April 18, 2010

Canon E-TTL flash settings – Average vs Evaluative flash metering

With TTL flash, (or E-TTL, as Canon call their specific flavor), the camera and speedlight working together according to various algorithms to control the flash exposure. The E-TTL flash exposure will therefore depend on various factors – the tonality of the subject and scene; the brightness of the scene; and how the camera interprets the sections of the metering pattern.  Other factors quite possibly also includes data from the lens.   How these factors inter-relate, we can only make educated guesses; and many photographers have taken time and effort to do test shoots to see how the cameras and speedlights work.

To make things even more interesting, Canon offers two modes of E-TTL flash metering: Average and Evaluative. This is set on the camera body via the custom functions.

The way I understand this to work, is that with Evaluative flash metering, the Canon camera takes the ambient light into account when calculating the flash exposure. With Average flash metering, the flash metering would appear to be de-coupled from the ambient metering, and the camera is less biased by the available light.  (I’m quite willing to be corrected on this.)

So which E-TTL flash exposure mode to use? Average or Evaluative?
In the end, I work in a fairly simplistic way …

With Canon I mostly keep it to Evaluative TTL flash metering. I then adjust my flash exposure compensation to taste. I do this by pre-judging the tonality of my subject and scene, and making an educated guess as to how much FEC would be needed. Then I fine-tune this by looking at the image on the camera’s preview.

This does strongly imply that you have to ride the FEC as you check your results while you shoot. Shoot, check and adjust.

Ultimately you HAVE to get used to how your camera and flash responds, by getting familiar with your equipment and shooting a lot. This to me is the key point here .. being able to roughly predict how my camera will react, and what the flash exposure would be like as a result, and dialing in a certain amount of FEC before even taking the shot.  It comes with experience and shooting a lot.

The general approach:
Keep to Average TTL flash metering mode if the flash is a dominant light source.
Change to Evaluative flash metering when the flash needs to act as fill-flash or when the flash is in relation to the ambient light .

This is the approach as generally advised on various websites and forums, and is how I used to do it with the Canon 5D / 1D mk II / 1D mk II N / 1D mk III

I would be change between the two modes, picking Average TTL Metering when the flash was a dominant source of light … and Evaluative TTL Metering when I wanted fill-flash, or needed the flash exposure in relation to the ambient exposure. With the Nikon D3, I would just use TTL BL mode and adjust from there. Since TTL flash metering is dependent on the tonality of your subject / scene (and all the other factors), you will have to ride your FEC anyway.

I mostly kept my Canon D-SLRs set to Evaluative flash metering, just riding my FEC higher to make up for the difference between how the camera.

My take on it is that I am better of using one of the modes the majority of the time, and get a feel for how my camera and flash would react. Rather that, than jumping between the modes, and hoping the camera and flash will sort it out, without much input from me as the photographer.

Once again, this means that there has to be some familiarity in how a specific camera and speedlight works together. This, for example, is how I noticed that I needed to adjust the FEC on my D3 bodies differently than I was used to on the D2x and D2H … which was close to what I was used to on the Canon D-SLRs I have used.

As an example – One instance where I know I would need to dial in a lot more flash exposure compensation, is with a back-lit subject.  The Canon flash system seems to be easily influenced by strong back-lighting, especially in Evaluative TTL metering mode.  So I would instantly dial in at least 1 EV more on my FEC in that case, as a start.

So whichever TTL flash exposure mode you use, for the same situation, you’d just set your FEC to different values … and still get correct exposure in an iterative way by:
- pre-judging the amount of FEC you might need,
- taking the image, and then looking at your camera’s preview,
- and making a visual judgment of how much more or less FEC you would need.

We need to accept a certain flexibility in our technique … as opposed to expecting that by meticulously analyzing how our camera and speedlight work in test situations, that we could get every image perfectly exposed from the very first frame. I believe this one of those areas where there is a danger of over-thinking it and expecting real world situations to fall withing specifically anticipated behavior … when it is much simpler and better, to simply adjust your FEC as you need, on an on-going basis while you shoot.


{ 22 comments. } Add a Comment

1 Brence April 19, 2010 at 10:27 pm

Thanks for the info Neil. When you are pushing the FEC, do you do this on the camera, or on the flash? I am aware that you can adjust this setting in either location. On my 5D I can only push/pull 2 stops on the body, but can adjust by 3 stops no the flash – are these settings both doing the same thing, or are the different? At the moment I tend to just change FEC on the camera body because this setting is quicker to adjust.


2 Neil vN April 19, 2010 at 11:02 pm

Brent, with the 1D mk3 it was ergonomically easier to adjust the FEC on the camera’s body. You also have the full 3-stop range that you can adjust the FEC on the 1D mk3 body as well.

With the 5D, I found that it was awkward adjusting the FEC on the top of the camera. It was like my forefinger just has too much work to do. So I preferred adjusting it on the speedlight itself.

There is no difference in the effect of flash’s output whether you dial it from the camera or the flash itself .. aside from the difference in possible range.

The one thing to take care though, is that the one way of adjusting the FEC, over-rides the other. (I don’t have a 5D on hand right now to double-check which over-rides the other, but it is easy enough to figure out.)

Neil vN


3 Bill April 19, 2010 at 11:11 pm

Hi Neil,

I struggled with this for quite some time with the Canon flash system.

I find your advice to be right on with regards to evaluative and average metering modes. One thing I do (for weddings, especially) is to use the flash exposure lock (FEL – *)quite a bit. It tends to speed the process up quite a bit, and you don’t have to ride the FEC as much.

If I’m shooting a Caucasian skin tone, I find that around +2/3 with exposure lock on the skin tone works pretty well “most” of the time. The FEC reads the center spot on the camera’s viewfinder (regardless of what metering mode you have set) so it’s always taking a spot reading of the center of the viewfinder.

Just thought I’d share some of my findings…keep up the GREAT work!!!


4 Neil vN April 19, 2010 at 11:26 pm

Bill .. thank you for the comment.

I just added an additional observation to the article – that with strong back-lighting, I know I have to dial in a fair amount of FEC already, as a start. Usually at least +1 EV more than I would need in a non-back-lit situation. This is the kind of pre-judging of a scene that comes with experience.

Neil vN


5 Athanasios April 20, 2010 at 7:05 am

Hi from Greece .
I think that it would be very helpful if we could use the same flash exposure lock for more than one photo. We could lock for the skin tones and shoot .


6 Paul Hodgson April 20, 2010 at 11:32 am

Hi Neil

Excellent easy to digest post as usual.

I’m a Nikon shooter and have moved away from Matrix meter now preferring average centre weighted as this combo seem to produce more predictable results – sometimes, Matrix throws a curve ball.

However, I notice the in a darker lit scene the introduction of anything much lighter, no matter the size, seem to greatly influence whichever Nikon metering mode you use.


7 Neil vN April 20, 2010 at 12:07 pm

Hi there John .. it’s the same idea that I wanted to put forth in the article on
Nikon’s TTL vs TTL BL flash settings.

While it is useful to know the general philosophy behind the two metering systems, I don’t think there is that much value to take endless test shots of things and hoping to parse this into very specific settings that you might be able to punch into the flash / camera as always-accurate settings.

I feel it is more useful to be able to predict what your flash is going to do … and then just guide it towards a more accurate FEC setting, depending on the situation you find yourself in.

So it’s like you noticing you would need a different FEC setting if there is a brighter area in the background.

Neil vN


8 Christian April 21, 2010 at 4:11 am

Hi Neil,

Just wanted to make a comment about one method of FEC overiding another method. I’m not sure about the Canons, but on the Nikons, the FEC on the body and the FEC on the flash are cumulative. If you use both they add up together (or subtract depending on which way you are going) e.g +1 on the flash and +1 on the body puts out +2 total. This can really be interesting.


9 Hugh April 21, 2010 at 4:30 pm

Hi Guys,
Christians comment above made me check the 580ex Mk2 Flash manual.
Quote….If FEC has already been set with the Speedlight, FEC cannot be set with the Camera. To set it with the Camera, first set the Speedlights FEC to Zero. p.32

The 430ex on p.14 says the same thing, the Speedlight FEC will overide the Camera.

So that rules out any cumulative effect from both Camera & flash on the Canon, pity.


10 Mark Benavides October 21, 2010 at 12:22 pm

Good morning. I’m STUMPED. I own Canon Mark I, II, III, IV. Mark I is the office workhorse but hotshoe malfunctioned and it’s currently being shipped to Canon.
Mark II and III set to camera defaults, and here’s the problem -
Mark II – if I FEC +3 one would expect washed out pictures. ‘something’ is compensating and histogram for FEC 0-3 look nearly identical.
Mark III – behaves as expected, i.e. FEC +3 extremely washed out and FEC +1 1/3 is pleasing. Oh yeah, MR14-EX ringflash (have 2) set for ETTL.
The long-winded question – what setting is messed up on the Mark II and keeping it from properly (over)exposing?
Thanks in advance,
Ottawa, IL


11 Neil vN October 23, 2010 at 1:57 pm

Mark …check your custom settings. The one camera might be set to Evaluative TTL flash metering, and the other set to Average TTL flash metering.

Neil vN


12 Doug January 31, 2011 at 3:01 am

The main difference between Average and Evaluative is really simple, no snake oil or miracle work needed.

Evaluative mode will set automatically to fill flash when the ambient light is strong enough to allow a correct exposure. This is -3 stops automatically. As you dial in less and less ambient light, the flash increases and eventually stops using FEC to arrive at a fill flash number of 3:1. For instance, if you are in manual mode and set a correct exposure, evaluative flash mode will use the flash at -3FEC as fill on the subject. As you dial down the ambient light, say by using a faster shutter speed, the flash will start to become the main flash, and expose the subject as if it were the only light. In other words, it no longer uses the flash to fill at -3FEC.

Average mode will simply ignore any ambient light and expose for the subject, w/o that inane Canon reasoning that all correctly exposed subjects should be fill flashed at -3EV. So if you have a subject in ambient light, and you want your flash to act like fill, dial in -3FEC and you have the equal of evaluative mode. Now here’s the reason I always use Average Mode: I can take a picture with a correct ambient lit exposure, and have the flash on the camera or bracket act at a power that produces a correct FLASH exposure on the subject, and still keep the background correctly exposed. This works great for backlit people who you want exposed correctly for flash, not filled–which creates a really boring flat light.


13 Cuki February 1, 2011 at 12:12 am

There’s a great video on Canon’s website, the Canon Flash Masterclass:

According to the video, in Evaluative mode the camera tries to figure out where the subject is (probably considering the area with the greatest flash contribution and also the focus information from the lens), then it interprets the preflash metering result accordingly; while in Average mode, the flash exposure is determined by simply averaging the light measured in all areas of the scene.


14 Blonnie August 23, 2011 at 1:22 am

Interesting. I just googled: “flash set from evaluative to average” because I was trying to figure out the mechanics of what was going on differently, and this page was listed -first-.

Every bit of new information I learn just seems to make me realize how very little I know.

Thanks for the helpful post. :)


15 Neil vN August 23, 2011 at 1:29 am

Blonnie .. well, you could’ve asked!

Neil vN


16 Blonnie August 23, 2011 at 10:13 pm

I didn’t want to keep bugging you to explain something in more detail on a busy day. You gave me the means to learn something more on my own and I ended up learning it from you after all! :p



17 Isaac June 26, 2012 at 10:43 am

Hi Neil, i jut bought your new book on off camera flash, so far its great. I have a quick question regarding E-TTL, I have a Canon 40D and a Rebel T2i and i seem to get great pictures when i use A-DEP mode and ettl seems to work fine. Do you use ETTL on M (manual) mode?? I can’t seem to figure it out,it acts totally different on Manual mode.. than other modes. Also i have a quick question The flash unit i have shoots off a red light to measure Distance i guess, what if its being obstructed by lets say a soft box does it throw the metering off? Sorry Im fairly new.


18 Shane July 17, 2012 at 2:06 pm

Hey Isaac. Im not Neil (obviously) but Ill try to help you out. ETTL will work the same in all modes. However, in M mode, the camera is no longer setting the ambient exposure for you. So, depending on how you set your settings in M mode, the flash can vary from the only light source to just a tiny bit of fill.
The red light emmitted from the flash is to assist in focusing in low light situations. The flash being in a softbox will not hurt anything with metering. :)


19 Rudy Winston May 8, 2013 at 9:58 am

Just to clarify on how Canon E-TTL metering works:

The text above isn’t exactly correct. EVALUATIVE E-TTL flash metering means the metering system (same sensor in the prism area that is used to read ambient light) breaks the scene into 35 or 63 zones (more with the EOS-1D X, but the concept is the same). Evaluative metering for E-TTL fires one pre-flash, looks at what areas show a reflection back to the camera, and THOSE metering zones are used for flash calculation. The actual exposure takes place, at this calculated power level, less than 1/10th of a second later.

AVERAGE gives the user an option to have the flash metering take an overall, average reading of that pre-flash, so it’s less likely to run into inconsistencies in situations like fast-paced flash candids at a party or wedding reception. Things like white or light-colored shirts, foreground objects and so on are less likely to require flash exposure compensation to get a good overall exposure.

In either case, ambient light is simultaneously read, but it’s the user’s choice of exposure mode — Program, Shutter-priority, Aperture-priority, or Manual (and the ambient light level) that dictate whether a Canon EOS will blend-in ambient light with E-TTL flash, or try to ignore it and go for an exposure where flash dominates. This is a different approach than Nikon takes with TTL vs. TTL BL settings. In bright light, from direct sunlight down to about the level of a dark cloudy day or deep shade (about EV 10), any of the auto exposure modes (P, Tv, Av) will deliver balanced-fill flash automatically, with ambient light properly exposed, and flash filling-in foreground subjects.

In lower light levels, P-mode starts to act like Nikon’s “TTL” mode — shutter speeds won’t drop below 1/60th second, so backgrounds will tend to be rendered relatively dark, unless you’re working at high ISO levels. But switch to Av or Tv exposure modes, and the camera will *always* try to balance ambient and flash, even in a totally darkened room. Users who like to work in Av mode, but prefer using faster shutter speeds, can call-up a Custom Function to limit slow speeds in Av mode to either the camera’s maximum sync speed, or to range between 1/60 and the max sync speed, depending on ambient light level.

Another difference (at least until the advent of Nikon’s D4, if I’m not mistaken) is that flash exposure compensation in the Canon system has ALWAYS been separate from ambient exposure compensation… you can adjust either one, in either direction in a flash picture, without influencing the other.

Hope this lends a bit of clarity to how the C-brand handles flash exposure, and how ambient lighting is factored in with flash.


20 Pabst Photo May 12, 2014 at 6:35 pm

I have noticed that when using evaluative and bouncing off of a wall – it works fine when I am using a wide (24mm) angle lens. It gets the exposure correct. However, when I zoom that lens to 105mm using evaluative the pictures comes out several stops underexposed. When using Average metering, the pictures all (from 24-105mm) look exposed the same. As such, I find that for run and gun work, like weddings – ‘average’ works better for me. Neil – great blog, great post – you continue to influence my craft.


21 Binh May 15, 2014 at 10:32 am

Great article.

Should we set camera in evaluative so the camera adjust the exposure based on the light of the room and the flash in average mode to adjust the flash light based on the near subject? We want the image count the light in the back for exposure but we don’t want the flash adjust the power for the back and over lightning the subject near by. Thanks


22 Neil vN May 16, 2014 at 11:01 am

Yes, that would be a good approach.

Evaluative TTL metering will tend to pull the flash exposure down when working in low light (when your ambient exposure is under, and you’re hoping the flash will pick up the exposure). So Average TTL metering tends to do better in that scenario.


Leave a Comment

Previous post:

Next post: