metering issues with 5D MkII

Discussion in 'Canon EOS' started by peterlyons, Jun 15, 2009.

  1. I'm a professional maritime photographer in San Francisco. And I've been shooting for years. I understand the difference between different metering patterns (spot, center-weighted averaging, and pattern). And I know, too, the kinds of lighting and subjects that are likely to fool a camera's meter, and have learned to compensate when dealing with those.
    I almost always shoot in aperture priority, and dial in exposure compensation to compensate as needed. For instance, I'll bump exposure up 1 or even 1 1/3 stops when shooting wide shots under gray skies, preferring to keep the clouds bright, and bring up the detail in my darker subjects. Or if a boat sails in front of a dark background, I'll compensate -1/3 or -2/3 to keep the dark areas dark, and hold detail in the white sails.
    Recently some of the exposures calculated by my 5D MkII have me wondering what I'm missing. I don't understand what is causing exposure to be so far off in some scenes.
    I will attach some samples below.
  2. 1/500 f/8, iso 400, +1/3 exposure compensation, pattern metering
    5D MkII, 24-105 L, Singh-Ray LB Warming Polarizer
  3. 1/1000 f/8, iso 400, +1/3 exposure compensation, pattern metering
    5D MkII, 24-105 L, Singh-Ray LB Warming Polarizer
  4. These two exposures were taken seconds apart, with the same camera, of the same subject.
  5. [[of the same subject.]]
    With vastly different framing.
  6. And how about these two. The first of the next two attachments was shot with the 5D MkII at the same settings as above, only this time the camera chose a still lower exposure of 1/1250 at f/8. And yes, this is still with +1/3 of exposure compensation.
  7. Compare the above with another shot, similar subjects on the same day, taken with a 5D and 100-400, set to Aperture Priority and iso 800, also set for +1/3 compensation.
  8. Rob-- Yes, I changed the composition, but I don't see anything in that shot that should fool the meter so badly.
  9. Comapring photos 1 and 2 I would say #2 has much more reflective water in the picture and secondly you are using a polarising filtern and have changed the angle of the shot quite drastically in relation to the position of the sun (maybe by 60 degrees of more).
    With Photo 3 and 4 seem almost non-comparable to each other or the other shots.
  10. I'm having the same problem too. Underexposure in certain daylight/overcast conditions in evaluative metering. Spot metering overexposes. I also think the metering is off from exposure to exposure, sometimes even when the subject has not moved at all. This is all under aperture priority.
  11. Your first pair of shots - assuming that we see the whole frame in both - confuses me. It could be that in #1 exposure was largely based on the very large area of shaded sail, and this allowed the much brighter area of directly sunlit boat stern to blow out. I'm not sure why the second is so dark except that it looks like it might have been better without the exposure compensation dialed in.
    In the second pair the cause seems more obvious. In the first shot a good percentage of the frame is the very bright boat, and with your compensation dialed in and the meter zeroing in on that the frame ended up dark. In the second, tighter shot the meter picked up on your central subject, the shaded persons in dark clothes, and allowed the bright white sunlit portion of the boat to blow out.
    In situation like these, if I had time, I would perhaps go with manual exposure.
    I understand that you have shot for years, and that you understand all the basics of metering and so forth. How new are you to digital as opposed to shooting with film? If you thoroughly understand the differences in how you need to deal with exposure issues, I won't offer any more than that question - except to say that film-based exposure methods do not always turn out to ideal for digital.
  12. I've never used a 5D, but I do know that with the Canon DSLRs that I have used, the matrix metering is highly dependent upon the focus point. Especially when using the cameras set up for using only a single middle focus point, the exposures can be wildly different. Don't know that this has anything to do with your problem, but it's probably a factor to be considered.
  13. I've never used a 5D, but I do know that with the Canon DSLRs that I have used, the matrix metering is highly dependent upon the focus point. Especially when using the cameras set up for using only a single middle focus point, the exposures can be wildly different. Don't know that this has anything to do with your problem, but it's probably a factor to be considered.​
    This must be it. Your explanation *totally* meshes with my experience on the 5DmkII. I was using the central AF point when I noticed much more variation in exposure.
  14. Thanks for all the feedback, folks. The first shot I posted to show an exposure that works for me. The stern may be very bright white, but I have all the detail I need in the RAW exposure, so I'm fine with that. In the second exposure, after recomposing, the whole scene appears very dark to me. There's no cropping there; you're seeing the whole thing. A blue sky isn't going to fool the meter, so I'm not clear about what did.
    Dan, while you said the second shot may have been better without compensation dialed in, I think you misunderstood. I had the camera set to OVERexpose by 1/3 stop. Without that, it would have been even darker.
    I thought the two tighter shots were pretty similar in lighting angle, but I accept the point that the darker one has more areas of sunlit hull closer to the center. Still, I've never had such trouble before. Exposures missed by such a wide margin are pretty rare for me, and I don't see the cause in these images.
    I'll definitely check on the exposure point selection though, to see if that's a factor here. I do select different points for different shots.
  15. Have you tried without the PL?
  16. Dan, I've shot film in the past, but my pro career has been completely digital, so I'm familiar with how to expose best for it. Typically I favor exposing to the right, as long as I'm shooting RAW. I get my highlights as bright as possible without blowing them (gray skies excepted--they can blow out), because shadow detail, especially crew faces, etc., is of primary importance to me.
  17. Arash, no, I almost always use the polarizer. IMO it's a necessity to get good contrast and color on the water, especially on sunny days.
  18. I've been shooting a 5D for over 3 years and a 5D Mark II for over 4 months. I too use a polarizer whenever I'm outside. I almost always use evaulative metering.
    I've never encounterd the kinds of stark differences you've shown in your pics. Mine have always been very consistent. The 2nd and 3rd pics I find to be very unusual and not anything I would expect.
    My recommendation would be to send it in to Canon, along with the photos, have the check the camera over.
  19. Here's what might be happening here. The exposure dilema seems to be a result of your use of that polarizer. Remember that with any polarizer the exposure value is affected by the relative position of the sun and the degree of rotation of the circular polarizer to attenuate the reflections.
    Given that the strongest attenuation of those reflections is 90 degrees to the right or left of your subjects position to the sun and based on what I see here in your posts your subjects all seem to be about 90 to 100 degrees clockwise from the suns position.
    My questions are did you adjust the degree of polarization for each shot or did it change, and if so, what type of metering was used? Backtrack your steps and I'll bet your question is easily answered.
  20. Peter: First, sorry for my misreading of the exposure compensation info. Second, it sounds like you do understand the ETTR business with digital. Good luck sorting this out.
    I'm a 5D and 5DII user and I haven't noticed any camera related exposure problems with my bodies.
  21. Peter,
    I understand that PL is essential to your work and in principal it should not affect metering because metering is TTL, but it is possible that the sensor is not fully calibrated with the meter when it comes to polarized light because of the various reflective layers that exist on the sensor, and thus at certain angles meter sees more light than the sensor, the best way would be to try without a PL first to see if you can get consistent exposure, this will eliminate the chances of meter being defective. If this turns out to be the case then you have to find a solution for the PL, maybe try a different brand that Canon recommends?
    If meter is inconsistent without the PL, I'd send the camera to Canon for calibration.
  22. I've noticed that in each image there is a piece of bright white that is near clipping. It's just a wild guess, but do you have highlight priority turned on?
    Probably not, but I just thought I'd ask to get it out of the way.
    My recommendation would be to use center weighted average metering with your compensations dialed in as normal, or better yet go full manual and spot meter (or chimp) if you have time to kill. I understand it's a pain to switch the way you operate, but if it was me, I would make sure the meter is working accurately with a grey card, and then avoid matrix metering in that particular lighting situation.
    It might also be the filter, as others have said. Perhaps it is somehow partially blocking the wavelengths that the meter responds to?
  23. I couldn't find anything in the manual about pattern metering, but if that also means evalulative metering:
    The manual states that when using AF (one shot) and evaluative metering, the metering occurs at the time focus is acheived, not when the shot is taken (like exposure lock). Are you perhaps recomposing, or is the scene recomposing itself on the high seas?
  24. Peter, I have had the camera for 6 months and almost 10k shots. I still can not figure out how it meters. I am not new to photography either but I am puzzeled by the various exposure I get with this camera.
    Maybe it does need calibareation as Arash suggested.
  25. Peter - I have shot film for over 30 years (35mm and MF) and recently committed to digital with a 5DII (I toyed with a rebel for a while). I found the metering took time to get used to - sepite the fact that I was using Velvia and have every pro Canon body from the New F1 onwards. I find that polarizing filters and ND grads produce much more variable effects on a digital body. Before someone writes in this is with circular polarizers - I understand the issue with liner polarizers. I am not sure if the issue is the digital sensor or the algorith used for the evaluative metering on the 5DII. I took a whole bunch of test readings from a grey card with 11 canon bodies under controlled lighting and found all of the 5DII metering results were within a 1/3rd of a stop of the film bodies (some of the older bodies such as the F1N and A1 only meter in 1/2 stops). Thus the reason is either the sensor or the algorithm. I should have taken the tests using a polarizer and grads but did not - I may try this next time I have a spare hour. What I have found is that I like to shoot in RAW, I use AEB a lot more than I used to (usually +/- 2/3 or 1 stop) and try and use manual with a polarizer or ND Grad. I also tend to add =1/3 of a stop to most "correct" exposures to get a velvia like result.
  26. I too get variations in metering with my XTi, In a burst of continuos shooting there will be one frame +1 while the others are ok. Looking through camera tests I notice that Canons metering modes have algorithm that takes a lot of notice of the central area, different metering mode selections just seem to make this area larger or smaller. This means that the frame doesn't have to move far to include or exclude a bright patch in the metered area. Your white sails and white boat could be having vastly different influences on the exposure setting depending on whether these areas are included or excluded from the metered area.
    Doesn't fix the problem but might explain why it happens.
  27. Philip:
    Just an observation... if every response is met with "I have a ton of experience (so don't tell me about basic stuff)" there is a possibility that you and those trying to help may decide that it is better to withold suggestions that might be useful.
    I certainly understand that you have a lot of experience with photography, but all of us are capable of having blind spots - I know that I sure do from time to time.
    Take care,
  28. Lots of good thinking here. Thanks to everyone for the intelligent responses.
    The more I look at my recent shoots, the more I'm inclined to think there's something wrong with the camera itself. Have a look at this. They're two different boats, but the composition is exactly the same, and they're shot within seconds of each other. Both on the 5D MkII with the 24-105, Aperture Priority, evaluative metering (I was calling it pattern because that's what Lightroom calls it), and both with +1/3 stop exposure compensation:
  29. Lots of good thinking here. Thanks to everyone for the intelligent responses.
    The more I look at my recent shoots, the more I'm inclined to think there's something wrong with the camera itself. Have a look at this. They're two different boats, but the composition is exactly the same, and they're shot within seconds of each other. Both on the 5D MkII with the 24-105, Aperture Priority, evaluative metering (I was calling it pattern because that's what Lightroom calls it), and both with +1/3 stop exposure compensation:
  30. underexposed
  31. properly exposed
  32. Dan-
    I'm not trying to sound like a know-it-all. Discussions of exposure metering can begin at a pretty basic level, and I was only trying to say that I have a good grasp on this stuff, and am familiar with the scenarios that I've seen fool meters. But clearly this has me stumped, so I'm all ears. If there's something basic that I've missed, I'd love to hear it!
    thanks for your input!
  33. That first image isn't even close to correctly metered. Look at the histogram and there's barely any white. :(
  34. Neill, the situtation you're describing with one shot in a series being off a stop from all the others...
    THAT sounds bizarre. I'm not sure what would cause that kind of result. It sounds like the kind of variability I get when I accidentally switch to spot metering.
    As to the polarizer being the source of trouble... I've heard about issues like this before, but have never seen them in my work. I wonder if there's something with the AF sensor in the 5D MkII that causes it to respond differently to polarizers than the 5D does.
  35. From the info you gave:
    1. Center point AF.
    2. Av mode.
    3. Center-weigh metering.
    Look at the 1st pix, AF center point meter system would see the light brown color of the sail. Thus it tells the camera to open up a bit. In the 2nd pix, the AF point see the white cloth of the sail, therefore it tells the camera to close down a little.
    Same thing happen in pix 3 and 4, except that in #3, probably it see part of dark color, part of white. In pix #4, it sees the dark color of the shirt.
    BTW, do not use the years of experience as a supporting argument in a talk, because it is a fallacy:
    1. Does the experience closely relate to the matter at hand? For example, a person with 20 years of indoor photography experience may trip when taking pictures outdoor.
    2. E.g., a person takes up a camera everyday in 2 years may be better than another person in 20 years, but one day per year; assuming they have the same IQ.
    3. Depending heavily on the IQ (not Image Quality, mind you) of a person, eg., how quick he/she learns and digests a lesson.
    And more, but these are the few.
  36. Peter, the camera is working fine. The issue is that you are wishing for consistent exposure more than correct exposure. If it's consistency that you want, then you have to use manual exposure in order to keep the exposure fixed from frame to frame. You can meter off a gray card or just take some test shots in manual mode, and once you are satisfied with the settings, leave them alone or if you need to change aperture, shutter speed or ISO, you have to manually (*sigh*) adjust one of the others to compensate.
    If you use auto-exposure, the camera will try to pick an exposure that it hopes is what you wanted while preserving some post-processing flexibility. From this perspective, all of the shots you posted look reasonably well-exposed to me. All the camera can do is guess at what should be exposed for mid-tone. It looks like it's always assuming that the center of the frame should be exposed as gray, and letting the top and bottom thirds of the frame to fall where they may, even if they might be black or blown out a bit. Some of your pictures have very wide dynamic range right in the center of the frame--from black to white--and I think the camera did an ok job at picking a compromise exposure that preserves flexibility in post-processing to go up or down to pick your black and white points.
    Now, a real complaint you could direct against Canon is their very weak manual mode. It would be genuinely useful if Canon had an "exposure value" mode in which you could choose some combination of shutter speed and aperture and lock it down. Then, when you make an adjustment to one would automatically shift up or down to compensate. Maybe they could associate this behavior with the "*" button in manual mode; pressing it would interlock the aperture and shutter-speed dials until you press it again. They could call it "manual shift mode", but I would call it "Hasselblad mode", since this is the way that the C-type lenses worked.
  37. PS: Re-reading your original post I realize now that you probably know all of this. But do give manual mode a try. I don't know why people think it's more work or takes more time. It's the easiest approach if your lighting is consistent (as it is in your examples).
  38. I don't understand why you wouldn't just meter manually. You are in a very consistent light space, shooting in the same direction. What I do in situations like this is take a test shot of something representative that I'll be shooting, check my histo and nail the exposure down. I change it depending on my position, but the results are as consistent as the light, very. I personally have not given up my incident light meters, they are the most consistent. In camera, I find the evaluative pattern to be the least consistent, and the average meter pattern to be the most consistent and easiest to anticipate where it will be fooled if I am going to use auto exposure in a situation. Technology is great, but remember that it is still pretty dumb when it comes to simple things.
  39. use film
  40. Evaluative metering is the most inconsistant metering mode one can choose, because it is the only one that has some "brain"! In this mode, the camera is going to use all the information available, from the different metering points, from the focusing distance and focusing points, etc. and it will try to come up with a good image. This can lead to very different exposures for a very small change in the view.
    Exemple: your camera has focused on a white part of the scene and the camera decides that that is what should be the reference, so it will under expose the picture; next frame you slightly recompose and the focus is made on some shadowy part, again this will be recognised as the relevant part to render correctly, and the white part will be blown.
    Keep in mind that this is a very simple exemple, the exposure algorithms in the evaluative mode are very complex. Citing the EOS5 manual: "After detecting the main subject position, brightness, background, front and backlighting, etc., the camera sets the proper exposure"
    But basically the result is that: the camera decides what IT deems is important in the frame and exposes accordingly.
    If you are used to compensate, just use the "center weighted average mode", this should give you a much more consistant exposure.
  41. Again, based on the last 2 photos you showed and my own experience with my 5D and 5D MK II, I would tend to point the finger at the camera.
    My recommendation would be to get Canon to check it over. Make sure you send in those last two photos
  42. If you want to know if the camera is broken, simply test it. Shoot a gray card. Vary the metering mode. If the frame is filled, each shot will be identical middle gray. If the images don't have histograms that show a centered spike, the meter is off.
    Do it with and without a polarizing filter.
    It is unlikely that the meter is accurate and the evaluative meter mode is broken. It is quite possible that the evaluation program assumes the sail is a face and is giving it a neutral 18% exposure value.
  43. Peter, as an advanced amateur with a love of the subject matter (sailing vessels), I certainly sympathize and share your exposure concerns. I upgraded to the 5DII as soon as I was able to obtain one (two months on the waiting list at my favorite store) and find similar issues with using pattern as well as center-weighted averaging. Prior to going digital with 4 different models of Canon EOS digital equipment from the 10D to my current body. Prior to the digital shift, I used a wide variety of film cameras over a 30+ year period. TTL metering has been a bugaboo with me unless I had no better option. Then I experimented with AA's Zone System which I found solved most of my problems when I could properly evaluate using spot metering of desired midtones (Zone V), which I still prefer over any other method.
    As for using polarizing filters, I have tried both circular and linear polarizing filters (PL), with varied success, but always with better exposure values in reflecting surfaces and sky areas. Given all of this, I note the following issues which were not noted in any of the previous discussions I read:
    Photos 1&2 show one major difference which I feel may have had an effect on the exposure given the other factors discussed; the angle of the sun on the subject of the image (the boat and sails) would have more of an effect on averaged/pattern exposure than the angle of the sun to the polarizer. Given changed angle of the background to the lens as shown by the diffence in the background, a significant change in the lens sun angle could cause the exposure difference shown given the use of the PL.
    Photo 3&4 show a significant change in focal length as well as in the amount of light toned areas (deck and background areas) which could have significant effect on pattern and averaging metered shots. Again, spot metering for a midtone might have enabled more consistant exposure.
    Photos 5&6 added on the 16th immediately indicated a difference in 1) the boats are different (sail numbers) and each is on a different angle of heel causing significant differences in the relative tone caused by minimal direct sun on the facing surface of the sail in photo 6, while much of the mainsail in Photo 5 is showing the shaded side of the sail (backlit) in relation to the nearly overhead sun.
    Again, I'm only an amatuer, but I personally prefer spot to average or pattern metering.
    Thanks for the opportunity to comment.
  44. My EOS 5D MKII consistently overexposes every daylight outdoor scene by 2/3. I have been shooting canon for 25 years, and have shot probably millions of shots with the EOS 5, 10D, 20D, 40D and never had single cause for complaint.
    The 5DII however is simply unusable outdoors. Even if you point it at a large white object, it over exposes that by 2/3. i went through phase of shooting every shot with the 40D, 5DII 0ev, and 5DII -2/3 EV. Guess what? near 100% of 5DII -2/3ev are the same as the 40D 0ev, which is so close to perfect I almost never have to compensate (unless there is a very strong backlight or similar).
    I called canon and sent them examples (I have thousands), but they said they had not heard of this problem and offered no help or alternatives.
    I shoot weddings part time. Outside, 5DII is unsusable, as it overexposes. the 40D is fantastic, as I can always get perfect shots. Indoors the 5DII is good, but not as consistent as the 40D.
    I went though a perod of just leaving the camera in -2/3, but occasionally the camera gets the exposure right, so Im left with a badly underexposed image. The anoyting thing is, I cannot work out when this is going to happen.
    I normally use evaluative meetinging, as have to shoot from the hip at weddings and sports events in lots of different conditions.
    Anyway, i posted some early comparison shots and got a lot of pointless replies. Note: fotki incorrectly reports that the metering was spot - if you download the full sized original you will see the correct exif data. look at the levels on these shots, and you will see all the 5D shots are shifted to the right by 2/3, and the 40D are bang in the middle.
    I only ever use the center point to focus. Most of the test shots were not recomposed.
    I have experimented with focusing on light/dark etc. but it doesnt seem to have much difference.
    People tend to say things like "use manual" or "use your light meter". I have not had to do that since the EOS 5 20 years ago, why do I have to go back to manual now? You can keep the 5DII if thats the price i have to pay, the 40D is far far better at meetering - I would go as far to say is for me, its perfect.
    P.s. I almost never use a poloriser - I tend to get better results in photoshop using levels and curves. Lightroom is not so good at this though.
  45. turn off bracketing
  46. I seem to have the same problem. I just purchased my camera last week, but to get good images I need to leave the camera on -2/3 and they are perfect, but it is a bit annoying as I was used to get perfect readings from my Nikon F100. Not sure what to do.
  47. I purchased my 5D MkII in the spring and finally received it about a month ago. I am a Las Vegas Wedding Chapel photographer... meaning some days I shoot 8 weddings... over 1,000 shots a day...outdoors and indoors. Indoors we have 3 chapels. The other photographers and I that shoot Canon have all been shooting with the older 5Ds and with either the 24-70 or 24-105 lenses. We have tested many different settings and finally found that shooting with speedlight 580 exII that we would shoot ISO 640, 60th shutter speed, f4 with internal comp up 1, neutral setting, 4000K... trying to neutralize color, exposure, etc... to get even results between all of us shooting. THEN... I got my new 5D. I knew that it would be different... but I thought for the better! I tried these same settings and they were SOOOO dark one minute and then COMPLETELY overblown the next.... doing the same thing (metering, bouncing flash, etc.) that I have been doing with the old 5D for over a year. That was my first problem. THEN... 3 DAYS INTO USING THIS NEW CAMERA... I leaned in to shoot a limo shot and the entire viewfinder lit up... all 8s, ever meter bar showing, etc... the viewfinder was completely unreadable. At this point I called Canon and then had to send it in for repair. They found that the pcb assembly was faulty. Now I've had the camera back for 2 days and last night I'm in the middle of a shoot and it just won't shoot. The lens is focused, red light blinks, and then NOTHING! I press the shutter button, red light flashes that it's focused and then NOTHING! At first I thought perhaps it was the lens so I changed out lenses and again NOTHING! I had to completely switch out cameras for the rest of the shoot. After the shoot I went back to the new 5D to test it out again and find out what was going on.... then it shot. I AM SOOOOO FRUSTRATED that I have what is supposed to be this great new camera and in the past month I've not even been able to shoot 5 days with it without problems and/or having it in Canon's hands! If anyone has suggestions for what I should do, things I should consider... PLEASE HELP! At least after reading this post I now know that at least I'm not the only one with the metering issues.
  48. Hello, I'm having problems with the metering, at this point I think is a problem with the algorithm in this camera.
    I'm getting lot's of underexposed pictures... even looking at the histogram...
  49. I too have exactly the same problems with my 5DmkII under-exposing 9/10. The frustrating part is that it gets it right sometimes. :)
    I've spoken to Canon and they refuse to believe that there would be such a problem with the camera.
  50. As with you Peter and the last three people who responded I am having the same sort of problems. I shoot AV, center spot focus, evaluative metering and the pics are hardly ever correct at 0+-. I am forced to alway run the exposure compensation up or down without any sort of pattern. Sometimes they are overexposed, sometimes under exposed. Its very very frustrating. 3 other things I have noticed happen to me. One time I turned on the camera and nothing happened. I took the battery out and put it in and then it decided to work. Also my 70-300 is lens just stops working on this body. Don't know why it just doesn't work so I again have to shut off and turn it on. Lastly, when I shoot multiple frames the exposures vary greatly from the first to the last. I posted this problem and some people said it was because of artificial light but it now has happened outside as well. I don't know what is happening here but I have sent it in and if they can't fix it. Its going back for good! I want my money back. My 20d never had any of these problems. JW

Share This Page