ETTL-II Evaluative vs Average Camera selection - What effects on PIC ?

My Camera User's Manual explains that the difference between 'Evaluative' & 'Average' ETTL-II mode is FEC (which is disabled in 'Average').

But, there's no other elaboration on the difference between the 2 ETTL-II modes.

Is it that 'Evaluative' gives preference to the illuminated subject, while 'Average' gives more emphasys to the shadows in the background ?

Try this https://photography-on-the.net/forum/showthre​ad.php?t=51698

roli_bark wrote in post #2806640
My Camera User's Manual explains that the difference between 'Evaluative' & 'Average' ETTL-II mode is FEC (which is disabled in 'Average').

The manual of my 20D, in the description of the "Average" setting under custom function 14, says:

As is typical of Canon instruction manuals, the term "automatic flash exposure compensation" is not found anywhere else in the book. I suspect it's a reference to automatic fill flash reduction, which is mentioned in Canon's Flashwork and other documentation. Flash exposure compensation (manually controlled) works with both Evaluative and Average flash metering.

I have not conducted systematic experiments on the differences between Evaluative and Average flash metering. Hopefully Leo will fix the links in the thread referenced above. Based on my own anecdotal experience, I have come to the following conclusions:

1) Average metering is not as susceptible to small, highly reflective areas in the frame which tend to cause inconsistencies with E-TTL. When the flash reflects off glass or polished metal surfaces and back through the lens, it can cause underexposure problems. These seem to be worse with Evaluative metering. I usually select Average metering indoors for this reason.

2) Average metering does not seem to account for the ambient exposure and reduce the flash output accordingly, at least not as well as Evaluative. Using Average metering outdoors without dialing in a serious amount of negative FEC tends to create overexposed subjects. For this reason I use Evaluative metering in outdoor fill flash situations.

3) Since Average metering uses the entire frame, it will tend to overexpose if the subject only fills a small portion of the frame and the background is distant (such as a person in a large room).

Again, this is based on my own experience and does not come from any official Canon documentation.

DrJack wrote in post #2806853
Try this https://photography-on-the.net/forum/showthre​ad.php?t=51698

Been there ... sample Pics [which demonstrate the difference] are missing ...

Curtis N wrote in post #2807081
The manual of my 20D, in the description of the "Average" setting under custom function 14, says:As is typical of Canon instruction manuals, the term "automatic flash exposure compensation" is not found anywhere else in the book. I suspect it's a reference to automatic fill flash reduction, which is mentioned in Canon's Flashwork and other documentation. Flash exposure compensation (manually controlled) works with both Evaluative and Average flash metering.

Thanks Curtis. My bad - I stand corrected...

Curtis N wrote in post #2807081
1) Average metering is not as susceptible to small, highly reflective areas in the frame which tend to cause inconsistencies with E-TTL. When the flash reflects off glass or polished metal surfaces and back through the lens, it can cause underexposure problems. These seem to be worse with Evaluative metering. I usually select Average metering indoors for this reason.

2) Average metering does not seem to account for the ambient exposure and reduce the flash output accordingly, at least not as well as Evaluative. Using Average metering outdoors without dialing in a serious amount of negative FEC tends to create overexposed subjects. For this reason I use Evaluative metering in outdoor fill flash situations.

3) Since Average metering uses the entire frame, it will tend to overexpose if the subject only fills a small portion of the frame and the background is distant (such as a person in a large room).

Thanks again - very helpfull.

roli_bark wrote in post #2806640
Is it that 'Evaluative' gives preference to the illuminated subject, while 'Average' gives more emphasys to the shadows in the background ?

Yes, that's basically it. Evaluative flash metering attempts to identify the subject and to expose the subject at midtone (plus or minus FEC), while Average flash metering attempts to bring the entire scene up to midtone (plus or minus FEC).

Average flash metering gives results similar to simpler flash metering systems: TTL, thyristor autoflash, and point-and-shoot. Scenes are brightly lit, and when shooting in dim light the subject is overexposed and the background is illuminated. In dim light with no reasonably close background, the flash pours out enough light that the subject glows in the dark. This results in serious blowout.

Evaluative flash metering—when the subject has been properly identified by the camera and when FEC has been properly set by the photographer—gives a correctly exposed subject. In dim light, the background will be much darker than with Average.

Many (most?) photographers are so accustomed to the overcooked look from simpler flash systems that they consider Evaluative metering to give "underexposed" pictures in dim light. I'm reminded of this story:

ESS was a speaker company back when I was a kid that allegedly embarked on a grand scientific survey to determine what kind of sound most people really liked. ESS researchers devised preference tests and queried large numbers of test subjects; the project was trumpeted in a series of ads. But as it advanced, things got awfully quiet, and finally the project was abandoned. The problem, it turned out, was that the sound all those Everyman test subjects were remorselessly vectoring in on was the sound of a cheap car stereo, which is what the greatest number of said Everymen were the most familiar with already.

Also, in a properly exposed non-fill flash photo the subject will be the brightest object in the scene. Everything else will be darker. This means that the histogram is almost entirely to the left of center, with a huge spike at the left end for all of the black. This is normal, this is desirable, this is not underexposure.

Doug Pardee wrote in post #2807217
Also, in a properly exposed non-fill flash photo the subject will be the brightest object in the scene. Everything else will be darker. This means that the histogram is almost entirely to the left of center, with a huge spike at the left end for all of the black. This is normal, this is desirable, this is not underexposure.

How do you define: "non-fill flash photo" ?
[obviously there's not a "fill flash" mode pre-ce nor in the Camera neither in the Flash ...]

roli_bark wrote in post #2807280
How do you define: "non-fill flash photo" ?
[obviously there's not a "fill flash" mode pre-ce nor in the Camera neither in the Flash ...]

Where the flash gun provides the primary source of light (to some extent) for foreground subject illumination. Commonly, using Auto, Program and smaller apertures/faster shutters in Manual under poor lighting often results in this (due to the insufficient ambience light present to properly expose background, but foreground subjects are exposed decently due to the flash gun - hence Doug Pardee's assertion on the [foreground] subject being the brightest object in the scene, with everything else [In middle/background] to be darker).

Doug, I concur with most of your post, except this part:

Doug Pardee wrote in post #2807217
Many (most?) photographers are so accustomed to the overcooked look from simpler flash systems that they consider Evaluative metering to give "underexposed" pictures in dim light.

Even with a close background, or the entire image at more or less the same distance from the lens, Canon's various automatic metering systems will expose it more brightly with ambient light than with flash as the primary light source. This is true with both Evaluative and Average flash metering. +2/3 or +1 FEC is required to make the exposures identical when you capture the same scene with ambient light vs. flash.

Curtis N wrote in post #2807375
Doug, I concur with most of your post, except this part:Even with a close background, or the entire image at more or less the same distance from the lens, Canon's various automatic metering systems will expose it more brightly with ambient light than with flash as the primary light source. This is true with both Evaluative and Average flash metering. +2/3 or +1 FEC is required to make the exposures identical when you capture the same scene with ambient light vs. flash.

What do you mean by "with a close background" ?
Do you have an explanation as to why these circumstances cause [an under exposed image w Flash illumination] ?

I included the phrase, "Even with a close background, or the entire image at more or less the same distance from the lens," because otherwise it's generally difficult for flash to match the ambient exposure of an entire scene. If the background is distant, the flash will always illuminate the subject more than the background, and any comparison between ambient-lit and flash-lit exposures of the same scene would be dubious at best.

I believe, as many people do, that Canon's calibration of its E-TTL metering is flawed, and these perceptions are not merely caused by improper expectations as Doug suggested.

Curtis N wrote in post #2807375
Even with a close background, or the entire image at more or less the same distance from the lens, Canon's various automatic metering systems will expose it more brightly with ambient light than with flash as the primary light source.

I don't disagree for those situations when there is significant ambient light, but my statement specifically said "in dim light". My intention was "when ambient lighting is insignificant". Maybe I should have put it that way.

roli_bark wrote in post #2807280
How do you define: "non-fill flash photo" ?

What Collin said. A "fill flash photo" is one where you could capture the scene with the same settings without flash and still get a good exposure; the flash is only being used to lighten shadows. A "non-fill flash photo" is one that would be unacceptably dark, probably black, if the flash wasn't used.

The Av mode is essentially fill flash, as is Night Portrait. Both modes will expose as if the flash wasn't going to be used—often resulting in very long shutter times—and then use the flash for fill-in.

The other modes can be fill flash if the camera is able to achieve proper exposure without flash. The most common example of this is outside in bright daylight, when the sun is casting harsh shadows on someone's face and you use flash to keep their eye sockets from being black holes.

Evaluative E-TTL flash is "fill flash" by definition. It accounts for the ambient light on what it thinks is the subject when calculating the required flash output.

The only way to make E-TTL flash the primary light source is to force the camera to underexpose the ambient. Usually the best way to do this is with Manual mode. It can also be done in P mode (which limits the shutter speed to 1/60). Or you can use Av mode with custom function 3-1 which forces the shutter speed to flash sync.