5d Mark III - Metering Underexposed

Question

So first off, I apologize as I think I've seen a similar thread here but I can't find it anymore!
I purchased a 5d Mark III as an upgrade from my 7d a couple of weeks ago and, although I am happy, there is one main issue that is bothering me.
When I meter a "perfect" exposure, the camera is delivering about a full stop UNDER that.
My normal settings are; 
Shoot in M
Center focus point
Spot metering
AI Servo.
I generally shoot 2/3 over anyway, and find that I'm needing to now shoot at least 1 2/3 over to get the same result as my 7d. In addition to that, I'm getting a noticeable vignette in lower (but even/consistent) light situations.
I have tried all the different metering options, including variations on the focus points. I have done a full factory reset and the issue is consistent across all my lenses. 
Any ideas?

Answer 1

rjs1981 wrote:
Thanks for the response!
I just did some additional test shots to answer your question more accurately.
I shot a grey card under tungsten light at ISO 4000, 2.8, with a 50mm 1.4. I shot the same image with both cameras set up exactly the same, and the 5dm3 is, in that shot, roughly 2/3rds underexposed.
I'm really baffled. 
ps. I notice you use the 60Da. Never spoken to an owner. Do you love it? I am starting to really enjoy astrophotography so that unit is interesting to me. Its pretty much geared towards that one purpose yes?
You don't, by chance, happen to own an incident light meter to determine which camera is actually telling the truth do you? 
As you are shooting in "M" according to your post, the exposure compensation is ignored (it's used to P, Tv, or Av modes... but not M).  It's the only setting I can think of that would cause the camera to createa a different exposure.
So barring that, it seems one of your cameras is having a problem.   I once had someone tell me that their Canon 5D II and Canon 5D III were giving different exposures... but I own two Sekonic incident light meters and three Canon bodies.... so I tested the light with both Sekonic incident meters, then tested my bodies pointed at the same gray card and all agreed on the exposure.
But it sounds like you've done some reasonable testing to make sure the exposures *should* be the same... and yet they aren't.  It seems reasonable that it points to a problem with one of the cameras.  I can tell you that my own 5D III body (and I also own a 5D II body) are metering accurately so I don't think there's a 5D III metering problem per se, but you may have a individual body with an issue.
As for the 60Da... I'm quite pleased with this camera.  So here's the story.
I belong to a fairly large astronomy club in the area.  I'm going to guess there are about 160 members.  Of those... probably about 20% are fairly seriosu imagers.  A few owned the old Canon 20Da (the first astro camera Canon sold).  Many members own modified and unmodified Rebel bodies as well.  But when Canon released the 60Da, several members took notice and bought one.  
So one day I'm at a friend's house and the previous night another club member was at his private observatory with his 60Da and took a photo of the whirlpool galaxy (M51).  I was looking at the EXIF data for the exposure.  At the time I did not own a 60Da... but I did have the 5D II (I did not yet have the 5D III).  I took the _same_ exposure (ISO & shutter time) with my camera using the identical scope in the same observatory for the same object.  
At the end of my exposure, I got almost nothing.
I doubled the exposure. 
I could see a hint of the galaxy... but mostly nothing.
I tripled the exposure.
Now I was starting to see the hints of the galaxy... but nowhere even remotely close to what the 60Da had captured.   (and I'm using a 5D II which blows the doors off the 60D when it comes to ISO performance.)
The following day I ordered a 60Da.
Human eyes are a bit wonky... we are "most" sensitive to greens because they are pretty much smack in the middle of the visible spectrum (which runs from 400nm to 700nm wavelengths).  We are less sensitive to blues and reds.  Traditional cameras compensate for this in several ways... rather than "truthfully" collecting light, the Bayer mask is already stacked to double the green reception vs. the blue or red.  But even the filters inside the camera have a slow ramp up to block the IR.  The "IR" filter actually starts to block the spectrum gently even as low as around 500-550nm.   It ramps up gradually as it approaches 700nm.  
90% of the universe is composed of hydrogen atoms.  Atoms give off light at very specific wavelengths as their electrons jump from one shell to anotther.  For hydrogen, it's the Ballmer series where the dominant light is at 656.28nm (Hydrogen alpha wavelength), then hydrogen beta, gamma and delta... but those are cyan, and a few shades of violet and safely sharter wavelengths then what a terrerestrial camera IR filter blocks.  It's mostly the Ha which is a problem.  
The 60Da is at least 3x more sensitive to Ha (and possibly closer to 5x more sensitive) as compared to a non-modified terrestrial DSLR cameras.    The result is that not only do you get more reds (the Ha is a fire-engine red color), but overall you get much shorter exposure times to capture the same image.
There are several companies which make high end dedicated astro-imaging CCD cameras... SBIG (Santa Barbara Imaging Group), Finger Lakes, Apogee, etc.  These are typically monochrome cameras with peltier cooling systems that can chill the CCD considerable colder than ambient temps (becaue there's a relationship between physical temps and noise), have incredible well-depth (basically a measure of dynamic range), and fitler wheels.  Since a monochrome camera does not have a bayer mask, the cameras are much more sensitive to light -- but they can't see "color".  To compensate, a filter wheel rotates in a "red" filter, "blue" filter, "green" filter, and usually a "luminance" filter.  They may also use special narrowband filters to pick up Ha, Hb, O III, etc.   They take numerous images in each part of the spectrum and them merge them to create color image.  These cameras tend to be expensive.    I sure would love to own an SBIG STX-16803 and filter wheel but it's the better part of $12,000!  
The 60Da has been working quite nicely so far - a good workhorse and I've put it to use numerous times capturing images for hours on end.
Here's an image taken by my 60Da.  This is based on 16 combined "light" images of 4 minutes each of the Dumbbell nebula  (Messier 27).  I also took 8 "darks".  What I did _not_ take were any "flats" or "bias" images and it shows.  You can see the obvious vignetting caused by the telescope ("flat" images would have allowed my software to compensate for this.)   The red colors you see in this image are Hydrogen atoms giving off light in Hydrogen alpha wavelength.  Without a modified camera, you get red... but not nearly as much.
BTW... in fairness I should mention that I'm getting pretty good at image "acquisition" in astrophotography... but I have a lot of learning to do when it comes to image "processing" for astrophotography.  Most of my club members blow me away (even using the same camera that I use) but I am learning quickly.
Tim Campbell
5D II, 5D III, 60Da