EOS 5D Mk III metering issue - underexposure

[janko]

Hi evrybody. I have been shooting with EOS 5D Mk III since May 2014 before I embarked on 5-week family trip  on the West Coast during July and August. I shot many and many pictures in all sorts of conditions. My major disappointment with the camera is its metering. At least my version  underexposes terribly. I read some discussion threads here and elsewhere, cleared whatever settings there may had been left from the factory but the problem is still there. Sometimes +1 exposure compensation does not help since in LR I need to push it another stop to get it right. When somebody says that Canon is little bit conservative in its in-camera meters I cannot agree less. I had a Nikon D80 before and at least there was no any metering issue. If  Nikon got it right on their cheap camera I cannot see why Canon should be behind. I also read about people sending their 5d mk iii to Canon servic and hed it re-calibrated.

Thanks for any suggestions but I am little bit tired of  tweaking the camera everytime i take a picture.

Jan (lives in Switzerland)

[Skirball]

I thought I remember there being quite a few discussions about severe underexposing with both the 5d3 and 6D when they came out.  I think some of it was just differences in expectation (Canon is a little bit conservative with metering, but not two stops), but some of them were actually sent back for service and fixed.  Two stops is huge, that shouldn't be happening.

If you're confident it's not user error, I'd send it in to get checked out.  THe 5d3 is one of the most popular pro/enthusiast cameras out there, I find it extremely unlikely this is normal and tollerated behavior.

I do bump up my exposure comp 1/3 always, and sometimes 2/3, when I shoot in Av with my 6D, but I rarely use automatic modes so seldom I don't fuss about it much.  I'm always checking my histogram and If I'm down a bit I move it up.

[RobertTheFat]

---

@janko wrote:

Hi evrybody. I have been shooting with EOS 5D Mk III since May 2014 before I embarked on 5-week family trip  on the West Coast during July and August. I shot many and many pictures in all sorts of conditions. My major disappointment with the camera is its metering. At least my version  underexposes terribly. I read some discussion threads here and elsewhere, cleared whatever settings there may had been left from the factory but the problem is still there. Sometimes +1 exposure compensation does not help since in LR I need to push it another stop to get it right. When somebody says that Canon is little bit conservative in its in-camera meters I cannot agree less. I had a Nikon D80 before and at least there was no any metering issue. If  Nikon got it right on their cheap camera I cannot see why Canon should be behind. I also read about people sending their 5d mk iii to Canon servic and hed it re-calibrated.

Thanks for any suggestions but I am little bit tired of  tweaking the camera everytime i take a picture.

Jan (lives in Switzerland)

---

Assuming you've checked to make sure no fundamental settings are amiss, I think I'd have it looked at. My 5D3 doesn't do that; it's exposures are commendably accurate in most circumstances. Like you, I've had mine since last May, and I don't think I've ever had to use more than a half stop of correction in post-processing. If the problem you're seeing were common, this forum would be alive with complaints. (Like with the focus problems on the 70D.)

[TCampbell]

I have a 5D III (and a 5D II), but I also have a couple of hand-held light meters (incident meters).  

When I compare the camera metering to the incident meter, I find the camera's metering is extremely accurate (Canon has been building these things for a while now... it's unlikely that they'd produce a camera that metered so far off).  

Keep in mind that when you use an "incident" meter, several things can throw the metering.  

(a) are you using full 'evaluative' metering, or have you switched to an alternate metering area? (e.g. spot metering, etc.)?

(b) have you dialed in exposure compensation without realizing it? (your EXIF data will tell you if you dialed in exposure compensation).

(c) are you metering subjects that are known to throw metering? (e.g. metering very dark things or very white things will throw the accuracy of the metering.  That's normal for in-camera meters because of the nature of "reflected" light vs. "incident" light.)

I'm wondering if you can post or link a sample (with EXIF data in the image -- very important).  Occasionally someone will review the image on the camera LCD screen... but that has it's own brightness settings and can trick you into thinking an image is either too bright or too dim.

[janko]

Hi Tim

Your interest and help is much appreciated. I will upload a picture and exif data. On our 5 weeks trip we have taken pictures in many situations including those that can trick in-camera meter.  I will come back with some samples for people to see. I cannot completely exclude  human error (photographer's lack of experience).

Best regards,

Jan

[TCampbell]

I just happened to catch a tutorial on time-lapse photography where the entire tutorial was based on the problem of shooting time-lapse when the time transitions from day-to-night or night-to-day (possibly the most difficult to do).  

Anyway, the photographer who does this says he used to have all 5D II bodies, but he started switching all his bodies to the 5D III... he made comments to how much better and more accurate the metering is on the 5D III.

Hopefully we can figure out what's happening to your underexposed images.

[janko]

Dear Tim,

Thank you for your interest to see my pictures. I tried to choose those where I see  underexposure with no exposure compensation. I uploaded them to my gallery (as poublic) for you to see. I was not sure with the exif data so I am sending them right away through e-mail.

With the best regards,

Jan 

[oldsurfer]

Reading the EXIF date with your examples, they all say Metering Method - Pattern.  What camera setting is that?  What metering mode do you have selected on the camera..  Most of my shots are set to Evaluative metering and are looking good.  Like the others have said if it's off consistantly maybe a trip to Canon Repair is in order.

[TCampbell]

The embedded EXIF data was stripped out of the images but I see you posted screenshots of it and I matched up the images names to the EXIF data.

I saw three images that appear to be underexposed... except I'm looking at the EXIF data and it tells me the following:

Image of the mountain scene (mix of sun & heavy shade)

  ISO 100

  f/10

  1/200th sec

Image of the RV (full sun)

  ISO 100

  f/8

  1/250th sec.

Image of group in front of tree (medium overcast)

  ISO 100

  f/8

  1/250th sec.

I'm wondering if you have any filters on your camera lens.  Do you, by any chance, have a polarizing filter?

If you do have a polarizer, do you know if it is a standard "linear polarizer" (sometimes called a "top polarizer") vs. a "circular polarizer"?

Here's why I'm wondering...

The RV shot was the most telling.  I can clearly see by looking at the RV and the ground, that the RV was in "full sun" when the shot was taken and based on the angles of the shadows I can even tell that this was taken near mid-day.  Mid-day full sunlight is _always_ meters out at about Ev 15 and can follow something called the "Sunny 16" rule.  We can take this shot all day in manual mode and never even glance at the light meter.  At f/16 (you were at f/10 ... more on that in a moment) the shutter speed can be set to the inverse of the ISO.  In other words at f/16 with ISO 100 you can use 1/100th.  At ISO 200 you can use 1/200th.  At ISO 400 you can use 1/400th.  These will nail the exposure (assuming no filters.)

You used Program mode and your camera used f/10 and 1/200th.  f/10 is 1.3 stops more light than f/16.  To compensate, the shutter speed would need to increase by 1.3 stops faster (to reduce light collection) so 1/250th would have been about perfect (assuming no filter).

You have something similar going on in the mountain scene... but this is mixed lighting.  The sky is partly cloud but the areas covered by clouds show the clouds are dense.  So you've got a mix of full sun + strong shade.  

The group photo in front of the tree is in medium overcast (light overcast still casts a shadow, but a weak one.  Medium overcast does not cast a noticeable shadow).  That means we'd open up by 2 stops to get this.  Your camera opened the aperture by 2 stops (good) but then reduced the shutter exposure by 1.3 stops (bad) giving you 2/3rds of a stop above a "Sunny 16" rule exposure, but you needed 2 full stops (it's 1.3 stops under-exposed.)

If you use a "linear" polarizer (or "top" polarizer) this will confuse the camera's metering system.  A "circular polarizer" should be used with any DSLR when a polarizer is desired.  A "circular polarizer" really is just a linear polarizer except it has an extra layer behind the polarizing layer which has a quarter-wave plate to re-orient the light after it passed through the polarizer.  This tweak allows the auto-focus and metering to work more accurately.

Probably less important, but I may as well ask what lens you are using and do you have this issue with any other lens (e.g. just in case the aperture blades are contstricting to an incorrect aperture opening size.)

Basically I'm wonder this because at the RV SHOULD HAVE BEEN fractionally over-exposed based on the exposure settings... and assuming no filter on the lens... and yet it's not.  The camera called for an exposure which was very close to a correct exposure and yet you ended up with an underexposed shot.  This may be caused by a lens filter or the lens itself.

[janko]

Dear Tim,

Thank you very much for your valuable thoughts. I use CPL most of the time when sun is out and there are blue skies, preferably with some white clouds.

The RV shot was 99% done with CPL, but still TTL should compensate for it, right? I am using 17-40 4L, 27-80 2.8L and 70-200 2.8L Mk II. When I look on my pictures in photoshop in the "grid" my general impression is that they are darker.

I used Nikon D80 before (not a great camera) but never had such issue. In fact it was opposite: bright sunny blue skies (mid) days meant -0.5 or - 0.75 compensation. With Canon 5D Mk III it is to have +1 compensation as default for any situation and bright sunny blue skies days it is up to +2. Recently I spent a week in Rome, taking pictures in churches etc and sometimes I had to compensate +2 at least. I am considering sending camera to Canon for check up. 

With the best regards,

Jan