some time looking at the histograms for all three shots, and there were only very small differences. It remains a mystery, but what the sets of shots did show can be seen here.

I don’t like to thrust my opinion or interpretation of visual results onto other people. People can judge for themselves. They can ask, is there as much detail? Does one image look degraded compared to the others?, and so on.

But if you ask me, they all look the same. And if there are small differences, you will only see them in 100% crops.

Which means you can use 800ISO without worrying about the image being worse than at 400IS0. You can add exposure compensation and it will not degrade the image.

You might have to modify the exposure in post-processing (easier to do with RAW files than jpegs) but if you have time to look at the histograms in the camera (not so easy with the D40 admittedly as you have to dig into the menus) while you are shooting, you can control that anyway.

And +1.0 is a lot of exposure compensation. I tend to use minus amounts, if anything, and -0.7 is about as much as I normally go, even in the brightest conditions. So here are the shots, and for me the bottom line is – be confident the camera can handle it – and go out and shoot pictures.

D200 panel of three shotsclick on the thumbnail for a much larger view (1000 pixels wide)
D200 three shot panel

D40 panel of three shotsclick on the thumbnail for a much larger view (1000 pixels wide)

D40 three panel

What is the real value of lens sharpness tests?

Some articles and comments I read around the web suggest that sharpness tests and crops of shots designed to show how sharp lenses are, are valuable.

Others suggest they are just pixel-peeping, or that somehow, 100% crops don’t present a valuable picture of what is going on.

I take a different view. The fact that each lens manufacturer publishes mtf charts showing the sharpness capabilities of their lenses shows they at least think there is value in knowing how sharp a lens is.

But there is another reason for testing a lens.

Read enough articles and you will build a picture that tells you that the variation between samples in consumer-grade lenses is greater than with more expensive lenses. No surprise there.

And then there’s the matter of how long that consumer grade lens will stay sharp, assuming it was sharp when new. Parts wear, lenses get knocked against things. What is to say that lens will still be sharp six months after you bought it?

Photodo published their independant test results for years and it was the first port of call for many people looking for a new lens. The Photodo site that is published now is somewhat different than under its old incarnation because it now publishes user reviews. But it still maintains the old database.

There are other sites that rely on user reviews, such as photozone but, no site can answer the question whether a particular sample is a dud, or if it was not a dud when it was bought, have the insides worn away or, what damage was done when it was accidently knocked against that door jamb yesterday?

And when I go out shooting, I like to know that the lens is known to be sharp and that my sample is sharp. And I want to repeat that test at some point so I have something to compare my results with. And I don’t want to make the tests so long and complicated that I forget what I intended to prove in the first place. So if I shoot at f8 and at a particular focal length a lot of the time, a good test shot like that should tell me what I want to know. If there is a problem, I will see it.

How to test
Tripods vary. Some tripods cause more shake than they prevent. Flimsy tripods, long shutter speeds, and jerky shutter releases all add up to blurred photos and results that have nothing to do with the showing the sharpness of the lens.

So I like to use a support that is not going to move. Brick walls are a good bet. Better than tables or other pieces of furniture (though heavy wooden furniture can be good). Put your ear against a wooden table and you will feel the vibration running through it. I can hear the vibration from the whizzing hard drive on this computer on this table now.

After a steady support, the next thing is an aperture that is small enough to ensure that any missed focus or focus errors are not going to show up in the shot. And as I often shoot at f8, that is an aperture that will work for the test and provide usable results.

I took these shots at f8, which on dx sized sensor, gives a reasonable depth of field. The subject to camera distance was approximately two feet (60cm) and the depth of field was therefore about 11 inches (28cm), which is enough to cope with any focus errors.

The next thing is to help the camera to expose correctly, which in this case meant two things. One was to take a shot that was predominantly darker rather than lighter. Too much of a light area in the frame means the camera will underexpose, and then you have to correct for that in processing the image.

The second thing was to cover the viewfinder window with my finger (near but not touching) so that no stray light came in and altered the exposure reading. Normally, the photographer’s eye would cover the viewfinder and prevent this problem, but I had the camera propped on a low wall.

Nikon provide a little plastic cover for the viewfinder, but it is sitting somewhere in my bag and I have never used it.

The Results
So, these are the shots from the Nikon 18-55mm kit lens that came with the D40. The exif data tells me the shot was taken at ISO 200, f8 and 1/4 second.

The crops are 475×475 pixels. I shot the images in RAW and converted them in Camera RAW 4.2 in Photoshop CS3. The sharpening parameters I applied were Amount 69, Radius 1.0 Detail 25 Masking 0


cones crop 1

cones crop 2

cones crop 3

Camera Raw 4.2 became available a few days ago and can be uploaded from the Adobe site, and a good place to start the search to find the exact download page is HERE

Nikon D200 and D40 image processing compared

I used a Nikon 105mm f2 DC lens, which is a very sharp lens, for both shots.

I shot both shots at ISO 200, in aperture priority mode at f4. The D200 exposed the image at one half a second, and the D40 exposed the image at eight tenths of a second.

Before taking these shots I shot a scene out of the window, but decided not to use it because the details were so small that it was difficult to tell what was happening. But what was clear from those shots and from these shown here, is that the D40 shoots warmer.

That is, it reads the color temperature of the subject as being warmer. Therefore the D200 shots tend more towards the blue end of the spectrum, and look bluer, and the D40 shots look more yellow.

This is easily changed in the RAW processor but it worth knowing how the cameras behave ‘out of the box’.

I shot in RAW and converted both shots together in Adobe Camera Raw 4.1.

Once converted to PSDs, I made the crops by using the rectangular marqee tool set to 475×475 pixels, and in each case copied the crop to a new file and then saved using save for web and devices.

The areas covered by the crops are different, one from the other, because the crops subtend a different area of the total frame. This is because the D200 has 10 megapixels, whereas the D40 has 6 megapixels.

D200 crop
D200 crop

D40 crop
D40 crop

I’ve written in other posts, that the 18-55mm kit lens on the D40 seems to produce images that are as sharp at 24mm as the 12-24mm Nikon lens produces on the D200. So now we have the same lens used on both cameras, and the message that shines out to me is:

  • Get out and shoot with the D40 because it is not inferior in image processing to the D200, and
  • Use the 18-55mm kit lens on the D40 because it is sharp.
  • There are other reasons to get the D200 and I will post about these, but for a lightweight camera that is far superior to any compact camera, and which does not cost a fortune, the Nikon D40 is a winner.