Flash mystery..

[lenelg]

I am a novice to TTL flash use, and decided to test a set of flash diffusers, to see which would work best for me. My first results are posted at http://www.flickr.com/photos/len_elg...7628131252489/ .

The results show up an unexpected mystery: Most exposures were spot on, but as you can see from the RAW Converter screenshots, a few came out heavily underexposed.

All photos were taken on the same occasion, using a Nikon D3100, an AF-S Nikkor 35 mm f1:1.8G lens, and a Yongnuo YN465 Speedlite in TTL mode. Exif data confirm that the camera was set at 1/60 sec, f1.8 and ISO 200 for all exposures, and that the flash had fired. I repeated the underexposed shots to make sure nothing spurious had caused this.

The flash modes I tested were:
- Bare flash head on (underexposed)
- Bare flash bounced at ceiling, built in reflecting card deployed (exposure OK)
- Lumiquest Pocket Bouncer (underexposed)
- Lumiquest Soft box (underexposed)
- Lumiquest Soft box aimed 45 degrees up (OK)
- Homemade softbox (OK)
- Demb Big Flip-it bounce card, straight up (OK)
- Demb Big Flip-it bounce card, leaned 15 degrees behind vertical (OK)
- Demb Big Flip-it bounce card, straight up, front diffuser mounted (OK)

I will of course have to repeat the experiment in a different setting. Meanwhile, does anyone have a clue what is going on here? My best guess is that when the flash is aimed straight forward, a reflected hotspot fools the TTL to underexpose. But I cannot really see this in the lighting patterns.

[Petiej]

Asking which metering mode are you using? Evaluative, partial,spot or center weighted..? Thinking maybe the focus point may come into play. Just saying, for example, if the camera were focusing on one of the dark or light objects in the room to obtain a reading.

[Colin Southern]

Hi Lennart,

A lot of people misunderstand flash diffusion. In a nutshell, the softness of the light is proportional only to the effective size of the light source (ie "how big the light source is" and "how close to the subject it is"). So putting things on the flash generally doesn't make any difference in terms of diffusing what comes out of the flash (because the effective size isn't changing to any significant degree). What CAN make a difference though are when they spread the light around the room, as the effective light source essentially becomes the walls and ceiling (although bouncing off ceilings normally gives very unflattering "office" lighting).

[lenelg]

Asking which metering mode are you using? Evaluative, partial,spot or center weighted..? Thinking maybe the focus point may come into play. Just saying, for example, if the camera were focusing on one of the dark or light objects in the room to obtain a reading.

According to Russell MacDonald´s excellent "Nikon CLS practical guide" , Nikon´s TTL system uses centerweighed metering for the flash, no matter what the camera is set to. Apparently, ambient light and pre-flash are measured independently. But I will test if this makes a difference..

[lenelg]

Colin,

What you describe is what I set out to test: I expected to see differences in the hardness of the light, but not such dramatic differences in overall exposure - which is why I am still puzzled..

[Colin Southern]

Colin,

What you describe is what I set out to test: I expected to see differences in the hardness of the light, but not such dramatic differences in overall exposure - which is why I am still puzzled..

Hi Lennart,

I can't help you there I'm afraid (I'm a Canon shooter) - but I'd be very surprised if you see any significant difference in the hardness of the light, as the size isn't changing (to any significant degree). Easiest lighting I've found is just to fire a flash into a shoot-through umbrella (which can be positioned close to the subject) (which results in an effective light source area thousands of times bigger).

[dmwj80]

I agree with Lennart. His post was very informative.

[Colin Southern]

Out of interest Lennart, did you take multiple exposures for each test? I've heard of 3rd party flash units giving inconsistent results on some occasions.