If I understood TTL correctly, say you are in M mode with a camera and you have intentionally underexposed. The flash in TTL mode theoretically would try to add enough light to correctly expose the scene.

Question is, the calculation is supposedly calculated from a preflash that happens once. Wouldn't it need to at least try a few times before getting it right? I mean, each scene is different, and the increase in power has a different effect in each scene.

In fact, the same question applies to other modes than M. Suppose you are in automatic mode. The camera already exposes the scene correctly. I suppose the preflash will try to "fill" the scene without overexposing it. But how would it get it right from just one trial? Say it went overboard during the preflash, how would it know how much to back off?

  • The short answer is that it's pretty dumb and gets it wrong on a regular basis! If you shoot somebody in a high vis jacket, it sees the flash bounce straight back off one of the reflecting strips, thinks "wow there's a lot of flash coming back, I don't need much power at all" and then ends up massively underexposing the photo.
    – Matt Grum
    Jun 27 '14 at 11:57

There isn't nearly as much to figure out as you seem to think.

Let's say that you have a scene in front of you that is nicely illuminated and doesn't really need flash at all, and you meter for, and set a manual exposure for, an ambient exposure that would have been absolutely perfect. Then, for some inexplicable reason, you decide to add manual flash (metered with an external flash meter and using manual flash settings, just to keep the camera's brain out of the equation) with no compensation at all. The result will be a one-stop overexposure (twice as much light as needed for a proper exposure) of the parts of the scene that are illuminated fully by both the ambient lighting and the flash, and one stop over isn't exactly an unrecoverable tragedy most of the time. So the worst a camera can do, assuming the TTL metering for ambient and flash are not connected in any way, and that the system is as brain-dead as the all-manual scenario described, is a one-stop overexposure of some parts of the image (say, certain planes of a subject's face). If you're using an older camera, or one of the less-sophisticated current or recent models, you can probably confirm that that's exactly what will happen when flash isn't needed and no compensation is dialed in.

The metering can be pretty unsophisticated. A single medium-low-power flash burst is fired (1/32 power is usual); a power level that is both high enough to register if full power would have been warranted, but low enough that it won't overpower the metering circuit altogether at the flash's minimum specified distance. Remember that the flash meter doesn't have to create a nice noise-free image, nor does it have to use nice small pixels to record image details. The metering sensors can have a much greater dynamic range than the imaging sensors, especially in an SLR-type camera where it's easy to use a completely separate sensor for metering. It's a simple thing to scale that reading up or down to a level that would have created a proper exposure for the "interesting" part of the picture. (The flash meter, if it's metering the whole image in segments, "knows" that it's unlikely that the flash will illuminate the entire scene evenly. It will either base the metering on the brightest part of the image or on the part of the image that is in focus or selected for spot metering - if you happened to be using flash exposure lock and a meter-and-recompose strategy.) And it may (depending on the camera) take into account ambient lighting as well as flash. Again, that can be done very quickly, since the metering sensor doesn't have to get a proper exposure, it just needs enough of a reading to tell what would happen at the settings you're using. It is not the case that each of the flashes is metered independently or that the system needs to make several attempts to arrive at a correct exposure; everything can be determined from one master flash exposure reading.

After that, it's simple matter of telling each of the flashes what power level to fire at, given both the metering and the exposure compensation you have selected for the group the flash belongs to. That may take a while with an optical triggering system, since it involves timing pulses, channel and group codes, and a power setting for each of the groups you have active in your flash setup. If you are using multiple flashes in multiple groups, how you have the flashes arranged will have a great impact on the resulting exposure - you may have several flashes ganged together in a modifier; you may have flashes in the same group at different distances from elements in the scene, or have different modifiers on each, etc. It is easy to make things go very badly with a multiple flash TTL setup if you don't understand that (which is probably why a lot of folks bail out of TTL and go all-manual instead of trying to figure out what the problem is). TTL flash is just an aid; you still have to understand the basics of lighting when using more than one flash.


The metering flash is almost always relatively low powered. The camera compares the amount of light the metering flash produced to the amount of the pre-flash light that is reflected back to the camera by the subject. It then computes what percentage of light was returned from the metering flash, assumes the same proportion of light will be returned at any particular power level and then computes how much light must be output by the flash in order to get the reflected amount of light at the needed level.


It only requires two values to figure out. Before the pre-flash it knows how much light is ambient. For the pre-flash it knows how much light the scene gets from the flash at a fixed power. It knows how much more or less light is needed for a standard exposure and simply sets the flash power accordingly since it knows exactly how effective the flash is from the difference between the pre-flash and the ambient scene and knows exactly how much as to be added overall from the ambient scene.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.