7D : metering problem

Started Jan 13, 2011 | Discussions
fpix
fpix Regular Member • Posts: 346
7D : metering problem

Hello,

I do observe following situation with 7D. Av, evaluative metering. For exemplification purpose, let us suppose a scene with rich contrast, say a medium/low toned subject, 10...15% of frame, on light background.

1. When shooting in One Shot camera sets a certain exposure. This exposure would be almost ok, I do observe camera is taking in account focus point and luminance of subject.

2. Same scene, this time in AI Servo AF , continous focus on subject, I get a different exposure, usually -1/3 ... -2/3 compared to situation at 1. But there is always some difference in exposure vs. situation 1. In my thinking this should not happen since the camera knows where the subject is and subject is continously focused.

Has anybody here observed this? Is this a bug?

Thanks for you contribution to finding out what happens.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
Topaz Contributing Member • Posts: 588
Re: 7D : metering problem

fpix wrote:

1. When shooting in One Shot camera sets a certain exposure. This exposure would be almost ok, I do observe camera is taking in account focus point and luminance of subject.

2. Same scene, this time in AI Servo AF , continous focus on subject, I get a different exposure, usually -1/3 ... -2/3 compared to situation at 1. But there is always some difference in exposure vs. situation 1. In my thinking this should not happen since the camera knows where the subject is and subject is continously focused.

Has anybody here observed this? Is this a bug?

Hi fpix. I just tried this experiment on my camera. I could not get the same results as you. On my 7D, I get the exact same exposure in One Shot as AI Servo. Like you, I used Av mode with evaluative metering. The only thing I changed between shots is the AF mode.

I would not expect exposure to change due to the focus mode, and I would be surprised if it did. I wish you good luck in discovering why your 7D is behaving strangely!

Brand New Regular Member • Posts: 256
Re: 7D : metering problem

Unless you're using a tripod to do test shots, I think you're experiencing what I believe: that the evaluative metering in the 7D is extremely biased towards what's under the primary point of metering - almost like it's spot metering. Check out my old thread here: http://forums.dpreview.com/forums/readflat.asp?forum=1019&message=33336970

Your issue might not relate to AI Servo vs. One Shot, but rather that the subject is in a slightly different position in each scenario.

fpix
OP fpix Regular Member • Posts: 346
Re: 7D : metering problem

Thanks, Topaz,

The issue seemes to have something to do with image content. I get this behavior only with, say small subject on plain dark or light backgrounds. But I see something which makes sense below. I will come back.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
fpix
OP fpix Regular Member • Posts: 346
Re: 7D : metering problem

Brand New wrote:

Unless you're using a tripod to do test shots, ...

Makes sense. I'll test again and come back.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
fpix
OP fpix Regular Member • Posts: 346
Samples shot from tripod (Re: 7D : metering problem)

Brand New wrote:

Unless you're using a tripod to do test shots...

So, now 2 samples attached, shot from tripod. Some magnetic sticking picture on my white refrigerator, warm light, AWB (I assume this has no influence anyhow). I would say there is clearly a problem regarding metering in One Shot vs. AI Servo.

Now, I know that under evaluative metering the camera has some complicated and confidential algorithm to determine exposure, that it takes in account the metering zone where the fosus is and the luminance of adiacent metering zones. But what I show here should not happen. Or my camera needs service?

I would like to emphasize (@ Topaz ) that the phenomenon happens only with relatively tiny subjects in contrast with background, background being plain light or plain dark.

I usually use AI Servo AF and focus with AF-ON button. It is very annoying to know matrix exposure in AI Servo AF is failing vs. One Shot AF. In this case the difference is 2/3 stops, but I have seen also 1 stop difference. For a black "Bird In Flight" -1 stop in exposure means throwing the image to the recycle bin...

Can someone replicate this problem? Thanks in advance.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
Brand New Regular Member • Posts: 256
Re: Samples shot from tripod (Re: 7D : metering problem)

Yeah, that shouldn't happen, and that makes me wonder if my inconsistent evaluative metering is something other than what I suspect. I noticed you had firmware 1.2.2. Maybe you can try updating to 1.2.3 and seeing if there was a "silent fix" that was thrown in there?

akin_t Veteran Member • Posts: 3,322
Interesting at best, not really a problem per se.

Interesting at best, not really a problem per se ...

I mean, if you're consistently getting slightly darker exposures in AI servo, I am going to guess that's how the algorithm works. "Problem" or "bug" is when the camera is doing something it's not supposed to do.

You're the one assuming focus mode plays no effect on exposure. What's the basis of your assumption? You're a photographer not a Canon engineer.

I'm not dismissing your claims, it's just you're fretting over nothing. It's an interesting discovery, it really is, but I don't see how it's a problem if you're consistently getting the same results.

As others have suggested, it might have been changed in FW revision. Try updating or contacting Canon about it.

Don't know if my Camera does this, as I haven't had a reason to look for the problem yet, but I'll try it out and post my findings.

Good luck either way.

 akin_t's gear list:akin_t's gear list
Fujifilm X-E1 Fujifilm X-T2 Fujifilm X-E3 Fujifilm XF 18-55mm F2.8-4 R LM OIS Fujifilm XF 55-200mm F3.5-4.8 R LM OIS +4 more
balloonchasers Veteran Member • Posts: 3,189
Re: Interesting at best, not really a problem per se.

Maybe the camera is trying to guess where the subject will be moving in AI Servo Mode (since it is in a constant mode of measuring) so it adds a little more weight to the edges of the focus point where as in One Shot mode it measures from the center of the focus point? See what happens if you use a dark background.....

At the end of the day these meters are only providing the best reading they can and you must determine if the reading and results are correct for your purposes or not. I would certainly not send my camera back for a 2/3 stop difference between two different focusing modes.

John

-- hide signature --

Feel free to use any of these additional letters to correct the spelling of words found in the above post: a-e-t-n-d-i-o-s-m-l-u-y-h-c If you find any extra letters, please place them here for future use...

balloonchasers Veteran Member • Posts: 3,189
Re: Interesting at best, not really a problem per se.

Also since your sample was taken inside I have to ask if you have encountered this problem outside? I think I have read before about varying exposure problems when shooting under fluorescent lights.

John
--

Feel free to use any of these additional letters to correct the spelling of words found in the above post: a-e-t-n-d-i-o-s-m-l-u-y-h-c If you find any extra letters, please place them here for future use...

rodbam Veteran Member • Posts: 9,046
Try the same test

On centre weighted average. I found Evaluative mode on the 7D has a learning curve to it as it doesn't act quite the same as my previous 350D & 40D. I now use centre weighted average & it seems to give me the same results as my older cameras on Evaluative. It will be interesting if in this mode your metering still changes when in servo mode.
One day I must learn the benefits Evaluative mode on the 7D.
Regards Rod

fpix
OP fpix Regular Member • Posts: 346
Re: Samples shot from tripod (Re: 7D : metering problem)

First thanks to all for interrest and feedback!

Now, one by one:

Brand New wrote:

... Maybe you can try updating to 1.2.3 and seeing if there was a "silent fix" that was thrown in there?

Would be happy to be so. I will install it and give a feedback asap.

akin_t wrote:

Interesting at best, not really a problem per se ...

I mean, if you're consistently getting slightly darker exposures in AI servo, I am going to guess that's how the algorithm works. "Problem" or "bug" is when the camera is doing something it's not supposed to do.
...

Actually this was only a sample to show the problem. First I observed the phenomenon outside with white pelicans on dark lake. This was overexposed! Somehow the same tiny subject in contrast to background, background being large and uniform, but now dark instead of light. I will not produce another sample here but I will verify this assumption anyhow and I guess it will happen as supposed, only the other way around - overexposure. More than this, the phenomenon depends on how large the subject is in frame and on the ratio of contrasts. In normal scenes it does not happen. I know to expose manually using the Zone System, but this another issue. For the moment I am trying to find out if this a bug or if my camera needs service.

balloonchasers wrote:

Maybe the camera is trying to guess where the subject will be moving in AI Servo Mode (since it is in a constant mode of measuring) so it adds a little more weight to the edges of the focus point where as in One Shot mode it measures from the center of the focus point? See what happens if you use a dark background.....
...

Could make sense. I will try to imagine an experiment to check if camera assumes AF point could move away. But it would be silly, in the samples showed camera was on tripod, remote released. Dark background - see above.

balloonchasers wrote:

Also since your sample was taken inside I have to ask if you have encountered this problem outside?
...

I have not tested thorrowly outside, but as mentioned above, I observed the phenomenon outside the first time. And with white subject on dark background, with overexposure. This was worse since I obtained burned whites. So, I guess fluorescent light is not an issue. But I will check.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
katman68 Regular Member • Posts: 487
Real world is what matters, i'd suggest go out and shoot that bird...

I think John is maybe on to something there...the metering method may change for servo. The 7D uses very complex methods beyond common understanding unless you're a Canon engineer or want to be that's for sure. I assume that's not why you bought a 7D?(don't get me wrong i've played lab rat too, but....)

What matters is real world, hence my subject line. Just go shoot some birds etc and see what happens? My bets are you'll be very impressed w/the results. I found my original 7D(haven't put new one to the test but likely the same) did a great job w/chasing things across the sky. In fact it track-focused dragonflies like some missile guidance system!!! Very accurate exposure, and gave a brilliant blue sky too, all with no exp compensation.

If anything it can be overly sensitive, no rhyme or reason. Barely discernable frame changes can alter exposure and not always better. I've debated trying out center-weight metering in certain cases for this very reason, like the last commenter said. It may very well stop that. Then again, bird-in-flight it seemed dead on! At least w/digital you can quickly find out exposure & go back to shooting right then...worst case it may need an adjustment but not likely. Let us know how you make out in the blue yonder!

Mark

fpix
OP fpix Regular Member • Posts: 346
Re: Real world is what matters, i'd suggest go out and shoot that bird...

Hi Mark,

I partially do agree with you. Actually I make a lot photos outdoors (eventually check http://www.fpix.eu ). I noticed the phenomenon outside, it annoys me and now I am trying to understand what happens. I just did not put samples with real photos from the field since these would have been difficult to replicate.

Of course, I do not know real exposure algorithms and even if we knew them we could not use these. Not this is the problem. My concern is the following. Aswell in One Shot as in AI Servo the camera sees the SAME scene and KNOWS where the focus is and I expect it should propose the SAME exposure. Unless John has a correct guess of what is happening - this I have to try out somehow.

Further, the problem is not to find a workaround. CW would fine, of course. I prefer partial (yes, partial, not spot) metering in action photos. For example on a bright colored flying bird I work with Tv and partial metering and I dial in +1 ... +1 2/3 exposure compensation, depending on how bright the bird's color is. This works fine. Sometimes the partial metering zone matches better the size of the bird in frame. The partial metering zone is larger than with spot and smaller than with CW.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
fpix
OP fpix Regular Member • Posts: 346
Found : Metering in AI Servo diregards any tracked focus point

So, last night I found the reason.

1) In Ev. Metering / One Shot the camera takes in account the placement of subject in the frame (it knows in which metering zone the focus is aquired) and the tonality of the subject itself and produces an exposure accordingly. A pretty good exposure. This can be observed if you have a light toned object and a dark one in the same scene. By focusing alternatively on the 2 objects the camera will respond with different exposures. This is ok.

2) In same Ev. Metering and One Shot when using manual focus (switch on lens) the camera produces an exposure for the whole scene, without taking in account any particular focus point. This is why there will be a longer exposure if you have a scene with dominant dark tonalities and a shorter one if you have a scene with dominant bright tonalities. Normal.

3) Now in Ev. Metering / AI Servo AF, with some AF point having focus aquired things happen as in 2). That means the camera calculates an exposure as if there would be no subject in the scene. This can be observed in a test like my 2 sample pictures attached yesterday. Thus with the first sample picture if I switch the lens to manual focus I get an exposure exactly as in sample 2 (I get a darker image). And the way around: being in situation of sample 2 if I instantly switch to One Shot I get a brighter image, exactly as in sample 1.

Now my conclusion:

  • This is definitevly not a bug. Since the manual does not state how exposure is calculated in One Shot vs. AI Servo. And for various other reasons.

  • This can be ok in sitaution like fix framing and tracking subject. It would make no sense to change exposure in serial shots while tracking the subject in the SAME scene.

  • But one should be aware tiny subjects on contrasting plain backgrounds might come out overexposed or underexposed. This is quite known anyhow. Alternatives: CW, partial or spot metering, eventual with adequate compensation depending on subject tonality.

  • Quite annoying for people (including myself) who keep the camera always in AI Servo and focus with separate AF-ON. AF-ON pressed --> AI Servo. AF-ON depressed --> One Shot. But in second case you get a different exposure. For static subject this is worse than using standard One Shot mode. So better switch to One Shot. C.Fn IV-1 allows you to aasign the DoF button for instantly switching One Shot AI Servo.

Maybe quite complicated for the normal photographer, but good to know for people trying to constantly reach best possible exposure.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
Topaz Contributing Member • Posts: 588
Re: Found : Metering in AI Servo diregards any tracked focus point

fpix, nice work. I think this is an interesting conclusion. Since I very much like the AF-linked behavior of evaluative metering, it's good to know this does not work during AI Servo. I never realized that before.

This made me ask another question: for shooters who use back-button focus in One Shot focus mode, does evaluative metering link to the AF point? After all, the focus was obtained before the shutter button was pressed and there is no focus lock, per se, at the moment the shutter is pressed. I wondered: perhaps back-button focus with One Shot also behaves like manual focus and AI Servo (i.e., no AF point linking to evaluative metering).

So I tested this. My conclusion: with back button focus in One Shot mode, the evaluative metering is, indeed, still linked to the AF point. First of all, it achieves identical exposure to the case where I focus with the shutter button. Also, even after locking focus, if you simply select different AF points while the viewfinder is still live, you can see the metering change accordingly.

So I agree with your 3 points below. And I would add that for point #1, it's true regardless of which button you use to start AF and regardless of how much time elapses between AF lock acquisition and shutter activation.

I also agree that this is important info for those who use the technique of tapping AF-ON in AI Servo mode to "simulate" a one-shot focus. They should realize they are losing the benefit of AF-point-linked evaluative metering.

fpix wrote:

So, last night I found the reason.

1) In Ev. Metering / One Shot the camera takes in account the placement of subject in the frame (it knows in which metering zone the focus is aquired) and the tonality of the subject itself and produces an exposure accordingly. A pretty good exposure. This can be observed if you have a light toned object and a dark one in the same scene. By focusing alternatively on the 2 objects the camera will respond with different exposures. This is ok.

2) In same Ev. Metering and One Shot when using manual focus (switch on lens) the camera produces an exposure for the whole scene, without taking in account any particular focus point. This is why there will be a longer exposure if you have a scene with dominant dark tonalities and a shorter one if you have a scene with dominant bright tonalities. Normal.

3) Now in Ev. Metering / AI Servo AF, with some AF point having focus aquired things happen as in 2). That means the camera calculates an exposure as if there would be no subject in the scene. This can be observed in a test like my 2 sample pictures attached yesterday. Thus with the first sample picture if I switch the lens to manual focus I get an exposure exactly as in sample 2 (I get a darker image). And the way around: being in situation of sample 2 if I instantly switch to One Shot I get a brighter image, exactly as in sample 1.

Now my conclusion:

  • This is definitevly not a bug. Since the manual does not state how exposure is calculated in One Shot vs. AI Servo. And for various other reasons.

  • This can be ok in sitaution like fix framing and tracking subject. It would make no sense to change exposure in serial shots while tracking the subject in the SAME scene.

  • But one should be aware tiny subjects on contrasting plain backgrounds might come out overexposed or underexposed. This is quite known anyhow. Alternatives: CW, partial or spot metering, eventual with adequate compensation depending on subject tonality.

  • Quite annoying for people (including myself) who keep the camera always in AI Servo and focus with separate AF-ON. AF-ON pressed --> AI Servo. AF-ON depressed --> One Shot. But in second case you get a different exposure. For static subject this is worse than using standard One Shot mode. So better switch to One Shot. C.Fn IV-1 allows you to aasign the DoF button for instantly switching One Shot AI Servo.

Maybe quite complicated for the normal photographer, but good to know for people trying to constantly reach best possible exposure.

Keith Reeder Veteran Member • Posts: 3,725
Good man, Fpix...

Nice to see someone actually evaluate and test an issue thoroughly and come to an informed conclusion rather simply insisting (often supported by contributors who have no clue and automatically assume everything is a camera fault) that you've got a broken camera.

Excellent analysis - and I'm glad to have learned something very useful about the camera here.

Keith Reeder Veteran Member • Posts: 3,725
This is useful

http://usa.canon.com/dlc/controller?act=GetArticleAct&articleID=2547

I guess this is why the AE Lock and AF-On buttons are right next to each other on the back of the camera...

fpix
OP fpix Regular Member • Posts: 346
Re: Found : Metering in AI Servo diregards any tracked focus point

Thanks for appreciation, I am glad this may be useful.

As I said, I was quite sad to find this behavoir. I previously had 40D and was very unhappy with it's inconsistent evaluative metering, which brought me to learn the Zone System and use M/spot whenever possible. I was glad to observe that 7D has a much better evaluative metering in One Shot and automatically believed the same would apply in AI Servo, continuing to "emulate" One Shot via AF-ON released, camera being always in AI Servo, as usual.

In AI Servo the camera actually knows where the focus is, the algorithm is there already (used in One Shot) so there should be at least a Custom Function to allow evaluative metering linked to AF point as in One Shot. Would be nice to see it in a firmware upgrade, but I have no clue how we could bring this piece of information to Canon. I assume it is quite impossible. And I believe only few photographers are actually affected, most would not go in such details anyhow.

I believe I will completely abandon matrix metering. I'll stick to partial or CW with adequate AEC for action photo or M/spot (as in Zone System) whenever possible.

Just to mention:

I tested a D300 of a friend today. Not very thorrowly, but it seems matrix metering works the same (AF linked) both in Single Shot as in Continous Focusing mode...

Topaz wrote:

...

This made me ask another question: for shooters who use back-button focus in One Shot focus mode, does evaluative metering link to the AF point?

Sorry, what do you mean by back-button focus ? AF-ON or AE-Lock?

I guess this is why the AE Lock and AF-On buttons are right next to each other on the back of the camera...

Sorry again. I do not understand what you mean. Could you detail? Anyhow, I use AE-Lock for getting an additional focus point ("registered AF point"). AE Lock function is not worth a separate button. It is already implemented in One Shot, as explained in the link you kindy indicated.

-- hide signature --

Cristi@fpix.ro

 fpix's gear list:fpix's gear list
Canon EF 135mm F2L USM Canon EF 16-35mm F4L IS USM Canon EF 100-400mm F4.5-5.6L IS II Canon EOS 5D Mark IV Canon EF 85mm F1.8 USM +2 more
Topaz Contributing Member • Posts: 588
Re: Found : Metering in AI Servo diregards any tracked focus point

fpix wrote:

Topaz wrote:

...

This made me ask another question: for shooters who use back-button focus in One Shot focus mode, does evaluative metering link to the AF point?

Sorry, what do you mean by back-button focus ? AF-ON or AE-Lock?

Either one, since the 7D lets you customize both those buttons. I was referring to the common technique of disabling AF-start on the shutter button, and using any of the back buttons to start AF at a separate time. Anyway, what I was trying to point out was that I answered my own question, and the answer was "yes, it still links".

I guess this is why the AE Lock and AF-On buttons are right next to each other on the back of the camera...

Sorry again. I do not understand what you mean.

That comment was from another poster, but actually I do not get the point either.

Keyboard shortcuts:
FForum MMy threads