After several discussions on Nikonians Forums concerning how TTL and TTL-BL work with respect to off-center subjects and distance, I thought I'd do a study to really find out for sure. Until now, all my comments on this topic have been based on what I've seen in my wedding photography. Now, I can say for sure exactly what is taking place.
This is more of a practical study than a scientific one. I have mostly been looking for effects, and not setting up control shots for precise measurements. I'm interested in how to use this knowledge to take better pictures, not for a review of the system for a specification.
The first thing I wanted to experiment with was the issue of how TTL and TTL-BL differ when shooting a light subject on a dark background at the same distance.
All the images in this study were shot with the following:
D200 Camera on a tripod
SB800 flash in the hot shoe with the head aimed forward
Nikkor 17-55mm f/2.8 AF-S DX lens
f/2.8 and 1/60th sec unless noted otherwise
The location was indoors during daytime with blinds closed and a low intensity incandescent ceiling lamp on
AF-S Focusing unless noted otherwise
How Subject Position Affects Flash Power
In this set of tests I used a dark blue bathroom towel as a background and I placed a round white paper targets (the 'subject') at various places within the frame to see how the power of the flash would adjust. The reason I used a bathroom towel was that the texture of the towel provided a good AF target and I consistently got sharp and solid focus.
Image 1 Reference image, no flash , f/2.8, 1/1.6 sec
In Image 1, I used available light only, just as a reference. Notice the color of the bright 'blue' towel has shifted due to the Auto WB, which tried to make it 18% gray. The histogram shows a narrow peak representing the narrow range of brightness values for this image.
Image 2. Flash TTL, f/2.8, 1/60th sec
Image 2 is also for reference. It also hows how the Auto WB gets the color right when flash is used. The histogram shows a similar pixel distribution as in Image 1.
Image 3. Flash Off, f/2.8, 1/60th second
Image 3 is another reference image using the same f/ stop and shutter that will be used for the rest of the images in this study. Notice that the ambient is so low that will not be a factor in any of the remaining images in this study.
Image 4. White disk top right corner, Flash TTL
Image 4 shows the effect on the histogram of introducing a white disk of about 3 inches in diameter into the upper right corner of the frame. If you compare this to image 2 you can see that the background (the towel) has gotten slightly darker as a result of the flash metering system 'seeing' the white disk. Now, carefully look at the histogram, and you can see that the main hump of pixels representing the dark blue towel has moved left, and there is a small hump of pixels bunched against the very right edge. This small hump of pixels is the white disk, and its presence as a highly reflective subject has forced the flash power to decrease, which is why the background (the towel) got darker. The fact that the pixels for the white disk are climbing the right edge means that the flash has blown them out. Those pixels are overexposed and if there were any detail in such an area, it would be unrecoverable.
Image 5. Disk in upper right corner, Flash TTL-BL
Image 5 shows the effect of using the flash in TTL-BL mode. Comparing to Image 4, which used TTL, we see that TTL-BL sees the disk and tries to match it to the background. The result is actually quite good due to the fact that TTL-BL also uses distance information from the lens to set the power. This all has the effect of making the entire image much darker and the background is nearly black. The pixel hump representing the background has moved nearly to the left edge of the histogram, while the smaller pixel hump representing the disk has also moved left so it is no longer climbing the right edge. There are no blown out pixels in this image and it would be usable.
I am skipping Image 6 and 7 where I moved the disk to the bottom left corner, because the brightness and histogram distribution were essentially identical to Images 4 and 5.
Image 8. Disk centered, Flash TTL
In Image 8, I moved the disk to the center of the frame, and used the flash in TTL mode. You can see that now the flash metering system is more greatly affected by the disk more than it was in Image 4 when the disk was in the corner, and it decreased the flash power further. You can tell by the left shift of the background hump (towel pixels), and if you look closely, you will see that there is one thin line of pixels climbing the right edge indicating that the disk is totally blown out. So, even with the substantial darkening by moving the disk to the center, it is still not enough to prevent the disk from being totally blown out. This is because the disk is smaller than the center weighted measurement area, which allows a considerable portion of the dark towel to influence the metering. For proper monitor preflash measurements, the subject must be centered and occupy 18% of the frame. In this case it occupies only 7% of the frame, so lots of the dark towel is getting involved.
Image 9. Disk in center, TTL-BL
In Image 9, I switched the flash to TTL-BL and you can see that once again, the flash metering system tried to reduce the brightness of the subject (the disk) to match the brightness background (the towel). This moved the histogram to the left so that it looks essentially identical to Image 6, and indeed it should, because both were balanced to the dark towel. The fact that the disk was in the center made no difference. To state this another way, TTL-BL works differently when the background is darker than the subject. This is an area that is the subject of much research, and newer cameras than my D200 will likely deal with the dark background problem much better. It is my understanding that they are working on a way to switch to regular TTL when the background is dark. If this works out, it may obsolete regular TTL.
Focal Distance Tests Using a DX lens
Image 10. Disk in center, Flash TTL-BL, focus distance set manually to infinity
In Image 10, I am studying the effect that focal distance has on flash power.
I used TTL-BL and manually refocused the lens to infinity, and you can see the dramatic effect. Obviously, the image is no longer sharp, because it is way out of focus. The lens is a DX lens, which reports its focal distance, and it told the flash metering system that the subject was at infinity, so the flash increased to a very high power for the shot. I have no way to tell exactly what power the flash chose, but you can see that the histogram was shifted so far to the right that the background pixel hump is now in the center, and if you look really close, you can see the subject pixels are climbing the right edge compressed into a thin line, totally blown out.
Image 11. Disk in the center, TTL, focused at infinity
In Image 11 I switched the flash to TTL and refocused the lens to infinity. If you compare Image 11 to Image 8 you will see that they are essentially identical in brightness (not image sharpness), which indicates that the flash power was not affected by changing the the focal length of the lens. To state this another way, distance is included in the flash power calculation when using TTL-BL but not when using TTL.
1. In TTL flash mode, the subject reflects more light back to the flash metering system when the subject is centered. This is consistent with the comments I made in my earlier blogs that the flash monitors the center weighted area of the frame. I have shown through this study that the flash power is influenced more strongly by the subject when it is centered, but it is also influenced to a lesser degree all the way to the edge of the frame.
2. It is totally clear that in TTL mode the subject will often be overexposed if the flash ev is not reduced. I have found in my own wedding and events work that the flash works best when the flash compensation is set to about -1.0 ev for indoors.
3. TTL mode is NOT influenced by the distance that the lens reports.
4. TTL-BL mode is heavily influenced by the distance that the lens reports. I did not do a full study here, so I don't know exactly how much it increased, but by rough comparison with the flash in manual mode it appears to have increased to at least 1/2 power if not higher. This further implies that whenever we use TTL-BL mode we must make sure the distance from the camera to the subject is equal to the distance of the flash to the subject. In other words, don't use TTL-BL with the flash on a flash cable unless it is kept at the same distance to the subject as the camera.
5. It is totally clear that when the background is darker than the subject TTL-BL reduces the power of the flash in an attempt to balance the subject to the background. This works OK much of the time, but it is still best to use TTL indoors, where the subject is usually brighter than the background. However, I expect this conclusion to change as the newer versions of TTL-BL emerge on newer cameras.