[News] 5dmk3 Light Leak Issue Surfaces

It seems a new issue has popped up for the Canon 5dmk3. Multiple people are reporting that there is a light leak through the top LCD panel that directly affects metering. The easiest way to detect this issue is to put the body cap on the camera body as well as cover the viewfinder. Then place the camera in "P" mode at ISO800 and press the LCD backlight button.

As you can see in the video, the shutter speed changes depending on whether the backlight is activated. Similar results are achieved when a flashlight is shone onto the top LCD. Some users are even reporting that direct sunlight or walking into a room with bright overhead light is enough to sway the meter. If you are one of the lucky ones who managed to purchase a 5dmk3, it might be worth doing this little test to see if you need to seek a replacement version.

Test Procedure:
Remove lens and place the body cap on the camera body.
Place the viewfinder cover on the viewfinder.
(This should theoretically block any light from entering the camera.)
Set ISO to 800 and set the camera mode to "P".
(You should see a shutter speed of around 10 sec.)
Now either shine a flashlight near the top LCD panel area or simply turn on the backlight for the top LCD panel.
This will cause the shutter speed to fluctuate between 5-8 seconds, indicating light leakage.

Video example of sunlight affecting metering:

1st/featured video contributed by Sean Jaeho Yi

Lee Morris's picture

Lee Morris is a professional photographer based in Charleston SC, and is the co-owner of Fstoppers.com

Log in or register to post comments

ouch,.. what wrong with canon ...

 would be interesting to know if that is happning with every 5dIII. If it is so... it's a huge fuck up..

OMG what a fail 3500$. I hope not every one mk3 has it :I

Both of my Mark IIIs do this, but does it really affect real world use at all? I don't see any change in the metering when I actually have a lens on my camera.

it might if you're shooting star trails or something at night and use the back light to check your settings, it's easily avoidable but something that should be fixed for a camera of this calibre.

Guys! Have you tried this on the Canon EOS 5D Mark II? Guess what... it isn't a new issue. The 5D2 does the same thing and it isn't a light leak thru the LCD - it is an issue with light leaking thru the viewfinder. That's why there's a rubber cap to go over the viewfinder that comes with the camera.

I put my 5D2 thru the same tests and the 5D2 does the same... cover up the viewfinder and guess what... no exposure changes!

From the DPreview forum thread: (http://forums.dpreview.com/forums/readflat.asp?forum=1032&message=41...

From page 114 of the Canon 5D Mk III manual (and, coincidentally, page 86 of the Mk II manual):"If you will not look through the viewfinder when you press the shutter button, attach the eyepiece cover (p.185). If stray light enters the viewfinder when the picture is taken, it may throw off the exposure."

Did you miss the part where they covered the viewfinder and still got this problem?

Light leaking from the viewfinder is a problem that has ALWAYS existed on SLR's, and that's why expensive camera bodies has ha switch on the side that lets you cover the viewfinder without attaching a cap. My Nikon F2 (a model that came 1971) has this. 

What is shown in these videos is not light from the viewfinder, but from the LCD area.

Yup, I missed that step - sorry I blew up.

Well, this is an issue at ISO800 @ 10 secs (adds a stop). I'd imagine that at ISO12800 some photographers working at weddings or other low light jobs may experience same inconveniences.

Mine does it too...

and I have noticed (in my one week of ownership) that it likes to underexpose everything about a stop...
Not sure how much this "light leak" contributes to that... 

Ok, after reading the DPreview forum posts and trying some of my own tests... I'm not concerned. Maybe if you're doing long exposure night photography you'll want to not shine a light on the LCD area... but even so, nobody's shown that it impacts the photo being taken... maybe just gives the wrong exposure.

Seems like a non-issue.

but even so, nobody's shown that it impacts the photo being taken... maybe just gives the wrong exposure.

Ehhm, perhaps that's exactly what impacts the photo being taken: exposure is off by 2/3 - 1 stop. And yes: this is an issue when you're in the right circumstances (images at night, photographing people at a reception when you're just under a bulb, etc).

I agree that this won't be a problem at your usual outdoor situation, but please do keep in mind that the 5d and successors are being used rather a lot by wedding photographers... If you are at iso6400 or higher and your exposure is 1 stop off you probably won't like your result.

It isn't clear here if this is only affecting the meter reading, or if there is actually light falling on the sensor that would corrupt a long night exposure.  If the former, than I am not concerned (yes, I have a MK III).  For those night shots, I don't rely on the meter reading anyway.

However, if this actually means that the star-trails shots will have a haze or perhaps a corner that is ruined by spill light, then I am not happy at all...  I've only had the camera a couple days, and haven't had a chance for any night shots yet.  So far I am very happy with the studio performance of the camera.

mine doesn't leak.  just tried the above test. 

Just to be clear, it doesn't HAVE to be a light leak. Electronics used to Light the LCD could be drawing current that then drops the sensors current for Metering and causes it to misinterpret the true reading and give a false meter reading.

Heh that would be even worst for canon. Will see how canon will explain that problem.

 It appears from the second video that the sun will cause the fluctuations.

:) OK so dont use the lcd light :) Use a seperate flash light or a lighter. And for those who dropped 3.5k for it, be patient and canon will probably fix it for free.

Speical thanks goes out to you beta testers who paid the 3.5k for the untested camera bodies. 

 using a flash light on the lcd screen would probably cause the sensor to interpret it as a light-leak as well =P

test that theory, let us know ;-)

My 5dmk2 has the same issue. Also AF light will change exposure when tested with cup on. I didn't notice this before but now after 2y of usage I want my money back!!! or mk3 ;)

BTW there will be no problem in real life situation. The issue exists when light coming through lcd is much brighter then light coming TTL.

I think it's good we found out now, that if you're going to take a photo without a lens, and just a lens mount cover on, at night, at iso 800, with the view finder covered, you better not use the lcd illuminator, or you may not get the pic.... :-)

But wait, the guy from Adorama said it was the best camera ever and I should send in my check post haste!

Oh gosh... imagine if this camera was film. Would ruin every roll. 

I'm an avid Canon boy, but these continual issues with the 5D are really bringing me worry.

Pobody's Nerfect.

NEVER buy first product off the line.

 mk2 has the same "issue". Fortunately it doesn't cause any problem in real life...

I know right, but folks love a sensational story that absolutely doesn't affect them in anyway.

I just tested both my MK III and my older MK II.  The MK III _did_ exhibit this effect, though not as much as the original report.  Mine went from 4 sec to 3.2 sec in the test.  The older MK II had no change at all.

Again, if it is a meter reading only, I am not really worried. After all, if the body cap is on and there is NO light at all coming in, then an exposure of 4 sec at ISO 800 won't show anything anyway.  That is why I don't use the meter for night or special effects shots.

I have seen several night/high-ISO shots in online reviews, and none seem damaged by this.  Haven't done those myself yet with my new camera (just got it a couple days ago), but can't really say I am all that worried.  I expect Canon will come up with some firmware fix to this molehill-as-mountain problem.  In the meantime, I will just revel in my new-toy-happiness... :)

A hand held light meter is a nice thing to have. MAYBE invest in one of those? Just a thought :-)

Test Procedure:
Remove lens and place the body cap on the camera body.
Place the viewfinder cover on the viewfinder.
Set ISO to 800 and set the camera mode to “P”.

SO..... WHY IS IT Av MODE, 400iso, with a lense in the video ????!!!! 

The question is if this happens when a lens is on the camera as well... If so then this is a catastrophic failure from Canon!
It means that when out in bright light or situations where light shine onto the top of the LCD that it might throw the exposure of which could end up ruining your picture... That is a MAJOR fault!

You really have to wonder what is going on at Canon - The last 3-4 years their cameras have been plagued by one fault after another.

Over react much? 

Michael, relax... not happening with a lens on as far as anyone has been able to tell.

Plagued by faults one after the other? Huh? Canon has been outselling the market. The market doesn't think they are inferior products.

Well if you have paid a bit of attention to Canon for the last 4-6 years you would have noticed that we have seen quite a lot of issues. For instance the really bad autofocus issues with the 1D Mark 3 series... So bad in fact that many newspapers/sports photographers switched to Nikon to avoid the issue. You can find many of these issues with Canon... Seems like their QA isn't what it used to be - Just take a look at the numerous issues which keep popping up with the 5D Mark 3.

The point is that this problem is more than just a theoretical problem - It will directly effect your cameras ability to give a correct exposure which, in most photographers minds, is classified as a BIG problem.
Not to mention that something like that should NOT happen on a camera at this price level.

I get a 1/3-2/3 stop exposure change with the lens on if I turn the LCD on or shine a pen light on the LCD, so it definitely is happening with a lens on. I'm not really worried about any real world problems at this point, though.

Thanks for confirming the issue. But that is a real world issue - It means that the exposure of your images will be off which is definately a problem... Not to mention that something like that shouldn't happen in a camera at this price.

From what I have seen, there is no real world scenario in which this will affect me or change the exposure of my photos. Sunlight on the LCD doesn't seem to change my exposure, I'll never be shining a bright light directly on it, and any long exposure low light photography I do would not be in an automatic exposure mode and I wouldn't be turning the LCD light on. Unless I start to see my exposure changing with direct sunlight on the LCD, I am not a bit worried about this.

Oh Jesus, listen to you.

Yeah really useful comment... What part of the negative effect on correct exposure didn't you understand?

I think I did the test a little further.  The leak doesn't seem to actually put any light on the sensor, but affect the metering instead.  The following is what I did.

With the lens cap on M mode ISO25600 f/22 15s.  f/22 is to minimize light coming into the sensor from the front.  I then take 1 picture with the LCD cover, and 1 with the LCD not covered.  Check image through the historgram, you get identical image.  At this point I worry that this can cause a problem to my daily shooting, since the exposure metering can be a inaccurate.  So here comes test#2

Take picture in bright sunlight ISO100 f/2, meter stays the same regardless of rather I block the LCD or not.

That says, the metering error is probably not affect anyone in real life shooting, because the amount of light coming into the light meter throught the lens easily over power the error.  It's not like you are going to have bright sun hitting the LCD while you taking picture of a dark cave everyday.

Test#3 (Like the cave situation above)
Take picture inside my room with and without the sun hitting the LCD.  The room is at 8EV, while the sun is at 15EV.  The meter stays the same as well.

Therefore In situation where you are taking picture during the night where the lens is seeing something very dark, while you are inside a bright room.  It still wouldn't cause any error with at least 7 stops of light.

My copy has this effect...

I so rarely shoot using any kind of TTL metering, that I'm not worried, unless light actually hits the sensor, than that can be a problem.

Ive' heard so many pros saying that we should all shoot in Manual mode, so that we can get used to our eyes metering light, so we should maybe start doing that.  There have been a few things I didn't listen to instructors about, and later I realized how much less work you have to do in post, if only you spend a few more seconds before you start shooting, so that you get the shot right in the camera.

Canon will probably fix this problem/glitch, but do yourself a favor and stop relying on TTL metering, cause if you're using any kind of TTL metering settings, there is a good chance that your photos aren't really that important to you

There isn't really anything any of you can say to change my mind about what I just posted, so you can either take my advice or leave it, and I'm definitely not trying to be an ass, it's just something I feel strongly about.

Not relying on TTL metering might be fine for you, but for me? Ok then, I'll just try and pull an exposure out of thin air when I'm shooting my next band... I have no idea what the lighting guy is doing, but that's ok, I'll find a great exposure by the end of my 3 songs... Sounds like a terrible idea to me, so how about I buy a separate meter? Yeah great idea... I'm sure the performers will love me jumping up on stage with them to check the exposure... Even better is how it will change by the time I jump back down for the shot.

If I had the benefit of taking my time, or of real access to my "models" or control of the light, you might be right, but your situation doesn't apply to everyone...

Has no one taken time lapse into consideration? This is a huge problem for time lapse, especially those who shoot on av and then deflicker in post. Thats a lot of deflickering.

It does not affect anything. Stop crying kids.

Clearly, you don't know a thing about time lapse -.-

try without lens cap. then try without lens cap in very very dark places...
yea thanks its fine

@  rmond: Why would you shoot time lapse in AV?  Any  change in your frame will affect metering. Time lapse is supposed to be shot in manual.

What bothers me most about these canon cameras is that i still have to go out there in person with them to get an image. At 3500,00 dollars you'd expect this camera to do the work for you.

Epic fail Canon!

Many who don't actually own or use the 5d3 are getting overly excited on this light leak issue. Yes, it should not have happened for a camera $500 more expensive than the competitior. However...


Yup, the metering is still affected by the light leak with a lens and lens cap on, as least mine does. However, once you take off the lens cap, no amount of light in bright outdoor or dark indoor is going to affect the metering, simply because the light leak is so insignificant compared to what is coming through the lens.

Maybe it MIGHT affect the metering in scenes like nightscape and cityscape, though mine doesn't. However, unless you belong to the 0.01% of canon users who actually use the 1960s style viewfinder cover for a 15s exposure, it probably doesn't matter the least bit.

Maybe it really does affect the metering in really really low light, perhaps something that requires a 5mins exposure. For such low light condition, you have to shoot in bulb mode which is not affected by metering...


good point: your limited testing :)

So lets wait for more complex and full test to confirm that. And this is strange that 5DMK3 has the problem and the other doesn´t like 60D hmm.

Was just trying to be humble with "limited testing". I pretty much tested mine from bright daylight to dark rooms at night. Any darker, you will probably have to go bulb mode. This light leak is only an issue when the lens is off, or the lens cap is on. I am not so sure who shoots with the lens cap on...

You are taking it out of context by comparing it to a heater fan of a car. A new car with a bad heater fan is more like a new camera with outer focus points that don't work with wide angle prime. Sure, not everybody use a 24mm f/1.4(or the car's heater), plus not being able to focus with a wide angle prime on the outer af points doesn't affect the photo taking capability. I would agree with you if the light leak was causing issue with metering when the lens cap is on, but it just isn't.

More comments