I've seen a lot of threads here and in other forums where someone will complain that their camera or their flash isn't working right because their pictures are always underexposing or overexposing. A lot of people have even resigned themselves to permanently setting FEC to a specific amount of f-stops, just so they can get a correctly exposed image. After reading so many such threads I sometimes can't help but wonder if ETTL or ETTL-II really is that unrealiable or is it more a case of "user error".
I came up with a test, described below, that some of you might find useful to see if ETTL on your camera is spot on or off by a certain amount. I was really pleased with the result that I got with my 1DmkII. ETTL-II was spot on (to within a tenth of a stop).
Attached are two images from my test. The first one was taken with the flash in Manual mode and the second one is a composite of two test shots in ETTL-II mode.
For those interested in performing a similar test of their own, here's how I did it. Note that this test will only work with flashes that can be set to manual mode like the 550EX and the 580EX. Also, this test should be done in a room where the ambient lighting is low enough so that the flash is the primary source of lighting for the test shots.
1. You will need a background that is as close to neutral gray as you can get. The farther the backgroud deviates from neutral gray, the more the discrepancy will be between the manual flash exposure and the ETTL flash exposure.
2. You will also need a target subject that is neutral in tone. I used a Black/Gray/White card since the average of the three colors should be neutral in tone.
3. Set up the background and the target subject exactly 10 feet away from the flash. Use a tape measure to accurately measure the distance. Position the target so that the glare from the flash is not reflected off the surface of the subject back to the camera. (NOTE: Instead of 10 feet, you may end up using a different distance, as determined by Step 6 below).
4. Set the camera and the mounted flash on a tripod. Remember to keep the flash exactly 10 feet away from the subject.
5. Turn on the camera and set it to manual mode, shutter at 1/250 or whatever your highest sync speed is.
6. Turn on the flash and set it, also, to manual mode. Note the distance scale indicator on the flash and adjust the camera aperture so that the distance marker is on the 10 feet position. If you flash does not have a 10 ft. marker or is in meters, then select a distance marker that is close to 10 ft. You will need to reposition the target so that the distance between it and the flash matches the distance marker you have selected on the flash.
7. Once you've determined the aperture appropriate for the flash-to-subject distance, take a shot of the target. The shot will be close to being perfectly exposed if you did a good job of accurately measuring off the distance to the subject. This shot will be your control image.
8. Set the flash to ETTL mode and take another shot. Make sure that FEC is 0 on the camera and on the flash. If ETTL on you camera was calibarted correctly at the factory, the exposure of this image should closely match your control image.
9. Use EVU (or any other program that shows the histogram on a segmented graph) to compare the two images.
How'd your camera do? 
.
please log in to view hosted photos in full size.
please log in to view hosted photos in full size.
I conducted my own test to see if ETTL-II will really perform as advertized when there is a very bright specular light source in the frame. The specular light source, in my test, was caused by a reflection of the flash off of a mirror. I've attached two images below of my results.
