Strange behaviour 580EX with omnibounce?

[René Damkot]

Well, here's a strange thing:
Recently got a 580EX, and it performes better than my 550EX.
Came across a stringe thing though: When using bounce flash with omnibounce it sometimes overexposes *a lot* even with negative FEC. When I take off the Omnibounce it works as expected (even underexposing a bit too much), *just* filling in the audience. (first shot -2 2/3 FEC, second shot -1 FEC !!)
(Bounce flash, even though the hall is mainly black, there's enough smoke to diffuse the flashlight and allow it to reach the subject; the 550EX without OB performed the same as the 580EX with OB: overexposing, so I don't think limited flash power is the reason for the second shot getting less flash exposure)
Full exif attached. Any thoughts? Sorry for the compression artifacts :p

[tim]

I can't explain this - the omnibounce should make little difference, assuming the flash can't directly hit the lens. I have a couple of related thoughts but nothing overly helpful:
- I would expect to have to use negative FEC in those scenes, they're quite dark and the camera will be aiming for 18% grey.
- I notice the darker photo is 1/250th, the lighter 1/200th. That could explain why the band on the stage is darker, but doesn't explain anything about the crown (since shutter speed is irrelevant).

I'll think some more tomorrow when i'm not so tired. By then Scott or Bob may have the answer for you anyway. The answer, for now, is don't use the omnibounce unless you're bouncing the flash.

Like the 2nd photo btw :)

[René Damkot]

Both pics were made using the flash pointing straight up.
I would expect to have to use -FEC as well, but not more than 2 stops...
I was thinking maybe the preflash would be to weak to be measured, so the flash thought "what the heck, let's give it all I've got" ;)
With the 550 I did some test once using manual flash output, and the pics came out more or less as I wanted at around 1/32 in similair circumstances if I remember correctly...
Maybe I'll have time to do some (more or less, as ambient is changing all over the place) controlled tests with 550 vs. 580, with vs. without OB...
Strange thing is, the combo 580EX - Omnibounce gives good results when using it as main light on the subject:
http://www.atak.nl/inc/img.inc.php?pic=plaatjes/actpics/051013bg_%20kowalski_voorn/rhd20051013BGKV087.jpg&s=fotos:%20Ren%E9%20Damkot

[scottbergerphoto]

Big dark backgrounds fool the flash into kicking out alot of light. You can zoom in to your subject, hit FEL and zoom out to take the picture or just know in advance that less flash is needed then what the camera will think. The opposite is true for big white walls or snow as background.

[René Damkot]

Big dark backgrounds fool the flash into kicking out alot of light.
Yeah, well, I know that, but the first picture is *no way* more then two stops under medium gray (look at the EXIF: FEC: -2 2/3).
In fact I would expect it to look like that *without* FEC. And why the difference with - without OB?
Oh, and zooming wouldn't do much good, since I use primes ;)
If I get around to testing, I'll give FEL a try as well... Thing is, without the OB, the 580 works pretty much as expected (read: as the 540EZ on my 1n would)

Exif:
Shooting Mode
Aperture-Priority AE
Tv(Shutter Speed)
1/200
Av(Aperture Value)
2.2
Metering Mode
Evaluative Metering
Exposure Compensation
-2/3
ISO Speed
1000
Lens
35.0mm
Focal Length
35.0mm
Flash
On
Flash Exposure Compensation
-2 2/3


Exif 2nd image:

Shooting Mode
Aperture-Priority AE
Tv(Shutter Speed)
1/250
Av(Aperture Value)
2.2
Metering Mode
Evaluative Metering
Exposure Compensation
-2/3
ISO Speed
1000
Lens
35.0mm
Focal Length
35.0mm
Flash
On
Flash Exposure Compensation
-1

[Jon]

I'm not entirely clear about the conditions for the two photos? Were they both taken with the Omnibounce in place? You talk about it being on, and off, and about different FEC settings, but the viewers I've used only show EC of -2/3, and no FEC. And the presence or absence wouldn't show up in the EXIF.

[René Damkot]

Both taken with 1D2 / 580EX, Av, EC -2/3 (otherwise the stage would overexpose)
First pic: With omnibounce, FEC -2 2/3
Second: without omnibounce, FEC -1.
EXIF in last message was copy-pasted from EOS-viewer utility, so it should be visible there...

[Jon]

OK - thanks. I'd suspect that the hazy look of the first one is backscatter from the Omnibounce's diffused all-around lighting. You may have been overestimating the effect the smoke would have on the undiffused flash, especially with a black ceiling. Just in case, was the flash at default (24 mm zoom when in "bounce" mode) setting, or had it possibly gotten manually zoomed (possible, even though auto-zoom doesn't function in bounce, or slave, mode)?

[René Damkot]

OK - thanks. I'd suspect that the hazy look of the first one is backscatter from the Omnibounce's diffused all-around lighting. You may have been overestimating the effect the smoke would have on the undiffused flash, especially with a black ceiling. Just in case, was the flash at default (24 mm zoom when in "bounce" mode) setting, or had it possibly gotten manually zoomed (possible, even though auto-zoom doesn't function in bounce, or slave, mode)?

The hazy look is the (direct light of the) omnibounce lighting up the smokey hall.
About overestimating the effect of the smoke: 'With the 550 I did some test once using manual flash output, and the pics came out more or less as I wanted at around 1/32 in similair circumstances if I remember correctly...' So no way the flash is out of power. We're talking ISO 1000, f/2.2, -1 FEC here...
Flash was at 24mm.
The problem I have is not "why is the second picture dark", but rather: "Why is the fist on massively overflashed, despite dialing in almost -3 FEC?"

[Jon]

Do you get similar results when there isn't smoke but the other conditions are similar? Frankly, I don't think your flash is doing much of anything in the second picture, and what it's doing isn't getting bounced back straight into the lens, so the shot is essentially available light shooting. In the first, the smoke is getting front-lit from the Omnibounce, but the pre-flash wasn't able to penetrate far enough through the smoke to get an accurate reading, so the system overestimated the required power setting.

One other thing - smoke won't "allow" the light to reach the subject by diffusing it; if anything it will interfere with it reaching the subject.

[René Damkot]

Well, If the second picture would have been taken without flash, the audience would have been pitch-black. Everything you see there is the flash. If I'ld have to guess, I'ld say flash output was around -1 1/2 to -2 stops. (While I set a FEC of -1).
I was thinking along the same lines about the preflash not being able to penetrate far enough...
If I have time and opportunity I'll do some decent tests in the same hall this weekend with the 550 and 580, both with / without OB... I'll see if there's a lot of smoke then. Difficult to control that factor though ;)
I do think the smoke lit by the flash and the ditto black ceiling work as a (terribly ineffective) reflector, since the 'lit smoke' is above (me and) the audience, it's more like a light source above the audience, so I get less interference by the smoke.

[gmen]

The pre-flash idea seems the likely candidate.

Might be a very good idea to avoid ETTL at all costs in these circumstances and get some control over the flash.

If you're shooting a batch of pics like the ones you've posted from the same distance why not use the flash on manual power? A couple of test shots and some chimping should enable you to tweak the manual power to get the illumination you really need... and no more chancing to 'luck' with ETTL.

OK... it's only an idea... :lol:

---- Gavin

[René Damkot]

Probabely a good idea ;)
It's just that I like to *understand* how something is working. That way I figure maybe I can anticipate on errors before they happen...
I'll make sure to include FEL and manual in my tests with both flashes...

[René Damkot]

Well that took long enough ;)
Forgot all about this thread.....
Solved it: the preflash isn't powerfull enough. Came across the same 'error' recentely, ,and found out FEL worked 'intermittently'. Turned out that if it didn't work the flash symbol in the VF was blinking: not enough light.