The Light Leak Issue

[Mt Spokane Photography]

Perhaps the reason that this shows up with the cap on is because the sensor is actually really really sensitive, more so than on older models, and that is why it seems to show up on the high end gear only.
IE: The same minor leaks as on all gear, but now the sensor can see it.

Boom.

Actually... this is probably the best explanation I've heard so far.  The 5D3 does seem to be MUCH more sensitive to dark situations than any other body I've ever had the pleasure of trying.  I'd be very curious to see if the 1DX also does the same thing in idiotic lens-cap-on tests.

The image sensor has nothing to do with the metering settings on the display.  The metering sensor is under the mirror and is used to set aperture and shutter speed.  This means light is getting to the metering sensor.  Obviously, if it gets to the metering sensor, then it probably gets to the main sensor as well, its just not very much light.
 

[bormo]

Ok, just  made test on my 5dIII, have leak,

one more test
- close both caps (viewfinder and lens)
- set max iso
- set 30"
- make shot with backlight and without
- check difference (I have minor one, kind of 2 vertical lines)

On iso 6400 no difference

[Lstop-photo]

I have also notice a2/3rd of a stop difference on the 5D3. When I received the new camera I tested the exposure against gray targets, just to see how the new metering system was faring. I have noticed the same difference in my everyday shooting, and have set the cam at +2/3 pretty much consistently.

I don't know if this is related to the light leaked issue or simply the way the new metering system behaves, but I would be interested in knowing if other people have noticed.

Luc

[javier]

I have also notice a2/3rd of a stop difference on the 5D3. When I received the new camera I tested the exposure against gray targets, just to see how the new metering system was faring. I have noticed the same difference in my everyday shooting, and have set the cam at +2/3 pretty much consistently.

I don't know if this is related to the light leaked issue or simply the way the new metering system behaves, but I would be interested in knowing if other people have noticed.

Luc

Ok thanks I now know I am not crazy and this actually happens.

If I take a picture for example: Manual 4.0 / 1/200 / iso 800   &   I compare the same but in P mode the manual picture is a bit darker like 2/3 also you can notice that if you check the histogram.

since I start to take pictures with my 5d mark iii I notice the picture was a bit dark like 2/3 so if I want to take the picture right I have to set my meter to +2/3

[altenae]

I have also notice a2/3rd of a stop difference on the 5D3. When I received the new camera I tested the exposure against gray targets, just to see how the new metering system was faring. I have noticed the same difference in my everyday shooting, and have set the cam at +2/3 pretty much consistently.

I don't know if this is related to the light leaked issue or simply the way the new metering system behaves, but I would be interested in knowing if other people have noticed.

Luc

Ok thanks I now know I am not crazy and this actually happens.

If I take a picture for example: Manual 4.0 / 1/200 / iso 800   &   I compare the same but in P mode the manual picture is a bit darker like 2/3 also you can notice that if you check the histogram.

since I start to take pictures with my 5d mark iii I notice the picture was a bit dark like 2/3 so if I want to take the picture right I have to set my meter to +2/3

First do the test with the lenscap off.
With me and others including the test of CR no issue at all.
The metering is not responding to any light at all.

Test this first before you think there is a real problem with the 5d mark iii

[justsomedude]

... it just doesn’t appear there is an issue to me.

Wrong.  It's a HUGE issue for anyone who's favorite photographic subject is the inside of a lens cap!!!

All sarcasm aside... does anyone not see how this could be a serious issue for astrophotographers? 

If you are using a headlamp or the LCD backlight to set your exposure for a night sky capture, you could be in serious trouble.  I'll be doing some tests tonight and will follow-up tomorrow.

I fear the 5D3 may not be suited for astrophotography at all if this problem affects nighttime exposures.

I've not done any astrophotography, but I was under the impression that the onboard meter is perfectly useless for such work. Or any meter, for that matter -- I thought the serious stuff was all done either from well-established exposure charts or by stacking multiple exposures....

b&

Erm, no.

Because you would NOT rely on the camera's idea of what is correct exposure when doing astrophotography. Because if you did, the camera would seek to make the picture's lightness similar to that for normal photography.

You simply don't use the camera's exposure system when photographing star fields, because the overwhelming majority of the field is near-black.

You go to manual, or bulb.

Maybe I've been taught the wrong way, but I've regularly used the camera's metering system for astrophotography.  By ramping up to a high ISO (say 3200), and performing a short exposure, you can then calculate back out to an equivalent exposure at 100 ISO using a bulb.  But if your initial "base" (for calculating) exposure is wrong, your 3-hour equivalent will be shot, and no one likes to waste three hours under a night sky.

Granted, most astrophotographers will thoroughly check the LCD and histograms before moving forward with a longer exposure - it just sucks knowing that the initial calc made off the LCD display may be totally bogus.

This isn't really a deal breaker - since most longer exposures are done manually - it just means the LCD could be useless for performing one's exposure calcs.  Kinda crummy, especially since that LCD backlight is invaluable in the hills when its pitch black.  Looks like I'll just have to use the LCD preview and INFO screen instead, once the test exposure is complete.

Meh.

[altenae]

Read this on FM and I think he is 100% right.

this finding of a purported light leak on the 5D3, is typical of this stage of the life of a new camera model, in a world where too many buyers of such stuff are nerds rather than photographers (I confess being a nerd myself):

someone sets the camera up in a weird way, ie blocking any light from entering
this someone apparently doesn't realise that this means any further "experimenting" will be basically measuring noise.

But doing so in ways much less useful or professional than the manufacturer would
the same person then proceeds to flash a light at the camera. Woooo, apparently a tiny bit of that light gets in!

Should one therefore panic?

Yes if the person regularly (1) shoots in very very very very low light, and (2) regularly flashes strong lights at their camera while shooting in these conditions.

WHO does this? NO ONE. Astrophotographers DO shoot in very very very low light (not THAT low - nowadays any decent DSLR can capture say Orion, hand-held).
But they do NOT flash lights at the camera while shooting.
Camera makers are of course happy with any additional sale made - but frankly one sometimes wishes there were more photographers out there, and less failed testers / nerds. Puh-lease!

[javier]

I have also notice a2/3rd of a stop difference on the 5D3. When I received the new camera I tested the exposure against gray targets, just to see how the new metering system was faring. I have noticed the same difference in my everyday shooting, and have set the cam at +2/3 pretty much consistently.

I don't know if this is related to the light leaked issue or simply the way the new metering system behaves, but I would be interested in knowing if other people have noticed.

Luc

Ok thanks I now know I am not crazy and this actually happens.

If I take a picture for example: Manual 4.0 / 1/200 / iso 800   &   I compare the same but in P mode the manual picture is a bit darker like 2/3 also you can notice that if you check the histogram.

since I start to take pictures with my 5d mark iii I notice the picture was a bit dark like 2/3 so if I want to take the picture right I have to set my meter to +2/3

First do the test with the lenscap off.
With me and others including the test of CR no issue at all.
The metering is not responding to any light at all.

Test this first before you think there is a real problem with the 5d mark iii

I am talking when I take pictures normally

[epsiloneri]

Maybe I've been taught the wrong way, but I've regularly used the camera's metering system for astrophotography.  By ramping up to a high ISO (say 3200), and performing a short exposure, you can then calculate back out to an equivalent exposure at 100 ISO using a bulb.  But if your initial "base" (for calculating) exposure is wrong, your 3-hour equivalent will be shot, and no one likes to waste three hours under a night sky.

It sounds like a good way, but from what you describe it seems you are using the test exposures to compute the longer exposure time, not the camera metering system. Thus you would not be affected by this "light leak".

It's funny how when the 60Da was surprisingly released last week, people were complaining about the lack of Canon leaks.  There is really no way to please everyone 

[altenae]

I have also notice a2/3rd of a stop difference on the 5D3. When I received the new camera I tested the exposure against gray targets, just to see how the new metering system was faring. I have noticed the same difference in my everyday shooting, and have set the cam at +2/3 pretty much consistently.

I don't know if this is related to the light leaked issue or simply the way the new metering system behaves, but I would be interested in knowing if other people have noticed.

Luc

Ok thanks I now know I am not crazy and this actually happens.

If I take a picture for example: Manual 4.0 / 1/200 / iso 800   &   I compare the same but in P mode the manual picture is a bit darker like 2/3 also you can notice that if you check the histogram.

since I start to take pictures with my 5d mark iii I notice the picture was a bit dark like 2/3 so if I want to take the picture right I have to set my meter to +2/3

First do the test with the lenscap off.
With me and others including the test of CR no issue at all.
The metering is not responding to any light at all.

Test this first before you think there is a real problem with the 5d mark iii

I am talking when I take pictures normally

Yes I understand, but don't blame your issues which this leaking light with lenscap on.

[neuroanatomist]

The metering sensor is under the mirror and is used to set aperture and shutter speed.  This means light is getting to the metering sensor.  Obviously, if it gets to the metering sensor, then it probably gets to the main sensor as well, its just not very much light.

The metering sensor was under the mirror in old designs, but these days it's actually up in the pentaprism housing (labeled in the upper right of this diagram taken from a Canon tech report):



So, light that reaches the metering sensor should be blocked from hitting the image sensor by the mirror when it's flipped up during the exposure.

[prestonpalmer]

I am glad you recognized that light leaks through thumbs!  Good catch.  Since my 5D2 acts the same way, im gonna say NO ISSUE here!

Now, back to wedding photography.

[javier]

I am glad you recognized that light leaks through thumbs!  Good catch.  Since my 5D2 acts the same way, im gonna say NO ISSUE here!

Now, back to wedding photography.

What about the 2/3 stop off in normal use of the camera.

[altenae]

I am glad you recognized that light leaks through thumbs!  Good catch.  Since my 5D2 acts the same way, im gonna say NO ISSUE here!

Now, back to wedding photography.

What about the 2/3 stop off in normal use of the camera.

Javier, If you think it is the camera, return it.
My friend and I do NOT have 2/3 stop off in normal use.

[Rob Wiebe]

Hmmm only affects  my 5D3 without a lens and thumb instead of cover too. No issue with lens on or rubber cover on eyepiece .... LOL"