Evaluative vs. Average flash metering

[sambahri]

i got my rebel XT 350D . i know the difference between evaluative and partial flash metering. but my question is in which condition of flash phtography to use each of them. by the way i got a distance information lens as well as a 420EX flash. pls i need a pro answer in detail..thanks

[tim]

Maybe you could explain partial flash metering to me, and how you achieve it on your 350D/420EX.

[PacAce]

Originally Posted by sambahri

i got my rebel XT 350D . i know the difference between evaluative and partial flash metering. but my question is in which condition of flash phtography to use each of them. by the way i got a distance information lens as well as a 420EX flash. pls i need a pro answer in detail..thanks

By partial flash metering, I assume you mean using FEL to lock the exposure of the area covered by the partial metering area. You would use the normal evaluative metering for most situations where the subject would have a normal mid-tone reflectance characteristic, neither too light nor too dark. If that is not the case, then locking the exposure on something more mid-toned by placing the partial metering area over it and pressing the FEL button would give you a better flash exposure. Obviously, the object your are trying to FEL should be about the same distance as the subject you are trying to photograph.

[sambahri]

sorry i meant avearage flash metering

[roli_bark]

Appologies for the Bump - Just to get the correct answer ... PacAce ?

[Curtis N]

Evaluative E-TTL is the default mode. It tends to be easily confused by white or bright elements such as white shirts or reflections from polished metal surfaces. For indoor use, most people find that Average mode yields more consistent exposure, as long as the background is relatively close.

But Average E-TTL mode is looking for light throughout the frame. So if you don't have a fairly close background to reflect light from the pre-flash (such as outdoors or in a large room like a church), the metering system will only receive light reflecting off the subject and therefore tend to overexpose unless your subject fills most of the frame.

So, my usual recommendation is Average mode for most indoor situations and Evaluative mode for outdoor fill.

[PacAce]

Originally Posted by Curtis N

Evaluative E-TTL is the default mode. It tends to be easily confused by white or bright elements such as white shirts or reflections from polished metal surfaces. For indoor use, most people find that Average mode yields more consistent exposure, as long as the background is relatively close.

But Average E-TTL mode is looking for light throughout the frame. So if you don't have a fairly close background to reflect light from the pre-flash (such as outdoors or in a large room like a church), the metering system will only receive light reflecting off the subject and therefore tend to overexpose unless your subject fills most of the frame.

So, my usual recommendation is Average mode for most indoor situations and Evaluative mode for outdoor fill.

As much as I hate to do it, I'll have to sort of disagree with Curtis' take on Evaluative vs. Average. And here's why. Curtis says that indoors where the background is relatively close to the subject, you will get more consistent flash exposure in Average mode than you would in Evaluative mode. That may be true but does that mean that the exposure will be correct? Most homes that I've been to have light colored walls. That being the case, if the wall were considered in the metering of the exposure, wouldn't that tend to underexpose the shot since the wall is light in tone? Yes, it would. Check out Exhibit #1.

Exhibit #1
Exhibit-1.jpg


Curtis also claims that ETTL-II is easily confused by reflections and shiny surfaces? But the way ETTL-II is supposed to work, that shouldn't be the case. AAMOF, from my experience, the opposite is usually true. It's Average metering that easily gets confused by reflections from mirrors or glass panels. See Exhibit #2.

Exhibit #2
Exhibit-2.jpg

[PacAce]

While I agree that with Evaluative metering, a white-clothed subject will tend underexpose the white clothing, how often do you see someone dressed all in white unless the subject happens to be a bride. And in that case, you can also add a little bit of +FEC to compensate. BTW, did you notice that Average metering underexposes in this situation, too, although not as much as Evaluative?

But on the flip side, if the subject is in all black, Evaluative will tend to expose better than Average would.

Exhibit #3
Exhibit-3.jpg


And just for funzies, I put a black background to simulate taking a flash picture outside or in a very large hall or auditorium. As you can see, Curtis was right about the results one could expect in this environment.

Exhibit #4
Exhibit-4.jpg


Needless to say, I'm a great fan of ETTL-II. A lot of people complain about it but most of the results I get from using ETTL-II suit my taste just fine. Of course, ETTL-II isn't perfect but neither are any of the flash algorithms out there, including Average metering and its very close cousin, the self-contained Auto flash.

[PacAce]

Originally Posted by roli_bark

Appologies for the Bump - Just to get the correct answer ... PacAce ?

Unfortunately, there is no right or wrong answer here. It boils down to which you like better and can adapt to more easily. Both have their pros and cons.

Both metering modes require some knowledge of what's going on with the exposure so that you can apply the appropriate exposure corrections. Or just good old chimping helps a lot, too. If all one wants to do is slap on a flash, turn it on and shoot away, one is going to be very disappoints with a lot of the results one gets no matter what mode is used.

[monty28428]

Wonderful examples Leo!! Thanks for sharing this.

[johnstoy]

Well live and learn... There is a lot of versatility with the ETTL-II...I've my sights set on a new 580EX II... Great info...

[Curtis N]

Hey Leo, that model looks familiar!

Originally Posted by PacAce

It boils down to which you like better and can adapt to more easily. Both have their pros and cons... Both metering modes require some knowledge of what's going on with the exposure so that you can apply the appropriate exposure corrections.

This I certainly agree with. I stated that Average mode is more consistent, but perhaps I should have used the word "predictable." You always need to think about what the camera "sees" and adjust FEC accordingly. Indoors, I find that it's easier to predict how Average E-TTL will meter a scene and I can guess right more often.

As for reflections from glass, I have never been able to get the kind of shot you got in the lower left of exhibit #2, in any metering mode. That scenario always leads to gross underexposure in my experience. And I think we'll agree that shooting flash directly in front of glass is something to avoid anyway, and most flash shooters learn to do that.

Likewise, a scene with a lot of white, such as a full length bridal or a table shot with a white tablecloth, will need +FEC in any metering mode.

The scenarios that tend to make Evaluative E-TTL frustrate me are the ones that are harder to predict. Like a six-person group photo where only one person is wearing a white shirt. In my experience, Evaluative E-TTL will expose that white shirt to medium gray every time. Reflections from metallic surfaces such as the aluminum frame on a chalkboard also tend to throw things off. Wedding shooters also complain about similar effects from satin fabrics in bridesmaid's dresses, and it varies, depending on how much of the fabric happens to be angled just right to reflect light back into the lens. I'm talking about small areas creating reflections much more subtle than what you get shooting into a window, and it's really hard to see these things coming.

But with Average E-TTL, as long as the background is close, I can look at the relative reflectance of the scene and adjust FEC, just like I would adjust EC for an ambient light shot. And without a complex algorithm "evaluating" the scene and making assumptions, small reflective areas will not affect the metering as dramatically.

[roli_bark]

Still, some things confuse me here: In some other threads some of you mentioned two basic differences between Average & Evaluative:

1. That 'Average' tends to be more accurate exposure, when the Flash is the dominant [or almost dominant] light source on Scene.

2. That 'Evaluative' tends to meter light 'Subject' based. That means that Evaluative has algorithms to differ between the illuminations of subject and background.

How these two factors sit with the findings you've both have described ? Especially (1) with Leo's Exhibit #4 ?

Another note:
I think this thread (from 2004 !) - How ETTLII Works - by Scott Berger ('scottbergerphoto') sums up the ETTL-II Flash Evaluative mode quite accurately:

http://photography-on-the.net/forum/...t.php?p=319540

"... The ambient and pre-flash readings are compared. The metering areas having a small difference are selected as the flash exposure metering areas. (Areas with very big differences between ambient and pre-flash readings are excluded or down weighted because they are assumed to contain a highly reflective object or that the subject is not in that part of the frame. The algorithm avoids chronic underexposure problems in such situations.) These readings are weighted, averaged and compared with the ambient light reading, and the main flash output is then set and stored in memory. E-TTL II weights and averages the flash metering for the subject and all other objects at the same distance as the subject. Even if the subject's position, reflectance or size changes, the flash output will not change radically. The flash exposure will be highly accurate and stable..."

Finall note:
A detailed explanation for the difference between Flash Average & Flash Evaluative is due. Uptodate, I couldn't find ANY official elaborate reference to this from Canon !.

Thanks Leo, Curtis, Scott Berger - I see you'all as the de-facto Flash gurus here [along with few others].


Moderator(s):
A. Can title be changed from 'evaluative& partial flash metering' to 'Evaluative & Average Flash metering - Differences'.
B. Can this become a STICKY ?

[PacAce]

Originally Posted by roli_bark

Still, some things confuse me here: In some other threads some of you mentioned two basic differences between Average & Evaluative:

1. That 'Average' tends to be more accurate exposure, when the Flash is the dominant [or almost dominant] light source on Scene.

2. That 'Evaluative' tends to meter light 'Subject' based. That means that Evaluative has algorithms to differ between the illuminations of subject and background.

Re #1, I, for one, never said anything to that effect. AAMOF, I have usually tended to shy away from discussions about Average vs Evaluative because I personally never felt that it was a cut and dry "use this or use that" type of a discussion. Because of the algorithm used in Evaluative mode, I think that it would produce more accurate exposures more often than Average would. But I'm sure there are others who won't agree with me.

Re #2, Evaluative does bias exposure towards the subject. But no matter how the exposure is determined, there can only be one exposure setting. There is no way for the flash to determine correct exposure for both the subject and the background no matter how complex the algorithm used. The best it can do is try to expose the subject properly and let the exposure for the background fall where they fall.

Originally Posted by roli_bark

How these two factors sit with the findings you've both have described ? Especially (1) with Leo's Exhibit #4 ?

Exhibit #4 is a perfect example of #2 above. It also perfectly demonstrates the flaw of statement #1.

Originally Posted by roli_bark

Another note:
I think this thread (from 2004 !) - How ETTLII Works - by Scott Berger ('scottbergerphoto') sums up the ETTL-II Flash Evaluative mode quite accurately:

http://photography-on-the.net/forum/...t.php?p=319540

"... The ambient and pre-flash readings are compared. The metering areas having a small difference are selected as the flash exposure metering areas. (Areas with very big differences between ambient and pre-flash readings are excluded or down weighted because they are assumed to contain a highly reflective object or that the subject is not in that part of the frame. The algorithm avoids chronic underexposure problems in such situations.) These readings are weighted, averaged and compared with the ambient light reading, and the main flash output is then set and stored in memory. E-TTL II weights and averages the flash metering for the subject and all other objects at the same distance as the subject. Even if the subject's position, reflectance or size changes, the flash output will not change radically. The flash exposure will be highly accurate and stable..."

I think my examples show that ETTL-II is working as advertised for the most part. Some may question the "accurate and stable" part, though. I have seen inconsistencies between exposures in Evaluative mode but I've also seen them when using Average mode. What I haven't determined is whether the inconsistencies are caused by ETTL itself or the hardware, i.e. the flash unit. In either case, the difference in exposure, for me, isn't much to be concerned about.

Originally Posted by roli_bark

Finall note:
A detailed explanation for the difference between Flash Average & Flash Evaluative is due. Uptodate, I couldn't find ANY official elaborate reference to this from Canon !.

Thanks Leo, Curtis, Scott Berger - I see you'all as the de-facto Flash gurus here [along with few others].


Moderator(s):
A. Can title be changed from 'evaluative& partial flash metering' to 'Evaluative & Average Flash metering - Differences'.
B. Can this become a STICKY ?

Re A, consider it done.
Re B, I'll include this in the FAQ thread which is stickied at the top.

[roli_bark]

Thanks Leo. one more thing:
In many of your above references to ETTL-II I'm sure you meant "Evaluative" ... Please fix.