Updated: April 11, 2012
After doing some more testing last night and again this morning it seems that the viewfinder / ambient light problem I mentioned in yesterday's update
extends to the Canon 5D Mark II (and I suspect other cameras as well). My Canon EOS 60D does NOT exhibit this problem, so this seems to have something
to do with the positioning of the light meter along the light path inside the 5D series (and maybe others. Now the question I ask myself is "If you have
owned the 5D Mark II for 3.5 years and never noticed this, should it bother you now." I'm still trying to figure that one out. I think I will forever
be aware of it at this point and will start to use the rubber eyepiece cover a lot more often - or will at least adjust the meter accordingly. Is this
a real problem? It's real, but whether it's an issue we should be concerned is another matter. I think you should probably do your own testing with your
camera and then answer that question for yourself.
Updated: April 10, 2012
Today I discovered that the light leak on the 5D Mark III goes beyond the LCD panel.
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, just one ceiling light that is located in front of the camera.
Tomorrow I'll test this same issue outdoors in daylight to see if the condition exists and if it worsens. 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 thruogh the
viewfinder was enough to fool the camera's metering system, which changed the exposure setting.
To see if this was a normal, I did the same test with the Canon EOS 60D and did not have this problem. The 60D gave the same meter reading when I
placed my eye against the viewfinder as I did when I pulled it away. This leads me to believe that the 5D Mark III does indeed have a light leak
issue that needs to be corrected.
I also discovered from a past service notice that Canon had a light leak issue way back in 2007 with the Powershot A650.
Canon needed to repair the affected cameras, which probably means this issue cannot be corrected with a firmware update.
Covering the eyepiece when your face isn't up against the camera
has always been a recommended practice, despite very few knowing about it - but this pratice was reommended mainly when shooting with the sun at your
back. I have never noticed any of my camera's changing exposure values when shooting indoors - until the 5D Mark III.
How will this effect your pictures?
This has the potential to over or under expose your images without you knowing why one image came out great, and the next came out much darker. If you
shoot outdoors with the sun behind you the difference in epxosure can be up to 2 or 3 stops. Indoors it seems to be about a 1/2 to 2/3rd stop difference.
How to avoid the problem?
You can avoid the problem, both indoors and out, by shooting in LiveView mode whenever you're eye is not in the viewfinder. By switching the
camera to LiveView the mirror inside flips out of the way to allow the light from the lens to go pass directly to the sensor, instead of to the
viewfinder. This is why the viewfinder goes dark when you switch the camera into LivewView mode.
Canon EOS 5D Mark III - Top LCD Backlight Leak Issue (April 9th, 2012)
Over the past week or so it has been reported by some users that the Canon EOS 5D Mark III has a problem with the exposure changing when users turn on the top LCD backlight.
I first noticed this problem two nights ago before I knew there was a potential problem with the camera. I was out shooting long exposure time-lapses two nights ago and thought it would be fun to
see what the HDR mode in the camera would do with a full moon sky and castle. I set up the time-lapse and pressed the start button on the intervalometer. After the camera started shooting
I was curious to see what the camera was selecting for the exposure during the three shots that make up each HDR image. Since it was dark outside I needed to use the backlight button on the LCD.
After letting the camera go for about 5 minutes (about 10 exposures) I thought it might be smart to see if I was getting the results I was expecting - before letting it run for 2 hours. When I
played back the first few captured frames I noticed that every other frame was lighter / darker than the previous one. I only scrolled back & forth through the first four images, but based on
the inconsistent exposures I decided to scrap the HDR idea and went back to the normal long-exposure time-lapse.
The next day I read a post on twitter about the possibility that the backlighting on the LCD display could cause an exposure shift. It then dawned on me that this was the problem I was having
with when I tried to capture the HDR time-lapse. Because I was turning on the LCD backlighting in order to check the exposure values the camera was underexposing those frames, but not the ones
where I didn't bother to check. As I researched the problem online I read conflicting information, which got me to do some of my own testing - based on what I remembered from my time-lapse the
night before. So here is how I tested for the problem, as well as the results.
I placed the 5D Mark III on my desk, aiming at a bulb blower on my desk. The scene was purposely lit with my iPad running "Soft box Pro," an app that makes the iPad act like a pseudo soft box for
close up photography. The iPad was positioned about 45 degrees to the left of the camera. Behind the iPad (not in the photo) is my 20" flat panel display, which was also providing some ambient
lighting to the left of the set up.
In this test I placed the camera in the Auto (Green) mode and took two exposures. One with the backlight on the LCD turned off, the other
with it turned on. As you can see in the photo below, when the LCD was turned on the camera set the exposure value to 1/5th vs. 0.3" when
backlight was turned off. You can see that when the backlight was turned on the exposure was slightly darker. Is this a major
difference? Not in my view, but in some cases it could be help determine whether you're maintaining detail in the areas you want. In the green
mode the camera selects all parameters, which is why I wasn't able to keep the ISO Setting at 1600. Regardless, the results are the same.
In this test I placed the camera in the Program (P) mode. With the backlight turned on the camera
selected an exposure value of 1.3" seconds vs. 2.0" seconds when the LCD backlight was turned off.
This is a 2/3rd stop difference.
In this test I placed the camera in the Av (Aperture Priority) mode and once again, with the camera
set an F/4.0 the camera selected an exposure value of 1.6 seconds with the backlight turned on and 2.5 seconds
when it was turned off. Again a 2/3-stop difference.
In this test I placed the camera in the Tv (Shutter Priority) mode. The results are the same, but in Tv mode I
set the shutter duration and the camera then changes the aperture based on what it believes the exposure should be.
With the backlight turned on it set the exposure value to F/5.6 because the leak from the LCD backlight was fooling
the metering into thinking the scene was brighter than it really was. With the backlight turned off, the camera set
the aperture at f/4.5.
In this last test I placed the camera in the M (Manual) mode. In manual mode the camera cannot adjust any of the parameters,
which means the image in the example below was not affected. However, when shooting manual mode, we typically rely on the meter guides
to know when the scene is properly exposed. A proper exposure would have the mark on the meter dead center. With the backlight turned ON
the meter read dead center, meaning the camera felt was exposure was correct. When I turned the backlight off, the meter was about 1/4 way
to the left of center. If I had actually adjusted any of the values in order to get the meter back to the center the left hand image would have
turned out brighter, just as they did in the examples above.
Some have argued that this problem really isn't a problem and that it is similar to light getting in through the rear viewfinder, which all SLR style
camera's have issues with. I don't agree. Here's why. When I shoot star lapses or any type of long exposures at night I rarely worry about light
getting into the viewfinder. A) Because there typically isn't any and B) If there is, it will most likely be consistent throughout the entire
time-lapse sequence. I can also block it using the included rubber slide-in eyepiece blocker. However, when shooting in these same dark conditions
I have to rely on the top LCD as it's the only way I can check or double check my settings once a time-lapse or exposure is under way. When the
camera's shutter is open the viewfinder information goes black, so the only place to get the readings is on top. Since it's dark you'll need to light
the display, either using the backlight button or by using a tiny pen light or LED held close to the LCD. In either of these cases, light will leak in
to the camera, and can affect the exposure of the next image being taken.
In my testing it's important that I point out that the light from the LCD backlighting or from a pen light does not actually affect the photo. Once
an exposure is under way you can turn on the backlight and your image will NOT be affected. It will, however, alter the camera's metering, which will
have an affect on how it exposes the next image.
When shooting long exposures or night time-lapse sequences you will usually set the camera to manual mode, in which case your exposure won't change from picture to picture,
even if you turn on the LCD's backlighting. I only happened to discover it because I wanted to try the HDR mode while in Aperture priority mode. You will most likely not
be affected in video mode as well, since you'll most likely be shooting in manual mode.
I think it's important to keep in mind that this problem will affect only a small portion of 5D Mark III users since it only shows up, as far as I can tell, when light getting through the
LCD panel is brighter than the ambient light getting in through the lens. With that said, this IS something that Canon should correct. Either by reducing the camera's meter by 2/3 stop
whenever the backlight button is pressed in low-light or by having users send the camera back to properly seal the area around the top LCD display.