400D Issue

[snappd]

Hi all, my first post here after doing a bit of reading on the site & so far most of my questions have been answered without having to say a word! Anyway I have had my 400D for a little over 1yr....just out of warranty & I have realized that it is underexposing by at least 1 stop. Has anyone else had this experience? Is it a common problem, as i know 2 other people with 400D's that are performing the same way? I am not sure if I should approach Canon & what if anything they will do about it.
Appreciate your thoughts....

Btw This was a test conducted at Brisbane college of Photography.

[gstark]

Hi, and welcome.

This is a common problem in Brisbane. What happens is that as one moves closer to the equator the sunlight is stronger, and thus there's a tendency for images to be overexposed. As you move further south, this tendency decreases, and it's about right as you reach Eden on the NSW far south coast. Images made in Hobart are, as you might surmise, underexposed by about 1.3 stops.



Seriously, on what basisi are you determining that the images are overexposed? By viewing the image in the LCD on the back? On a calibrated monitor?

Could you please post a couple of images that you think are overexposed, along with their associated histograms?

[rmp]

What metering are you using? Not sure what the options are on the 400D, but if the metering is set to weight the centre and the centre is not indicative of the whole image, that could lead to overexposure.

Anyway as Gary says, post images -- but also with the settings used, namely shutter speed, ISO, aperture, shooting mode, metering mode and exposure compensation. Don't retouch the images.

[snappd]

Very funny......but I actually said it is underexposing. The test was done using a white wall, using the manual setting, finding 'correct exposure' taking the shot & then checking the histogram....which consistently showed that the shot was underexposed. It wasn't until I deliberately set the exposure meter to 1 stop over that the histogram was bang in the middle. From memory the camera was set to evaluative metering, ISO 100, Auto WB, I used a variety of shutter speed & aperture combos. I will endeavour to post the images when I get a bit more time.

[gstark]

I will endeavour to post the images when I get a bit more time.

Please do.

The histogram does not need to be in the middle, btw. It depends upon the content of the images that you're shooting: a high key image should correctly tend towards the rh side of your histogram, and pushing it towards the centre is not the correct course of action.

Posting your images, and the associated data and histogram is the best way we can assess what you're seeing and help you.

[rmp]

Sorry, I meant underexposure not overexposure. The 'wrong' exposure at any rate.

[snappd]

I understand the concept that 'high key' will be overexposed & 'low key' underexposed on the histogram, but if you were after a 'middle grey' so to speak shouldn't the histogram be in the middle if you set the exposure meter to '0' (needle in the middle).
If I posted the photos all you would see is a varying degree of grey pictures....not sure how to add metadata or 'associated histograms'? is there a quick way using cs3 or acdsee?

[rmp]

In ACDSee go to Properties, select EXIF, right click, Copy All, paste in here under the image. For histogram I don't know a better way than taking a screendump and cropping it.

[gstark]

but if you were after a 'middle grey' so to speak shouldn't the histogram be in the middle if you set the exposure meter to '0' (needle in the middle).

Perhaps.

Perhaps not.

If you shoot a person with very pale skin, where will you be metering from, and how would you expect that to be metered? And rendered in the histogram?

Now ... same questions, for person with dark skin?

Your meter, and your histogram, are merely tools to help you assess the light that's available for you to use. How you choose to use that light is what photography is about.

Consider that you can put a half dozen photographers into a room to photograph the same scene, but each of them will turn in totally different results. Why is that the case?

[snappd]

Ok to save time the histogram is on top of its corresponding photo....the grey bit around the outside is the actual picture. The last picture is the one I overexposed by 1 stop. Ok it's not bang in the middle but do you see what I mean? Am I jumping to the wrong conclusion here? It was the same piece of wall each time just the shutter speed, aperture & focal length were changed, but I kept the exposure on '0' all except the last one.




Filename _MG_0972
File Type JPEG Image
Size (KB) 2,063 KB
Image Dimensions 3888x2592
BPP 24
EXIF
Model Canon EOS 400D DIGITAL
Date/time 2/04/2008 6:53:18 PM
Shutter speed value 1/20 s
Aperture value f/5.6
ISO speed ratings ISO 100
Exposure bias value 0.00 eV
Metering mode Pattern
Flash Flash did not fire, compulsory flash mode
Focal length 70 mm
White balance Auto white balance



Filename _MG_0977
File Type JPEG Image
Size (KB) 1,962 KB
Image Dimensions 3888x2592
BPP 24
EXIF
Model Canon EOS 400D DIGITAL
Date/time 2/04/2008 6:59:04 PM
Shutter speed value 1/20 s
Aperture value f/3.2
ISO speed ratings ISO 100
Exposure bias value 0.00 eV
Metering mode Pattern
Flash Flash did not fire, compulsory flash mode
Focal length 17 mm
White balance Auto white balance



Filename _MG_0979
File Type JPEG Image
Size (KB) 2,053 KB
Image Dimensions 3888x2592
BPP 24
EXIF
Model Canon EOS 400D DIGITAL
Date/time 2/04/2008 7:02:22 PM
Shutter speed value 1/6 s
Aperture value f/3.2
ISO speed ratings ISO 100
Exposure bias value 0.00 eV
Metering mode Pattern
Flash Flash did not fire, compulsory flash mode
Focal length 17 mm
White balance Auto white balance





Filename _MG_0983
File Type JPEG Image
Size (KB) 1,840 KB
Image Dimensions 3888x2592
BPP 24
EXIF
Model Canon EOS 400D DIGITAL
Date/time 2/04/2008 7:09:03 PM
Shutter speed value 1.299998 s
Aperture value f/8
ISO speed ratings ISO 100
Exposure bias value 1.00 eV
Metering mode Pattern
Flash Flash did not fire, compulsory flash mode
Focal length 48 mm
White balance Auto white balance

[gstark]

Am I jumping to the wrong conclusion here?

Probably.

These images are really not very good at all for judging overall exposure. Each of the histograms are well within the range for acceptable exposure, but we have to ask, exposure of .... what?

Where is pure white? Where is pure black?

Where is there a tonal range that we can assess?

Rather than shooting a wall, perhaps go out and shoot some real subjects: a building, a tree, some people ... let's then have a look at those images, and go from there.

[rmp]

Am I jumping to the wrong conclusion here?

Rather than shooting a wall, perhaps go out and shoot some real subjects: a building, a tree, some people ... let's then have a look at those images, and go from there.

Agreed. Don't use any filters, and if possible, maybe get another 400D (or Canon DSLR) along to shoot the same thing with the same settings. And I'm not sure what picture styles the 400D has, but set them back to default if they are changed.

Shoot some subjects with contrast light/dark, eg half a frame of sky, and some all light like clouds, and some all dark (say a building with little sky) etc.

[snappd]

Ok thanks for your advice, I will try that...my friend who I often shoot with also has a 400D & to some degree we have already tested the theory....but will make sure every setting is exact & I will give her my lens as well, as I thought some of our differing results could have been the lenses.....her shots always seem to be brighter than mine (both focusing on a swimmer in the middle of a 25m pool). btw, the test we did with the white wall was done at BCPA as mentioned in my original post & we were told by the tutor that this was an accurate test of a camera's ability to expose correctly.....we looked at how each camera in the class of about 8 performed in this situation & as I said before all 3 400D's in the class had similar results....varying degrees of underexposure, to which the tutor agreed was the case. I am merely going by what I have been told by someone who obviously knows alot more than I do.

[gstark]

I am merely going by what I have been told by someone who obviously knows alot more than I do.

Yep, and I appreciate that.

I've asked a couple of others to poke their heads into this thread to see if they can offer some further advice. In terms of straight exposure, this is not a test that I'm familiar with, nor is it one that I consider to be of great value, but I am (as always) happy to be corrected.

Let's see if we can sort this out for you. We have some truly great people here, and I'm sure their input will be most valuable.

[sirhc55]

I’m a little puzzled by your tutor using a white wall to determine correct exposure. A white wall can be used to evaluate WB but to get correct exposure one would normally use a target gray card such as the one produced by Perfect Pixels.

The mid-tone spike is based on the traditional 18% gray (actually more like 12.5%). If you look at the normal histogram and assume a 100% scale then 18% would be nowhere near the middle of the histogram. The card also incorporates a white scale (for highlights) and a black scale (for shadows). A good exposure would normally have a histogram that indicates no spiking on the left or right hand sides and a peaking that is towards the left hand side of the scale (off centre).

By shooting a white wall to evaluate exposure the histogram will tend to be towards the right hand end of the scale. By bringing the histogram back to the centre you will be underexposing.

[digitor]

Any scene will be rendered by an automatic exposure system at its average value. The histogram will vary considerably, according to the content of the scene.

Any uniform scene (such as a white, or grey wall) will appear on the histogram with a single predominant central spike, at the average value.

I don't have a Canon camera, but my D70 set to aperture or shutter priority will expose a uniform scene (the aforementioned white or grey wall) with the histogram pretty well exactly centered.

Note it doesn't make any difference what reflectance the wall is, if it's white, or black, the meter will try to render it as "middle grey" - whatever that value might be for the meter in question, 12% or 18%.

However, what the 400D represents the centre of the histogram to be, I don't know!

Cheers

[the foto fanatic]

Taking a picture of a white wall is not a good test for your camera's exposure meter, regardless of what your tutor may have told you.

What your camera is going to do is to render all of that white as an 18% grey, thus giving the impression that it is underexposed. In fact, it is underexposed. People often have this problem when they are photographing snow.

Perhaps the test is to see what exposure compensation you need to make the wall white - probably at least 1 stop towards the +ve side. It is almost counter-intuitive in that you need to increase the meter's suggested exposure for a largely white subject.

Any exposure meter is going to have difficulty with this scenario. If you took a picture of a black wall, your image would be overexposed for the same reason. The meter wants to represent that colour as 18% grey. You would need to change your exposure compensation to the -ve side.

I would suggest to you that you should be checking with your tutor to make sure that you have completely understood what he or she said.

[digitor]

What your camera is going to do is to render all of that white as an 18% grey,

Exactly the point that I made above.

Regardless of what the reflectance of the wall is, the meter will attempt to render it as (for the sake of argument) 18% grey. Thus, it is an effective test as to where the meter places that point on the histogram.

Of course the resulting image will not be representative of the actual reflectance of the wall, any shade will be reproduced the same, regardless of whether it is black, grey, or white in reality.

Cheers

[gstark]

A white wall can be used to evaluate WB but to get correct exposure one would normally use a target gray card such as the one produced by Perfect Pixels.

Precisely.

What your camera is going to do is to render all of that white as an 18% grey, thus giving the impression that it is underexposed. In fact, it is underexposed. People often have this problem when they are photographing snow.

Excellent point, Trevor. Thank you.

I would suggest to you that you should be checking with your tutor to make sure that you have completely understood what he or she said.

Good suggestion.

Regardless of what the reflectance of the wall is, the meter will attempt to render it as (for the sake of argument) 18% grey.

Yep.
Something seems to be amiss here, and I don't (yet) think it's within your camera.

[snappd]

What your camera is going to do is to render all of that white as an 18% grey, thus giving the impression that it is underexposed. In fact, it is underexposed. People often have this problem when they are photographing snow.

Yes...giving the 'impression' that it is underexposed when in fact the histogram should show that it is correctly exposed if the camera is metering correctly......to get anything white you would have to overexpose by up to 2 stops.....& that would be reflected in the histogram as being overexposed.

Any uniform scene (such as a white, or grey wall) will appear on the histogram with a single predominant central spike, at the average value.

Note it doesn't make any difference what reflectance the wall is, if it's white, or black, the meter will try to render it as "middle grey" - whatever that value might be for the meter in question, 12% or 18%.

This is exactly what the tutor (in my opinion) was trying to make.....I will of course try to check with the tutor tho & get some further clarification. Logically it makes sense to me that the histogram should be in the middle when it has no 'tonal range' or variables to 'confuse' it.......

[sirhc55]

This might be worth a view:

http://www.luminous-landscape.com/tutorials/understanding-series/understanding-histograms.shtml

[gstark]

Note it doesn't make any difference what reflectance the wall is, if it's white, or black, the meter will try to render it as "middle grey" - whatever that value might be for the meter in question, 12% or 18%.

This is exactly what the tutor (in my opinion) was trying to make.....I will of course try to check with the tutor tho & get some further clarification. Logically it makes sense to me that the histogram should be in the middle when it has no 'tonal range' or variables to 'confuse' it.......

Well, no. The point here is that is is already "confused". A camera meter will always try to render a scene as if it's 18% grey. Always: that is exactly how it's programmed, because that is how a so-called average/typical photo often ends up.

So, if your subject is white ... how the hell does your camera know this?

The fact is that it doesn't, and the fact is that it's expecting to see middle grey. Thus, it sees white (consider Trevor's excellent snow scene analogy) and makes a decision that this is a brightly illuminated grey scene, rather than a normally illuminated white scene.

That's a very big difference, and I suspect that it may well be the key to this discussion.

Metering has therefore been done according to that (incorrect) assumption - I said the camera was already confused - and as a result, when it takes the PoV that this is a brightly lit grey subject, it must give you underexposed images because it's compensating for a non-existent problem.

As I've already pointed out, the camera's meter is but one of a number of tools that we use in photography, but it is the job of the photographer to learn and understand not just how to use those tools, but also to understand when and under what circumstances those tools might show some weaknesses.

This is exactly such a time, and, as the photographer, it becomes your job to recognise those times when you may need to override the meter's suggestion. Why do you think camera manufacturers provide us with the ability to override the cameras' meters?

Getting back to the reading of the histogram for a moment, it's actually logical that a histogram of a correctly exposed image of a white wall should be tending well towards, if not falling right off, the rh side of the histogram, because (a) the wall is white, and (b) the rh side of the histogram is where the whites live.

As Chris says, you don't really want it falling off at either end, but a correctly exposed image doesn't mean that you're looking for a perfect bell curve either. Correct reading of a histogram needs to take into account the subject matter included in the image; lighter coloured (or brighter) subjects will pull it towards the rh side, darker subjects will pull it towards the left.

So ... don't blindly accept the meter reading as gospel: look at the scene, and your subject; consider its brightness (or lack of) and use the meter as a starting point from which the correct exposure - which is a choice that YOU make - can be determined. Then use the histogram to review how it's all working for you. If it's not falling off either end, and is basically within the boundaries of the range given, then you're pretty close to correct.

Look for shadow detail ... what level of detail have you managed to retain in the darkest areas of the image. Look for blown highlights - they're not a good thing. Remember that your sensor might not be able to record the full dynamic range that you're able to see; technology does have its limits, and you may need to select a compromise point that gives you the results you're seeking, but is still not "perfect".

[Biggzie]

Well said Gary .... Ive worked on the theory that correct exposure is what I want, not what the camera tells me I want.

[the foto fanatic]

Thom Hogan has written a quite informative piece about the histogram in relation to a grey card on his web site:

http://www.bythom.com/graycards.htm

In summary, he says that modern camera meters aren't calibrated to 18% grey, they instead are set to 12% grey. He further says that this is the reason that the histogram of a grey card would produce a peak to the left of centre.

[digitor]

So, if your subject is white ... how the hell does your camera know this?

It doesn't - and that is the crux of the matter. It will render the scene at the meter's "calibrated" average point - which is why the test will show you where that point is displayed on the histogram. For my D70, that point is pretty well exactly in the centre.

You will never get a "correct exposure" (whatever that is) of the wall, be it white or black, but that is not the point of the test. The point is to show what the meter thinks is an average exposure value.

Cheers

[snappd]

This is exactly such a time, and, as the photographer, it becomes your job to recognise those times when you may need to override the meter's suggestion.

Yes Gary, I agree with you here....I do this quite often to get the results I want, not what the camera thinks is correct. I understand & agree with all that you have said.....but I am not talking about general shooting conditions......just this test....& after reading that link above from Trevor (thank you Trevor) I am tending to think that maybe the 400D isn't calibrated to 18%.

So, if your subject is white ... how the hell does your camera know this?

It doesn't - and that is the crux of the matter. It will render the scene at the meter's "calibrated" average point - which is why the test will show you where that point is displayed on the histogram. For my D70, that point is pretty well exactly in the centre.

You will never get a "correct exposure" (whatever that is) of the wall, be it white or black, but that is not the point of the test. The point is to show what the meter thinks is an average exposure value.

Cheers

Thank you Digitor.

[the foto fanatic]

You might also find this thread on Photo.net an interesting read (about exposure and histograms on Canon bodies):

http://photo.net/bboard/q-and-a-fetch-msg?msg_id=00L7Vz