Pentax P-TTL: Why and when does underexposure occur and what can I do?

Question

Pentax's P-TTL flash metering system tends to be unreliable and may underexpose at times. When and why does the system underexpose and what can I do? I want to know the situations that cause flash underexposure so that I can compensate for it accordingly, with as little trial-and-error as possible.

Note that I often manually set the ISO lower than what the Auto ISO would choose in order to reduce noise (but not so low that the flash doesn't have enough output), even though this means that the flash will need to discharge at or near full power. This often, but not always, causes underexposure unless exposure compensation is applied. In addition, direct flash never seems to cause underexposure, but the appearance of direct flash is generally poor, with heavy shadowing and red-eye detracting from the image.

The flash unit I'm using is a Metz mecablitz 50 AF-1 digital (guide number 50 m / 164 ft at ISO 100, zoom 105mm), Pentax P-TTL dedicated, running the latest firmware (3.0). The flash is primarily being used with my Pentax K-5, but both my K-5 and K-r suffer from this problem, and apparently to the same degree.

I'm now suspecting that the metering pre-flash is too weak, causing the camera to incorrectly assess the amount of flash exposure required. The camera's flash metering algorithms are also a possible suspect. Would a more powerful flash unit produce more accurate flash exposures, even if the 50 AF-1 does have enough power to correctly expose the scene with a given set of exposure parameters?

Edit

Upon further testing, it seems part of the problem is distance to the subject--the camera often underestimates the amount of flash power needed to properly light the subject at longer distances. This could be caused by the camera or the flash unit, and it's possible that the flash's pre-flash timing or power is inconsistent with what the camera body expects (keep in mind this is a third-party flash unit). Any workarounds? (I do know that the lens communicates focus distance information to the camera, but it does not appear to be taking advantage of this information, other than adjusting the aperture when not set manuallly.)

Answer 1

I have used P-TTL flash on some events, and have generally been quite satisfied with the results. I don't like the "deer in headlights" look of a direct on-camera flash, so I've only used bounce flash, or off-camera flash. That said, here are some ideas why underexposure might be experienced and how to avoid it.

In exposure modes where shutter time is not specified by photographer, the camera caps shutter time at 1 / (crop_factor × focal_length) when a P-TTL flash is used; if the calculated cap is faster than sync speed, sync speed is used. By this, the camera tries to avoid blur from camera shake, expecting the flash to provide enough light. This cap is applied even if the flash is obviously unable to provide enough light by itself. In Program mode, such cap will not be compensated by using larger aperture. Switch to Manual exposure mode to choose both aperture and shutter speed yourself. (Note that under sync speed, lower speeds will not gain you more available flash power, only the amount of recorded ambient light will increase.)

The flash might simply lack enough power; try using the flash with manual power setting at full power to determine if this is the case. Setting ISO near the capability limit and shooting different subjects while moving around, sometimes the flash will not have enough power. One way that power might become lacking is allowing camera to change shutter speed faster that sync speed; this will require the flash to pulse in High Speed Sync mode, considerably decreasing ability to light. Or the camera could be choosing a smaller aperture than when you were figuring out how low ISO you can pick. Use wider aperture and/or raise ISO to ensure sufficient firing power; switch camera to Manual exposure mode if shutter speeds get faster than your camera's sync speed.

• For calculating distance that could be covered with a flash, knowing and understanding the guide number might be useful. For bounce flash, the whole trip from flash to bounce area and back to subject has to be considered, with some loss accounted in depending how reflective the bounce surface is (say, about 1 stop for a white ceiling). Hint: stadiums, landscapes, and bounce flash in large halls is out of range whatever your GN is.

Pentax metering tends to sway towards avoiding clipping highlights when in doubt; with direct flash, reflectors, mirrors, chromed surfaces and even nearby light-colored objects may cause the camera to be careful and underexpose your subject. With bounce flash, you should be pretty safe here unless some of your flash reaches the reflective surface by a direct route. Use something like a black foamy thing to shield your flash from shining towards scene directly. Shoot in RAW for better ability to adjust exposure in post-production; you can't control the ratio between ambient and flash in post, but for precise control, you should prefer manual flash to P-TTL anyway.

Behavior of matrix metering is hard to predict in settings with varying lighting. You could try using spot metering to take control over what your camera meters by. Automatic exposure also aims for 18% gray, so if you're metering from a lighter surface, dial in some exposure compensation accordingly. Understanding where a subject should fall in Zone system would not hurt.

Answer 2

As a general rule, the exposure can be very sensitive to light reflected by shiny objects in the scene (such as door handles, windows, mirrors, etc). In some cases you simply can't guess the amount of light returned by these objects, so controlling your flash manually may help.

Alternatively, you may want to try the A-mode of your external flash instead of PTTL if your flash supports it. It is the mode when your flash uses its own light sensor to accurately measure the light returned by the object. This sensor is not sensitive to light reflections by small objects and as a result is more consistent (at least with Pentax cameras).