550D exposure problems with spot

Hello,

I was also wondering if anybody had problems with their exposure spot metering with a 550D.
I consistently got different results when the camera is turned in portrait mode.

When turned from normal landscape counterclockwise it overexposed, when turned clockwise, it exposed fine, even if the subject was symmetric between left and right side.

Anybody seen anything like that?

Here is an example, the difference between the two shots is only the camera turned by 180 degrees. This is using spot metering on the top of the tree.

Did anybody see such issues with their camera?

mbrowne wrote:

That is a poor scene to test the effect. Try something with more even illumination.

many other shots/subjects had the same issue - and my 400D with the same subject was fine.
also, many focal lengths, 10, 22, 18, 55, 100, (3 different lenses).
and no CPL filter used. (again, 400D with same optics fine).

Just wondering if there is something strange about metering I don't understand.

BAK wrote:

I do not understand why you'd use spot metring for this, but anyway, it certainly seems a stange result to me, too.

I had strange exposure results with all kinds of subjects, so I started to try to take more controlled shots, checking out also different metering modes (this difference also happened with central weight, and other metering modes, not just spot).

However, I reasoned that with spot it definitively shouldn't be there, since the metering should only be a circular area in the center (according to Canon brochure).

If the camera is pointed at a plain grey wall, like the side of a barn, with no sky in the shot, and you turn the camera, does the exposure change?

I was also shooting buildings, same thing happened - but then one could have argued that the background was not even enough - So, hence I choose the above subject.

But you are right, I was thinking I just should shoot the sky and turn the camera slowly and see what happens. That's why I also stated that I didn't use a polarizer. One could almost think the exposure meter is measuring through a polarizer, and getting different readings based on orientation.

I get the camera back in a few days, I had sent it to Canon, it generally overexposed anyway. I did state the problems and gave sample images. The repair report that I have already says "calibration done" (no repair). I can see that you can calibrate the general overexposure, but I can't see how one can calibrate the above issue. It looks more to me like the spot meter doesn't really work.

Anyway, that's why I started to post if others have seen similar issues.
Or if there is some obscure point one has to be aware of.

If you just turn the camera on and point it at the ceiling and reading the setting int eh viewfinder and then turn the camera sideways, do the settings change by a couple of stops?

Theory:

I don't have the book in front of me right now, but I seem to recall spot metering uses about 3.5 percent of the pixels located in the center of the sensor to determine exposure, and about 9 percent for center-weighted. Let's suppose it is possible the camera is actually using a sensor area off center (up or down) in portrait orientation. In one orientation you are going to spot on the sky, and at 180 degrees you are spotting on the trees - all the while keeping the same composition.

Perhaps Canon has 'calibrated' the spot back to the center of the sensor where it belongs.

Just a theory.

Hope you follow through with us when you get your camera back.

wave01 wrote:

Hi what mode were you when you took the picture as it shows aperture and exposure differences in the picture.

It was shot in P mode. So I let camera choose aperture/shutter. Overall there were 2 stops difference at least between the two shots when I looked at Exif data later.

pointshooter wrote:

Theory:

I don't have the book in front of me right now, but I seem to recall spot metering uses about 3.5 percent of the pixels located in the center of the sensor to determine exposure, and about 9 percent for center-weighted. Let's suppose it is possible the camera is actually using a sensor area off center (up or down) in portrait orientation. In one orientation you are going to spot on the sky, and at 180 degrees you are spotting on the trees - all the while keeping the same composition.

Yes, this would be a possibility. I also was thinking whether it still behaves like a center weighted situation, - with the camera thinking that the bottom of a horizontal frame is darker, and the top (sky) is brighter, which could give different results in Portrait orientation. But that's why I tried to have equal sky left and right here.

Based on the 550D review here, I was sort of the impression that spot is really only center:
http://www.dpreview.com/reviews/canoneos550d/page10.asp
so, such horizontal top/bottom stuff should go into this equation.

My old 400D with partial metering was pretty equal in the two different portrait modes of this subject. (live view was also fine).

Perhaps Canon has 'calibrated' the spot back to the center of the sensor where it belongs.

Just a theory.

Hope you follow through with us when you get your camera back.

Yup, I sure will test the camera seriously when it gets back.

This is the second time, just before it was in for a repair of the green display under the viewfinder - lighting up about half of the segments permanently. If that is some kind of liquid crystal display and the fluid leaked out over AE area, maybe that caused this anomaly?

-- hide signature --

Editor Bob

Camera was set to spot metering and four shots taken, all in portrait orientation. Subject is inside wall of a screened porch.

Top two are with the camera grip side up and bottom two with grip side down. Red dots show the center exposure spot. Left two shots spotted on the dark wall just below the screen; right two shots spotted just above the wall on the screen.

Camera was set to Tv, 1/4 sec and ISO to 3200 to permit hand-held shots. The posted composite has been reduced to 25% of original resolution, no cropping.

Aperture automatically adjusted to provide exposure and was consistent, regardless of camera orientation. In this case, the change in exposure was 1 stop - probably would have been more if I'd spotted differently.

Additional tests were made in P mode and landscape orientation. P mode varied the shutter speed most often, but I did notice a couple changes in aperture.

photonius wrote:

Yes, this would be a possibility. I also was thinking whether it still behaves like a center weighted situation, - with the camera thinking that the bottom of a horizontal frame is darker, and the top (sky) is brighter, which could give different results in Portrait orientation. But that's why I tried to have equal sky left and right here.

Based on the 550D review here, I was sort of the impression that spot is really only center:
http://www.dpreview.com/reviews/canoneos550d/page10.asp
so, such horizontal top/bottom stuff should go into this equation.

I have David Busch's book on the T2i (well worth the cost). He says the camera tries to make an "educated guess" about what kind of picture you are taking in evaluative mode, and indicates it will also use information from the rest of the frame in center-weighted mode. For example, lighter 'top' would indicate a landscape. He doesn't hazard a guess what happens to the 'top' when you turn the camera 90 degrees. In fact, I can't find much sage advice on portrait orientation anywhere. I would suspect it accounts for nearly half of my shots.

This is the second time, just before it was in for a repair of the green display under the viewfinder - lighting up about half of the segments permanently. If that is some kind of liquid crystal display and the fluid leaked out over AE area, maybe that caused this anomaly?

Really strange. Sorry to hear you are having all the problems with the camera. Hope they take care of you and make things right. It's a great model when all is working.

pointshooter wrote:

Camera was set to spot metering and four shots taken, all in portrait orientation. Subject is inside wall of a screened porch.

Top two are with the camera grip side up and bottom two with grip side down. Red dots show the center exposure spot. Left two shots spotted on the dark wall just below the screen; right two shots spotted just above the wall on the screen.

Camera was set to Tv, 1/4 sec and ISO to 3200 to permit hand-held shots. The posted composite has been reduced to 25% of original resolution, no cropping.

Aperture automatically adjusted to provide exposure and was consistent, regardless of camera orientation. In this case, the change in exposure was 1 stop - probably would have been more if I'd spotted differently.

Additional tests were made in P mode and landscape orientation. P mode varied the shutter speed most often, but I did notice a couple changes in aperture.

Thanks for this test. This is much more like I expect a spot to work.
I will try to see if I get the same results when the camera arrives.

-- hide signature --

Editor Bob

pointshooter wrote:

photonius wrote:

Yes, this would be a possibility. I also was thinking whether it still behaves like a center weighted situation, - with the camera thinking that the bottom of a horizontal frame is darker, and the top (sky) is brighter, which could give different results in Portrait orientation. But that's why I tried to have equal sky left and right here.

Based on the 550D review here, I was sort of the impression that spot is really only center:
http://www.dpreview.com/reviews/canoneos550d/page10.asp
so, such horizontal top/bottom stuff should go into this equation.

I have David Busch's book on the T2i (well worth the cost). He says the camera tries to make an "educated guess" about what kind of picture you are taking in evaluative mode, and indicates it will also use information from the rest of the frame in center-weighted mode. For example, lighter 'top' would indicate a landscape. He doesn't hazard a guess what happens to the 'top' when you turn the camera 90 degrees. In fact, I can't find much sage advice on portrait orientation anywhere. I would suspect it accounts for nearly half of my shots.

Yes, with the (old) 400D I didn't have to bother much, but now I recently used the 550D on vacation, and I was switching a lot between landscape and portrait. It drove me nuts because I had to constantly dial in corrections from -3 to +1. Later I figured out some of it was due to this very big exposure difference between the darker portrait orientation and landscape - landscape also overexposed like the overexposed portrait mode. Initially I figured it was like you said that information is also taken from the rest of the frame as you stated above, but the more I tried later - in particular then with spot mode - , the more it didn't make sense.

This is the second time, just before it was in for a repair of the green display under the viewfinder - lighting up about half of the segments permanently. If that is some kind of liquid crystal display and the fluid leaked out over AE area, maybe that caused this anomaly?

Really strange. Sorry to hear you are having all the problems with the camera. Hope they take care of you and make things right. It's a great model when all is working.

Yup, I hope so. Thanks for helping.

-- hide signature --

Editor Bob

I have now tested the "fixed" and returned camera some more.

Alas, not completely unexpected, the problem is still there. At least some of the overexposure issue has been corrected, and the spot does seem to work more as expected, but I still got weird results.
Anyway, I was able to track it further down.
It looks like the spot center is substantially off.

Again, to very differently exposed pictures, depending on the orientation of the camera.
This can be explained by the center being off to the right.

+++++++++++++

A series of pictures scanning across a lamp shows the effect quite clearly.
Screenshots, with AF point shown.

Here, in this 1. shot the spot sensor behaves as if it is already in the range of the bright light.

Here, event though the center is still in the bright light, the spot sensor picks up the dark area to the right and exposes brighter.

It is way too much off center (more than the circle in the viewfinder), that can't be according to spec??

photonius wrote:

It is way too much off center (more than the circle in the viewfinder), that can't be according to spec??

I see what you mean from those photos. I would like to have that fixed if it were mine.

Further testing showed that the center of AE Spot sensitivity is towards the right.

Looking at the viewfinder diagram of the 7D here at dpreview

, it looks like the 63 field AE sensor sits above the eyepiece, and looks down through a small optik on the projected image. Slight decentering seems therefore easily possible.

However, I've been informed by Canon service that this cannot be adjusted mechanically, but they do adjust it electronically. I guess it means using another of the 63 AE sensor fields as the center. Of course it would imply then that the 63 sensors overall would not lie centrally on the image, which may impact center weighted, or evaluative modes perhaps?

I tried to see if I could detect uneven cut-offs on the left edge versus the right, but it wasn't conclusive, my set-up was too simple.

Very sorry to hear you are still having problems. I think I see your point with the series of light shots.

I have always been confused by the claim that exposure priority is given to the active auto focus point(s). Does this override spot and center-weighted metering? Maybe this is only the case for evaluative metering. Perhaps your camera is giving exposure priority to one of the AF points on the right - regardless of setting. A couple things you might try.

1. Redo the light sequence with the extreme left AF point selected and spot metering.
2. Redo the light sequence in manual focus and spot metering.

3. Check the focus of the center and right AF points to see if they are spot on according to the camera settings.

I guess there will always be frustrations. Regarding overexposure, I routinely shoot at EV-1 on outdoor landscapes; and, I have learned not to expect any of my cameras (A590, G11, or T2i) to pick up cloudy or shady conditions with white balance set to AUTO. Maybe that's why they include DPP.
--
Editor Bob

pointshooter wrote:

Very sorry to hear you are still having problems. I think I see your point with the series of light shots.

I have always been confused by the claim that exposure priority is given to the active auto focus point(s). Does this override spot and center-weighted metering? Maybe this is only the case for evaluative metering. Perhaps your camera is giving exposure priority to one of the AF points on the right - regardless of setting. A couple things you might try.

Thanks for the reply. Actually, playing around with this, I learned some new things that partially answer some of the questions you pose.

A: Spot metering - (with center point AF).

I can focus on something with center point AF, and when I move the camera, exposure adjusts according to where the AE spot points at (in my case it's off to the right).

In fact, you can follow it easily in the viewfinder, just darken your room, have a single light source, and scan the camera across the light source. Set fixed ISO and aperture, and you will see how exposure time varies as the AE spot passes across the light source.

• You can turn off AF, go into MF, and the Spot AE metering is the same.

I haven't tested partial metering the same detailed way, but I think it's the same, the area of AE sensitivity is just larger.

• Center weighted measuring seems to work the same way too. You can scan the camera across the light source, the exposure changes, with strongest sensitivity in the center (well, not in my case).

B) Evaluative metering (one shot AF).

Here indeed the AE and AF seems to be linked/locked. After AF on a point, I can move the camera and the exposure does not change.

I have to verify that, as it is not entirely expected. Also, what happens with MF?

Regarding overexposure, my 550D had a quite bad case of it, like in your case I needed about -1 EV outside. I think that's not according to spec.

The overexposure was improved after the camera returned, but the wrong center wasn't fixed.

I have seen others mention dialing in -1 EV on a 550D or similar (unless it was you). I can't help that the AE sensor is not well calibrated on the 550D. The 63 AE sensor module was new on that Rebel body.

As to AWB, it seems ok to me. Color temperature is not the same with different conditions (sunny, cloudy), so it is normal to me if the shots don't have the same WB, because in reality they don't have either.

1. Redo the light sequence with the extreme left AF point selected and spot metering.

They were done with spot metering and center AF.

2. Redo the light sequence in manual focus and spot metering.

3. Check the focus of the center and right AF points to see if they are spot on according to the camera settings.

I guess there will always be frustrations. Regarding overexposure, I routinely shoot at EV-1 on outdoor landscapes; and, I have learned not to expect any of my cameras (A590, G11, or T2i) to pick up cloudy or shady conditions with white balance set to AUTO. Maybe that's why they include DPP.
--
Editor Bob

The camera has come back now, and it works as expected, i.e. like samples others have posted.

There are some lessons:

• Apparently it is possible to adjust the center of the AF (specially important for partial/spot) electronically. But I have no idea how far of the physical setup above the viewfinder was - perhaps the electronics was already off.

• Users should test their AF system, not just focusing system. Obviously it can be off and cause like in my case substantial inconvenience, i.e. constantly unexpected exposure variations. Use a small strong light source, and scan your camera across it.

• In Single shot mode with center AF point (that's what I use):

with Partial or Spot, after half pressing the shutter button, and then moving the camera, the exposure will continuously adjust to whatever is under the spot meter.
I.e. exposure evaluation and AF point evaluation will be different.

In evaluative mode, exposure is linked to the AF points. After pressing the shutter half-way, both AF and exposure stay fixed when moving the camera.

Evaluative metering with multiple AF points seems like pot-luck, it could choose any of the red lit up AF points, if I understand this by Chuck Westfall correclty
http://www.openphotographyforums.com/forums/showthread.php?t=8310

The evaluative metering with single point center AF weighs center, but also gives other areas more consideration, i.e. the center is not weighed as strong as in spot or partial metering.

Perhaps time to go back to center weight average, like on the good old EF and A1 cameras ...

I came across some comments when using lenses that have no chip, like Samyang lenses. It was said that there is a lens exposure correction curve, in particular for fast lenses. A quick test with the Samyang 8mm, which is not a fast lens anyway, shows that it exposes a little darker at f3.5, and gets brighter when closing the aperture, although it's pretty flat after f5.6.

If this is due to aperture problems of the Samyang or has to do with the curve, I don't know at this point. But good to keep in mind.