Weird ettl ec behavior - is this just me?

[Marsu42]

With daylight fill/bounce flash I'm mostly shooting Av mode with fixed iso 800-1250, min. shutter speed set (1/60s - 1/250s on my 60d) and a little negative flash ec to prevent overflashing the subject or positive flash ec because the diffuser eats too much light. I'm also often using camera ec to control the subject-background relation, so most of the time it's set between +1/3 and +1 to allow for a natural background.

The weird behavior I am experiencing: When changing *camera* ec a little from 0 to +-1/3 the background exposure changes a *lot*, while the following 1/3 steps are relatively minor - so on my 60d+600rt camera ec is not a constant, gradual scale. Is this just me, my camera or my flash? Is the flash exposure depending on the camera body used, or do all the ettl2 Canon cameras meter the same?

[7enderbender]

With daylight fill/bounce flash I'm mostly shooting Av mode with fixed iso 800-1250, min. shutter speed set (1/60s - 1/250s on my 60d) and a little negative flash ec to prevent overflashing the subject or positive flash ec because the diffuser eats too much light. I'm also often using camera ec to control the subject-background relation, so most of the time it's set between +1/3 and +1 to allow for a natural background.

The weird behavior I am experiencing: When changing *camera* ec a little from 0 to +-1/3 the background exposure changes a *lot*, while the following 1/3 steps are relatively minor - so on my 60d+600rt camera ec is not a constant, gradual scale. Is this just me, my camera or my flash? Is the flash exposure depending on the camera body used, or do all the ettl2 Canon cameras meter the same?

There are a few things that can make ETTL act a little weird or even inconsistent. Not sure about that specific camera-flash combination but I've done a bunch of testing a while ago together with some folks over at dpreview. I'd have to go back and look at the details but two things stood out when using a 5DII with a 580EXII: one was that ETTL behavior when the flash head is straight or angled came in different (less so with my 430EXII). The other one was that it can vary with which lens you use. Some lenses transmit their distance settings, others do not. The ones that do become part of the "TTL" calculation - and sometimes that's weird. In a way you could even say that these lenses (my 24-105 for instance) under some circumstances actually leads to a calculation in camera that is not really "TTL" but rather the good old GN/distance thing. Lenses without the transmission of this information came in a lot reliable.

So when you're compensating these factors may throw things off a bit. Ultimately, that is one reason why I've started to use manual settings a lot more instead of relying on ETTL all the time; at least in environments where light doesn't change that drastically.

When using ETTL I also find that Av is not the best mode but that manual camera setting are really the best for that. Syl Arena mentions that in his book also - something to the extent that M is really the default "automatic" ETTL mode. I know that Av is always said to be for daylight/fill flash. May be worth testing if other settings work better for you.

[Marsu42]

When using ETTL I also find that Av is not the best mode but that manual camera setting are really the best for that. Syl Arena mentions that in his book also - something to the extent that M is really the default "automatic" ETTL mode. I know that Av is always said to be for daylight/fill flash. May be worth testing if other settings work better for you.

My lenses transmit the distance information (though some are said to be more or less precise at it) - so it's "real" ettl2. But I recently also decided to practice more full m mode for flashing with ettl, esp. since Magic Langern now has a expo lock function that allows quickly changing aperture/shutter/iso without changing the exposure. And I know that Av mode has some quirks, I was just wondering if the camera ec behaves this way for all photogs or just me...

... but you seem to recommend the book Speedliter's Handbook: Learning to Craft Light with Canon Speedlites ? The other recommendation I read about is Canon Speedlite System Digital Field Guide

[Lawliet]

The weird behavior I am experiencing: When changing *camera* ec a little from 0 to +-1/3 the background exposure changes a *lot*, while the following 1/3 steps are relatively minor -

In my experience minimal changes in composition can throw exposure of - to you it looks the same, but some AF points lock/see something while others don't and that factors in the metering. (iFCL made me appreciate the handheld flash meter even for continuous light photography, and ETTLII doesn't give you a warning via the values in the viewfinder.)
And don't remind me of ETTL and obtuse angles or anything specular...

[DanielW]

... but you seem to recommend the book Speedliter's Handbook: Learning to Craft Light with Canon Speedlites ? The other recommendation I read about is Canon Speedlite System Digital Field Guide

The first one, from Syl Arena, is a terrific book; you won't regret getting a copy.
Daniel

[rlarsen]

I've never been overly impressed with Canon's auto-flash metering. I use RadioPoppers with my speed lights.

I alternate between TTL and manual, and when I use TTL I do a fair amount of exposure compensation. TTL is easily fooled by light and dark clothing as well as composition changes and background light. Flash exposure lock can be very helpful but the timer creates real limitations in convenience. With my older Canon 1-D MK ll bodies I changed the timer in custom settings to hold the locked exposure for a long time. My 5D MKlll doesn't have that feature.

I never understood why focus distance data wasn't more helpful in exposure accuracy.

The nice thing about TTL though is that you can have a speedlight close to the subject and still shoot at wide apertures like 2.8  Not really possible with manual flash setting, even at 1/128th power.

I hope Canon makes more improvements to auto flash metering in future cameras and firmware updates.

[digital paradise]

That is odd. Gotta think about it. Can you post some shots?

[brad goda]

any setting other than full manual exposure for flash and camera will give varying exposure and flash output.   setting one of either flash or camera on manual will reduce exposure changes... You have a good system... 60D and 600EX are good products... although the newest cameras now have rgb metering which reduces auto and TTL flux lots.. its still not perfect.
with changing focus points meter settings and background light exposure and compensation IS alot to ask the camera to do on a consistent level +- 1/3...
digital has immediate image proofing not like how film was... needing to pull polaroids...even with 35mm... trust your self to shoot manual and to trust your lcd for base exposure and setting fill.
good luck... have fun on the problem solve.

[Marsu42]

That is odd. Gotta think about it. Can you post some shots?

Unfortunately I always deleted them right away and set the correct camera ec, but next time I experience this problem I'll post them here.

You have a good system... 60D and 600EX are good products... although the newest cameras now have rgb metering which reduces auto and TTL flux lots.. its still not perfect.

Thanks for everybody's answers, and yes - my system is very good, after all afaik only the 1dx series has rgb metering (and of course cheaper Nikons, doh). That's why I'm so spoiled and expect my system to work all the time. And for bounce/fill near-macro shots and outdoors fill flash it usually works very reliable, though I have to fiddle with both ec values a bit to get what I want - but in comparison to full m flash this is of course extremely convenient.

But as written above I'll start shooting full camera m more often, just bought the Syl Arena book and will experiment with m flash more often - it's a lot to learn when coming from the "p&s" av+ettl2-experience, but I acknowledge it's more reliable in changing light or with reflecting surfaces, so I have to know how to do it.

[Chris Burch]

Just a thought that occurred to me as I was reading.  ETTL (especially when set to evaluative) can get thrown off by a lot of things, but the camera specifically looks for the highlights in the scene to avoid blowing them out.  That's why if you get a direct reflection of the flash off of something like a mirror in the background, your whole picture will drop down to a dark exposure based on the camera lowering the flash power.  If you change the EC of the image, the highlights of an image may also shift once the flash is applied.  For instance, if you're shooting someone in a white shirt against a light wall and you increase the exposure, the added flash power make the near-ground subject (person in white shirt) brighter in relation to the background wall.  Whereas under the original setting, the background may have been the brightest area in the pic and the target for eTTL exposure, the new values make the person's shirt the highlight of the image and the eTTL calculations adjust accordingly.  I would advise you to only use FEC when you're using eTTL.  That will only affect the flash power applied and minimize the number of variables.  I advise most people to only use one "auto" setting, which includes eTTL.  Then you don't have to worry about the effects of each change on the calculations of another variable setting.  So shoot in manual mode and if you want to change the flash ambient ratios, change your shutter speed or ISO.

[Lawliet]

after all afaik only the 1dx series has rgb metering (and of course cheaper Nikons, doh).

side note:
Depends on how you define RGB-metering. The iFCL-capable cameras are actually dichromates, not the best color sensitivity, but a rough idea instead of color blindness. Basically they can detect red and blue, but green and white look the same to them.

[Marsu42]

Depends on how you define RGB-metering. The iFCL-capable cameras are actually dichromates, not the best color sensitivity, but a rough idea instead of color blindness. Basically they can detect red and blue, but green and white look the same to them.

Ok, didn't know that, thanks - I'm always learning a lot here :-)