Canon Confirms “Light Leak” Issue in the 5D Mark III

[TCampbell]

Who meters their exposure while pressing the LCD backlight button? Seriously?!

Here's a quick firmware "fix": patch the firmware so that whenever the light meter activates, the LCD back-light DE-activates. That way it won't throw the exposure.

 

[Overread]

Problem is that it seems the light seal that the LCD light gets though also allows sources such as strong sunlight to also penetrate the body and cause exposure problems.

 

[Derrel]

I watched the videos on Pixelwhatever...the second video shows a scenario done in DAYLIGHT, with the lens cap on, to thus produce a long time exposure indication; when the man moved a small strip of paper near the front of the camera, on the lens side in front of the top LCD, the exposure shifted by basically 50%, depending on whether direct daylight was hitting the top LCD, or if the slight shadow the piece of paper was casting cut the light level reaching the LCD down a bit...so, it would seem that there is some type of "light leak" surrounding the LCD display...somehow it seems that light rays are influencing the metering through the LCD panel...

 

[TCampbell]

I'd need to understand how they re-created this in daylight.

True light is typically measured in "lumens", "lux" or "foot candles", but cameras measure in "EVs" (stops). Lux is measured linearly. EVs are measured exponentially. Each time the EV increases by 1, the lux doubles.

EV 0 = Lux 2.5
EV 1 = Lux 5
EV 2 = Lux 10
EV 3 = Lux 20
EV 4 = Lux 40
etc.

A daylight shot is EV 15 or Lux 81,920

If you're in an extremely dark place -- lets say it's EV 1 (which isn't completely black... but close). You have 5 lux. Now you expose the LCD to sunlight and note that it changes the exposure by half a stop (50% increase... a full stop would require a 100% increase). The real amount of light leakage would have to be 2-3 lux to have this much sway on the exposure.

And yet... the sunlight exposed would be over 80k lux... but only 2-3 lux managed to leak in. This could create a difference of 1/2 stop ... but ONLY at EV 1.

If you shooting outside in full sun (which is the only place where full sun should be hitting the LCD) and that same amount of light leakage occurs, the amount of lux is so tiny that it's basically not measurable. (it would be 1/40,000ths of a stop).

I'm sure the light leakage is real. But I don't think it's significant. I can't imagine how the camera got through testing with all the prototype bodies that went to pros and nobody noticed a problem. It seems as if the only way to measure the problem is in a contrived / non-real world test environment.

Let me ask the question a different way (since I'm still shooting with a meager 5D II and don't have a III).

If you stand outside with your camera at noon on a sunny day and meter a shot, can you get an EV 15 exposure reading with the LCD covered with light proof paper? (e.g. ISO 100, 1/100th or 125th and f/16) And... then removing the light-proof paper so the sun can now shower down light upon the exposed LCD... do you see the exposure change?

It's possible that I haven't seen the right video to demonstrate the problem and it's worse than I think. But so far, the videos that I've seen demonstrating the problem can only do so if the camera is in an extremely dark place ... OR require that you do something to make the camera think it's in a dark place (put the lens cap on... put a back-cover over the eyepiece, and then turn on the LCD or flood it with bright light.)

In essence, I suspect the reason this problem wasn't caught prior to production is because it doesn't impact real world shooting scenarios in any meaningful way.

 

[DScience]

I'm sure the light leakage is real. But I don't think it's significant. I can't imagine how the camera got through testing with all the prototype bodies that went to pros and nobody noticed a problem. It seems as if the only way to measure the problem is in a contrived / non-real world test environment.

In essence, I suspect the reason this problem wasn't caught prior to production is because it doesn't impact real world shooting scenarios in any meaningful way.

Friend, however you want to look at it, however you want to justify it, Canon is possibly selling a 'defective' $3,5000 camera. It does not matter in the least if it isn't affecting ALL real world situations. If the fault in the meter system affects ANY situation, then that is simply unacceptable in my opinion.

 

[unpopular]

??

isn't EV a measure of relative exposure not absolute energy?

 

[2WheelPhoto]

EV measures reflected light to simplify choosing among combinations of equivalent camera settings. Lux has not even been brought up by any of my professors so I doubt its really relevant to our pics or a leaky Canon =)

 

[TCampbell]

EV measures reflected light to simplify choosing among combinations of equivalent camera settings. Lux has not even been brought up by any of my professors so I doubt its really relevant to our pics or a leaky Canon =)

EV (exposure value) is an absolute measure of light using a logarithmic scale. You can directly translate between lux, foot-candles, EV, etc. But you have to double the amount of light to get to each successive level of EV. Think of it this way. You're turned on a 1 watt light bulb in an otherwise completely dark room. You have some given level of light. Now you turn a 2nd 1 watt light bulb. You've now doubled the light in the room. That's a full stop. That's a noticeable difference. Now imagine you have 100 of these 1 watt lights. You now add another 1 watt light. This time there's no noticeable exposure difference because of the logarithmic nature of exposures. You've only increased the light by 1% and that's not enough to effect exposure.If indeed you really ARE metering the light in a room with a single 1 watt light (one of the cases where the LCD light leak can be detected) then all you have to do is not meter WHILE activating the light on the LCD. If, on the other hand, you are metering in full sun, then the light leak isn't strong enough to impact the exposure.Either way it seems the leak is too insignificant to really be a problem.People are ignoring that allowing light to enter through the viewfinder eyepiece ALSO impacts the exposure... Pretty much on every SLR camera (digital or film) featuring a built-in light meter. Photographers have always known this and accepted it. The simple solution to that very old "problem" is to simple be aware of it and know to shield the light to get a more accurate reading.How is this any different?

 

[2WheelPhoto]

EV measures reflected light to simplify choosing among combinations of equivalent camera settings. Lux has not even been brought up by any of my professors so I doubt its really relevant to our pics or a leaky Canon =)

EV (exposure value) is an absolute measure of light using a logarithmic scale. You can directly translate between lux, foot-candles, EV, etc. But you have to double the amount of light to get to each successive level of EV. Think of it this way. You're turned on a 1 watt light bulb in an otherwise completely dark room. You have some given level of light. Now you turn a 2nd 1 watt light bulb. You've now doubled the light in the room. That's a full stop. That's a noticeable difference. Now imagine you have 100 of these 1 watt lights. You now add another 1 watt light. This time there's no noticeable exposure difference because of the logarithmic nature of exposures. You've only increased the light by 1% and that's not enough to effect exposure.If indeed you really ARE metering the light in a room with a single 1 watt light (one of the cases where the LCD light leak can be detected) then all you have to do is not meter WHILE activating the light on the LCD. If, on the other hand, you are metering in full sun, then the light leak isn't strong enough to impact the exposure.Either way it seems the leak is too insignificant to really be a problem.People are ignoring that allowing light to enter through the viewfinder eyepiece ALSO impacts the exposure... Pretty much on every SLR camera (digital or film) featuring a built-in light meter. Photographers have always known this and accepted it. The simple solution to that very old "problem" is to simple be aware of it and know to shield the light to get a more accurate reading.How is this any different?

Thats great, but I'm trying to figure out what it has to do with the scope of this thread or a Canon leaking?

 

[TCampbell]

Thats great, but I'm trying to figure out what it has to do with the scope of this thread or a Canon leaking?

Or maybe you're being deliberately obtuse?

 

[2WheelPhoto]

Thats great, but I'm trying to figure out what it has to do with the scope of this thread or a Canon leaking?

Or maybe you're being deliberately obtuse?

No, just not into the thread-jacking. Guess I got reeled in though

 

[Bitter Jeweler]

EV (exposure value) is an absolute measure of light using a logarithmic scale. You can directly translate between lux, foot-candles, EV, etc. But you have to double the amount of light to get to each successive level of EV. Think of it this way. You're turned on a 1 watt light bulb in an otherwise completely dark room. You have some given level of light. Now you turn a 2nd 1 watt light bulb. You've now doubled the light in the room. That's a full stop. That's a noticeable difference. Now imagine you have 100 of these 1 watt lights. You now add another 1 watt light. This time there's no noticeable exposure difference because of the logarithmic nature of exposures. You've only increased the light by 1% and that's not enough to effect exposure.

I just want to tell you how wonderful an explanation that is. :thumbup:

 

[Bitter Jeweler]

I'm sure the light leakage is real. But I don't think it's significant. I can't imagine how the camera got through testing with all the prototype bodies that went to pros and nobody noticed a problem. It seems as if the only way to measure the problem is in a contrived / non-real world test environment.

In essence, I suspect the reason this problem wasn't caught prior to production is because it doesn't impact real world shooting scenarios in any meaningful way.

Friend, however you want to look at it, however you want to justify it, Canon is possibly selling a 'defective' $3,5000 camera. It does not matter in the least if it isn't affecting ALL real world situations. If the fault in the meter system affects ANY situation, then that is simply unacceptable in my opinion.

Do Nikons come with a viewfinder cover? How often do you use it?

 

[STM]

It's a good thing I rarely photograph in sheer darkness or with the top LCD on.

But it said it strong light from the sun also affects the meter.

I would just add some small strips of black tape to the top then. lol.

Been there, done that with several Hasselblad backs where the foam which surrounds the dark slide needs replaced!

 

[Derrel]

Pro-level Nikon cameras have had a built-in eyepiece shutter since the early 1980's. This is used whenever the camera is set for long exposures that will be timed automatically (as opposed to the exposure being pre-determined and srt manually in M mode), or when the photographer will not have his eye at the eyepiece and in any conditions in which stray light could enter the camera and cause the light metering system to be influenced by "stray" light entering the eyepiece.

As to the 5D-III, I stopped by Canonwatch.com to read their analysis of the problem, and to check their links. It seems that this issues might go BEYOND JUST the top LCD...EOS 5D Mark III light-leak issue UPDATE | CanonWatch

Ron Risman at cameratown did some extensive testing to trace down the problem and provides test shots to illustrate the issue. The bad news: the light leak on the 5D Mark III goes beyond the LCD panel (says Risman):

"What I discovered today is that the Canon EOS 5D Mark III will give you a totally different exposure reading just by pulling your eye away from the viewfinder under normal indoor lighting conditions. We’re not talking sunlight [...]. With normal room light, turning on the LCD’s backlight does not effect the metering like it did when I had most of the lights turned off – yet regular ambient light getting in through the viewfinder was enough to fool the camera’s metering system, which changed the exposure setting."



​