The Canon EOS 5D Mark III “light leak” exposure meter concern

by planetmitch12 Comments

I'm sure some of you have heard the discussion that started on Saturday, but there's a minor issue that has been found with the Canon EOS 5D Mark III metering system when shooting in pitch black (or shooting with the lens cap on).

I debated about even reporting it, because some people have blown this way out of proportion, but it seems I really should at least let you know about it. My gut reaction was that posting about it might further inflame the passions, but I'm erring on the side of reporting it so at least you're aware of it.

First, let me say it isn't a “light leak” (which is what everyone is calling it – not my words) which implies that light is getting into the sensor and ruining your photos. In this situation, the light may be leaking thru the top LCD on the Canon EOS 5D Mark III and altering the reading the exposure meter is seeing.

Some of the reports

Here's Ron Risman's test as an example:

Ok, now, if you look at the videos – the vast majority of them are showing the 5D3 with the lens off and the body cap on, or with the lens on but the lens cap on – to make it really really dark. And yes, even on my Canon EOS 5D Mark III, I see the change in the exposure. Note: you also have to have the viewfinder cap on so there's no light coming in thru the viewfinder (which is a known issue – which is why they provide the cover already attached to the strap when you buy it).



My analysis

First, I'll give you the summary:

  • Yes, with the cap on, there is an exposure change when you press the top LCD backlight or put a strong light on the LCD
  • Canon is aware of the reports and is looking into the situation
  • In my testing during the day, there is no exposure change
  • In my night testing, I couldn't see any impact to the exposure
  • There may be some impact to timelapsers – but they shouldn't be putting a light on the camera in the middle of a sequence any way!

I verified that with the Canon EOS 5D Mark III set up the way it needed to be to verify the problem, I did verify my 5D3 does show the symptoms.

Yesterday – I took the time to go outside during the day and test. There have been several reports (if you read the threads above) where people say their Canon EOS 5D Mark III is underexposing when they are out in bright sunlight and it exposes properly in shade. I found this hard to believe so I went outside to check. First, I wanted to make sure that the scene wasn't changing (obviously if the scene or lighting changes, the exposure would change right? D'oh!). So I found a brick wall on my neighbor's garage and I could be in the shade of a tree and then in sunlight. I did see there was an exposure difference between sun and shade! Oops – soon I discovered that when I moved into the sun, there was some light getting on to the lens – thus invalidating the test… silly boy!

So I turned around and faced away from the sun and pointed my camera at the perfectly clear blue sky. I'd get a very consistent reading off of that! Plus, the sun would be directly shining on the top LCD! Guess what? I got the same exact exposure every time! I covered the top LCD… no change. I turned on the LCD backlight (which I couldn't even see in the sun)… no change in exposure.

I also went out about 10pm last night when it was completely dark and there were no clouds (here in the city, we get a brighter sky due to the lights when clouds are present).

I put the camera on a stable surface and pointed it at the back of the neighbor's house which was pretty dark. I set the ISO to 800 because that is what everyone said they were doing with the lens cap on tests (but it doesn't really matter – with the cap on, you can see changes at all ISOs). I noted the exposure (1.6 seconds) with the LCD light off… then I took the photo. Then I turned the LCD light on – no difference in exposure. I covered the LCD with my hand and covered the viewfinder (there was a single light over me on the patio) – no change in exposure.

I tested for about 15 minutes pointing the camera at different scenes and frankly I couldn't find any consistent changes to the meter readings under these night time shooting conditions. I didn't take a video cam with me so I don't have the documentation, but I was satisfied that under my normal testing, I didn't see any issues.

Final summary

While I agree that Canon should (and is) investigate this issue, I haven't yet found any video proof that there is a problem for either full sun or nighttime shooting. People say they see something, but without seeing the conditions, how are we to know if it is an accurate test? (And yes, I can see people saying the same thing about my tests since I didn't take a camera… but I'm a busy man and didn't want to take the extra time on editing and uploading).

There are people demanding a recall and claiming this is yet another example of poor quality from Canon, but they're yelling without anyone fully testing the situation! Calm heads should prevail here. Nobody should change their buying strategy nor should they be returning their camera to Canon. Let's see what Canon finds – I'm sure they've got excellent testing procedures.

(cover photo credit: snap from the video)

Comments

  1. This problem has been around since the introduction of TTL metering 50 years ago. It most-often occurred when the camera was on a tripod and the user’s head wasn’t blocking light entry through the viewfinder. Some cameras come with a cap to cover the eyepiece.

    It’s hard to believe that designers wouldn’t pay attention to keeping non-image-forming light away from the exposure sensor.

    1. Author

      William – this isn’t an issue with light coming thru the viewfinder… initially I thought that too… but this happens with the viewfinder covered as well as with the lens cap on.

  2. What am I missing here? Why are people exposing or metering with the lens cap on. Not being a spammer, but I don’t understand the logic.

  3. There isn’t an issue with image quality, but there is a real issue with exposure. When shooting from the dark in any mode but M the exposure will be affected when using the lcd-light. You’ll get slightly underexposed images. See these example photos here:
    nl.fotovideo.nu/nieuws/27104/test-5d-mark-iii-light-leak-bug
    (use your mouse to scroll over the photos to see the affected image)

    Translated:
    translate.google.com/translate?sl=nl&tl=en&js=n&prev=_t&hl=nl&ie=UTF-8&layout=2&eotf=1&u=http://nl.fotovideo.nu/nieuws/27104/test-5d-mark-iii-light-leak-bug

  4. Yes. I was taking some shots of my family this Saturday outside and it was sunny. This is probably why my photos are under exposed. On every photo, I was tweaking the exposure higher in LightRoom 4 because they were too dark.

    I’m starting to think this is an issue that Canon should take care of. The camera is great, but this is concerning me now.

    I guess I have to overexpose a little when I’m standing directly in the sun to get a good exposure, but then I’ll have to turn compensation back when I step into the shade. Such a hassle. I love this camera and don’t really want to return it. Please recall and fix it. I really don’t want to give up the Camera.

  5. I was shooting in AV if that matters. I just did a test tonight in my dim room with just the hallway light on in the other room. I mean really dim. It was exposing at 1/8 of a second. Turning on the backlight on the LCD or shining a flashlight on it didn’t cause it to change at all. I think this is more of a underexposure issue more than a wrong metering in the dark.

    I just realized that the view finder also leaks light into the meter. That’s normal. So outside when shooting in daylight and looking through the viewfinder. It would probably cause nothing. Unless your eyeball was pushed up against/into the viewfinder, which no one really does. I’m going to have to take some more photos when I have time off at high noon again and see. I’ll over compensate for exposure and normal exposure and see again. I hope it’s not really an issue. I could live with it, cause it seems minor, but for $3500. It shouldn’t be an issue. :)

  6. Just use exposure compensation, whats the big deal? Plus if you’ve got the experience to be shooting time-lapse at night you will most likely be on M anyhow.

  7. I’ve seen nothing but reliably operation over the designed 1EV to 20EV metering range. Go below 1Ev and the backlight does affect the metering. Which is the only thing that gives me pause on this, the camera appears to function properly over the designed operating range, and I have to wonder given that whether Canon will actually issue a fix or simply say the camera works correctly in the designed range. Personally, I hope they do issue a fix, but you never know.

    As for working outside/in bright light I see no difference in metering between my 1D-3 and 5D-3 under any exterior lighting conditions I’ve tried them in. Likewise, I need to dial in approximately the same EV comp when shooting bright subjects or in bright environments like a beach or other bright area. Shading the LCD, or even putting opaque black gaffer tape over it, causes no change in metering in any practical exterior shooting case on the bright end.

    In fact, the only use case that I can come up with that may be problematic is something akin to shooting a concert where there are bright spot lights flashing over the audience and potentially your camera. However, even then the subject on stage is likely lit better than 1EV, and thus the meter is unlikely to be affected. I’ve tried to test this scenario in a controlled way but I’m coming up short there.

  8. For what its worth, I’ve noticed that using the evaluative metering setting on the MKIII gives me slightly underexposed images most of the time. Using the center weighted setting gives me exposures which are more what I would expect (and closer to what my MKII gives me).

    This tells me that there is probably a difference in the metering computations between the MKIII and the MKII. Nothing I can’t deal with though.

Leave a Comment