How does E-TTL II handle a bounced flash

[vab3]

I'm currently using an auto-thyristor flash (bought used from the

photo.net classifieds, no less), and I understand how it works with

a bounce card or diffuser, because the flash measures how much light

is reflected, and stops the light when it is enough.

 

But the E-TTL calculates the amount based on information from the

camera. How does it know if the light is going to bounce off the

ceiling, and thus travel twice as far to the subject?

[rog21]

It still measures the returned light coming through the lens. In bounce mode, ETTL2 throws out the distance information anyway and measures the duration of the flash in the same way as ETTL1 did/does.

[mawz]

E-TTL (And E-TTL II) meter based on a pre-flash fired immediately before exposure, read by the camera meter (As far as I'm aware, E-TTL uses the ambient meter, which is what allows the camera to use spot metering for flash exposures). This gets around the issue that you can't use TTL-OTF metering with digital sensors (Unless you're Fuji, and they dropped that functionality for i-TTL with the S3 Pro).

 

E-TTL II uses focus distance information to improve it's calculations, but the metering is essentially similar to plain E-TTL.

[vab3]

Thanks for the responses.

 

So, when you press the shutter, it flashes quickly, takes a measurement using one of the cameras many TTL metering options, opens the shutter, then flashes again based on the calculations and measurements.

 

Is that right? Does this cause a delay?

 

If you turn off the pre-flash, I'm assuming the camera and flash just figure it out, like someone would with a manual flash?

[jim_larson1]

<i>E-TTL II uses focus distance information to improve it's calculations, but the metering is essentially similar to plain E-TTL.</i>

<p>

The flash "philosophy" of firing a preflash and monitoring the results on sensors is the same, but how the information is handled is very different. For ETTL, the calculated power output is strongly biased by the sensors in the region of the focus point. For ETTL-II, the focus point is ignored, and more of an averaging algorythm is used.

<p>

"distance feedback" is used as cross check to avoid overexposure (ie, the camera knows that it should never fire full power straight ahead when the lens is focused at 3 feet). There is also adjustments to ignore black areas (ie, you just can't properly expose clouds in the background at night!), and to tone down the flash if any one area is super overexposed.

<p>

The later feature is what probably results in so many complaints of ETTL-II underexposing. People have not yet figured out that you can't properly expose a frame when you have 1/4 of the frame being a filled by a white table cloth in the foreground.

<p>

<i>So, when you press the shutter, it flashes quickly, takes a measurement using one of the cameras many TTL metering options, opens the shutter, then flashes again based on the calculations and measurements. </i>

Yup. That's how it works. Except that with a 20D/1D/30D there are only TWO TTL metering options: CF14=0 (evaluative) and CF14=1 (averaging). Delay in inperceptable.

<p>

<i>If you turn off the pre-flash, I'm assuming the camera and flash just figure it out, like someone would with a manual flash?</i>You can't turn off the pre-flash. Without the preflash, the camera can't meter the scene. With an external speedlight (550EX/580EX), you can set the flash for "manual exposure". When you do this, then there will be no preflash. . .as the camera is not performing any flash metering.

[awindsor]

Unlike several of the KM cameras the delay on the Canon cameras is short enough that I have never had a problem with blinking people. However there is apparently a problem with birds who are able to blink fast enough to be captured. There is even an acronym for the phenomenon BEETLe (Blinking Eye E-TTL).

 

If you want to turn off the preflash then you need an auto flash (unlike Nikon, Canon does not build this functionality into their flashes) or use manual flash (find in a studio - pretty terrible in most other situations).

[jim_larson1]

True. You also sometimes end up with BEETLe if you use second curtain sync.

[john_schwaller]

ETTL uses two different meterings to determine the flash...an ambient reading and the pre-flash metering. The different flash CF flash modes were specific for turning on/off Auto Flash Fill Reduction (AFFR).

 

With ETTL, it uses the focal point to determine the "subject". The algorithm uses the differences between the ambient and pre-flash readings to determine the flash need.

 

ETTL-II builds on the ETTL base, but has a different method to determine the "subject". When CF-14 is set at the default '0'('evaluative' FLASH mode, which has nothing to do with 'evaluative' AMBIENT mode), the camera will measure the differences in each metering zone (different for 1-series vs. xxD/xxxD). Those which show a slight difference are assumed to be the subject and are combined/averaged and used by the algorithm to determine the flash required. Those with no change (below a threshhold?) are assumed to be background and those with a large change are assumed to be a reflective surface and are thrown out or downweihted.

 

With CF-14=1 (averaging...again, flash not ambient) all of the flash meterin points are compared to the ambient readings, averaged together and fed into the algorithm to set the flash. Canon will not discuss specific of the ETTL-II algorithm, but many believe that '=1' continues to turn off AFFR and the manual wording, taken literally, would seem to support that.

 

Personally, I find that CF14=1 is better indoors and gives more consistent flash lighting...this is in 'M' mode usually with +2/3 to +1 FEC, which needs to be adjusted +/- for the scene/subject. For outdoor, I use CF14=0 in 'Av' or 'M' and, for fill with -FEC....but each scene is different.

 

Sorry to go on so long....JOHN