A question about ETTL and underexposed picture

Hi, I thought I knew the Speedlite well enough. However, tonight when I took pictures of a few friends in a restaurant, I encountered the following situation that I don't understand. Please share me your wisdom.

It was dark (as expected). There were about 8 people. I used wide angle. In order to get everybody in focus, I used f/8. Shutter speed was 160. ISO=100. On camera top I had a 430EX II set to ETTL, pointing to a low ceiling which was just 3-4 ft away. I intended to bounce the light. The ceiling was covered with black material, but it's pretty low.

The first shot was totally underexposed. I could hardly see people's heads. I thought maybe the first shot was just a test. Yet the second shot was equally bad. Then I used f/5.6. Faces are hardly distinguishable. Finally I set ISO to 800. Then everything looked OK.

It was evaluative metering all the time.

My question is: Why couldn't the Speedlite that was set to ETTL give me the correct exposure in the first few shots?

Not enough light (flash power) for f/8 at ISO 100 while bouncing off a black ceiling, which is why changing the ISO to 800 worked.

What Dave said above, not enough flash power. With flash at max power it can only do so much eTTL or not.

Oh, I thought it should be powerful enough. Thanks for your explanation, Dave and Bobby!

I've made up a couple of these for my 430EX and 430EXII Speedlites..

They work wonders when bouncing..

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

yogestee wrote in post #12029247
I've made up a couple of these for my 430EX and 430EXII Speedlites..

They work wonders when bouncing..

Still won't help if you shoot at f8 with ISO100. If flash power is max out nothing else will help except increase flash power by adding more lfashes, more powerful flash, increasing ISO or using wider aperture, or combination of all these.

With the flash in bounce it has 50mm coverage angle and ISO100 GN...130. So shooting with flash head straight ahead, it should reach almost 15' with f/8 on the lens. Normally bounce loses about -1.5EV with a typical home white ceiling, reducing the effective max distance.

Speedlight guide numbers have been proven (over and over, via flash meter verification), though. Most every speedlight we have measured (Canon, Metz, Sigma, etc.) is -1EV less powerful than claimed by the manufacturer. For example, my Metz 54MZ (very close to 580EX, just a bit less power) should reach 14' at f/8 (with 'normal' lens coverage angle) per its LCD display...but measuring with a flash meter, it only gets out to 10' at f/8; not GN112, but GN80... -1EV! Bouncing to that same distance (10') I have to have f/4 +0.3EV on the lens in order to get a proper exposure! The flash is underpowered...and that is what you ran into. It was made EVEN worse by bouncing onto black fabric rather than a matte white ceiling!

Thanks, Wilt! I usually didn't pay attention to all the math about GN. But now it proves to be something I should not neglect.

bobbyz wrote in post #12031735
Still won't help if you shoot at f8 with ISO100. If flash power is max out nothing else will help except increase flash power by adding more lfashes, more powerful flash, increasing ISO or using wider aperture, or combination of all these.

I realise this..

The only thing you forgot was shorter distance between flash and subject.

My rule - indoor flash - automatically go ISO 1600!

Also f/8 is overkill - I can get 8 people in focus at f/2! use a lower aperture, at least something reasonable like f/4 or something. Just get them into 2 rows or something with front people crouching a bit

I don't necessarily agree with smorter.
If you understand that using artificial light is a combination of 2 exposures (ambient & flash) increasing the ISO will not only affect exposure of ambient, but also flash.

Also, the focal length of the lens and subject distance from camera will determine how much DoF you have.

If you want to expose the subject AND the fore/background, increasing ISO will resolve the issue, but introduce a lot of noise. I would suggest that you lower your shutter speed (to capture some ambient light) and open up your aperture (to allow for more flash exposure), increase ISO an an acceptable level as needed (affects both ambient & flash exposure), and maybe lastly increase FEC on your flash.
That's generally what I do when I shoot flash on camera.

Another tip is to pop your flash on Manual instead of ETTL. Since you're shooting indoors where the lighting is consistent, why rely on ETTL to guess the correct exposure for you when you only have to take a few test shots, adjust to find the settings, and you'll get consistent results every time?

I only say this because people wearing black or white can throw off the ETTL system. Same reason why you want to put your camera on manual in those types of situations too!

You have a point in that I am being extreme. It is what I do, but not suitable for all situations

I do have to disagree about ETTL vs manual though. ETTL is excellent but you have to know when it is (designed to) fail. It handles white tablecloths horribly, but it is consistently bad, and reliably bad...if that makes sense. It is totally consistent and reliable if you know it's weaknesses.

I would never dare use manual flash for anything other than setups or off camera flash. ETTL takes just a fraction of the time and headache of manual flash trial and error

Bold part is incorrect - changing aperture affects ambient just like changing ISO. I don't know when this shutter controls ambient and aperture controls flash nonsense will stop. Hope Dave Hobby and folks give proper information on their blogs.

smorter wrote in post #12037145
. ETTL is excellent but you have to know when it is (designed to) fail. It handles white tablecloths horribly, but it is consistently bad, and reliably bad...if that makes sense. It is totally consistent and reliable if you know it's weaknesses.

Just curious, Smorter...do you shoot with ETTL in Evaluative CFn, or in Average CFn?