Metering Question

[Sageg]

~first, please excuse the picture. It was taken to test metering mode, not for any other purpose~

I've been testing out the various metering modes to get a better understanding and improve my exposure. The attached picture was taken using partial metering (I don't have spot), as I understand that this is the best with backlit scenes.

Now when I look at the picture, it appears to me as if the middle 9.5% of the screen, which is all the meter would look at, is focused on the girl only. However, the girl is underexposed, indicating that the very bright light did influence the meter.

For this experiment, I don't mind that there was spill on the hair as I am only looking at the metering. I know the colour is off (yellow room caused reflections).

Is my understanding of how this metering mode works completely off base? Am I missing something?

Other settings: 1/125, f2.8, ISO 400

Thanks for your help.

Permission given to edit this photo to help explain what went wrong

[malcolmp]

I also find that the background light sometimes spills into the metering region, and it doesn't matter a huge amount which setting the metering is on. I find the most reliable way is to set the camera to Av or Tv, take a shot, look at the histogram, and then adjust exposure compensation to adjust the picture. Takes a bit of practice but it will give you more control.

The other "cheating" way is to fill the frame with the subject without backlight, press the AE Lock button and then recompose your image. The metering will be set to the subject only.

BTW you can easily improve the image using shadow/highlight in photoshop. It doesn't look like too much has been lost, just shifted.

Malcolm

[PacAce]

Could it be it's because you metered off of the little girl's face which is light to begin with? By how much was the picture underexposed? A stop or less?

[SkipD]

This is a difficult scene to meter with ANY meter, much less one built into a camera. Expect - with any meter - to have to use your head in addition to the meter and force the camera to do something different than its first decision from time to time.

The method to achieve the corrected exposure would either to use an automated mode (P, Tv, or Av) and induce some Exposure Compensation to tweak it, or to use M mode and set the aperture and shutter speed yourself (based on the readings and what you know the adjustment should be).

[Sageg]

Thanks--I appreciate you all taking the time to respond.

So partial metering is still susceptible to spill. That's good to know. I usually do shoot in manual & usually expose to the right, but I am trying to nail down my understanding of metering modes.

She is about 1 stop underexposed.

Again, thanks.

[PacAce]

Thanks--I appreciate you all taking the time to respond.

So partial metering is still susceptible to spill. That's good to know. I usually do shoot in manual & usually expose to the right, but I am trying to nail down my understanding of metering modes.

She is about 1 stop underexposed.

Again, thanks.
I wouldn't say that partial metering is susceptible to spills unless the bright light source is very near the fringes of the partial metering area (which is not the case in your picture). Partial meterinal only meters what's in the partial metering area. What you do have to considier is the reflectivity of the area that you are metering. If you meter something that is light in color, you will need EC to increase the exposure so that the image is not underexposed. Similarly, if the area within the circle is dark, you will need to use EC to decrease the exposure to get the right exposure.

[Sageg]

Totally understand now. The partial metering did its job perfectly. It was just the natural lightness of her skin that caused the underexposure. Thanks! I'm going to go back to using the histogram to check exposure. :)

[Robert_Lay]

There is one aspect of this that does not seem to have been covered properly - at least not in terms that I am familiar with.

I see reference to "spill", but I'm not sure you refer to the mechanisms that I think are at fault here.

There were two separate issues at work here to cause your picture to be underexposed. One of them may be your "spill" effect - whatever that is. However, what I see is forward scatter. I am not talking about the direct effects of the window light on the girls face - rather, I'm referring to the effect of the direct rays of light coming into the lens and forward scattering throughout the image causing the captured image to lose contrast. In addition to that you have the same effect on the metering. The metering is seeing light that is not from the objects in the image but rather light striking the lens directly and resulting in forward scattered light. That effect (which may be what you are calling "spill" will generally cause the meter to register more light than is actually coming from the scene - thus causing underexposure.

Correction for that is to shield the lens from direct rays from a light source (hood, or whatever it takes).

The second issue is the one of metering facial tones. Facial tones have always been described as Zone VI, whereas your meter is expecting Zone V. That right there says you are going to be 1 stop underexposed if you meter off of caucasion facial tones.

The correction for that is that when you are metering facial tones and you want to place them on Zone VI then you have to dial in a +1 stop Exposure Correction.

[Sageg]

Thanks Bob! That makes a lot of sense. I do use a cap outside, but didn't even think of using one inside. I guess subconsciously I was thinking it would make the shot too dark, which it probably would if i was facing away from the light source, right? But as I was facing into the light source, I'm dealing with the same issue as shooting into the sun.

I've also got to be more diligent about understanding the zone system. I have a basic understanding, but don't always remember when shooting. What I usually do is try and expose to get my histogram as much to the right as possible without blocking the highlights, which is obviously a challenge in this shot because of the backlighting. But closing in tighter on her face, adjusting to compensate for her lighter skin & using a lens cap would have helped deal with the problems in this shot.

Again, thanks for your help.

[Wilt]

Bob is on target...I was about to write the same thing, until I read his post!
Reflected light meters try to render what they see at 18% grey (actually, to be precise, 12% grey). Caucasoid skin is brighter than a grey card, so metering skin of a 'white' will result in inherent underexposure, and depending upon time of year and how tanned they are from proplonged sitting in the sunlight they could be brighter or darker than that.

[nevilleb]

Most digital cameras tend to underexpose a tad bit. This is to prevent the highlights from getting blown out.

Of course, spillage can affect metering too.

I trust you did dial in positive EC? If not, it explains why the image is underexposed.

nevilleb

[Sageg]

No, I didn't add any EC. I'm actually taking a course, and the prof keeps on telling us how accurate the meter is & how digicams don't underexpose, despite all I've read to the contrary.

[DavidW]

That flies in the face of what most people, including Canon themselves, say.


The meter attempts to meter for 18% grey. If you're shooting a dark scene, the meter will tend to overexpose, as it pulls the darkness up to midtones, so negative EC is needed. If you're shooting a light scene, the meter will tend to underexpose, as it pulls the highlights down to midtones, so positive EC is needed.

There are also times when evaluative metering is fooled; the use of partial or spot metering (with appropriate EC if you didn't meter on a midtone) can help there. There's other tricks to do with metering when you've got more than about five stops dynamic range - but this isn't an attempt to write a book.


The meter, after a fashion, has some intelligence (in evaluative mode, anyway), but it can be fooled. Understanding how the meter works and checking the histogram and "blinkies" is the right way to expose - not slavishly following the meter. Even though shooting RAW allows you to shift the exposure in either direction, you are best off starting with the correct exposure in the first place.



David

[PacAce]

No, I didn't add any EC. I'm actually taking a course, and the prof keeps on telling us how accurate the meter is & how digicams don't underexpose, despite all I've read to the contrary.
Yes, that's ver true.....that is, if you're shooting a 12% gray card or a scene that averages out to the same reflectivity as a 12% gray card. :lol: Otherwise, as was already pointed out previously, the picture will come out under- or overexposed unless EC is used to compensate.

[PhotosGuy]

Otherwise, as was already pointed out previously, the picture will come out under- or overexposed unless EC is used to compensate. ...or you learn to shoot on "M".

[PacAce]

...or you learn to shoot on "M".
Same difference. You still have to compensate from what the meter tells you is the "right" exposure although, in the case of "M", you're not really using EC but the Av and Tv dials.

[Sageg]

...or you learn to shoot on "M".

I'm most comfortable in "M". I find that when I switch over to AV or TV, I keep on accidentally adjusting the EV, when I think I'm adjusting something else. :LOL: And the histogram is my friend! (Again, the prof doesn't like using the histogram. He also prefers to underexpose EVERYTHING then adjust exposure in raw. Not sure if I am getting my $$s worth. :( )

[DocFrankenstein]

I learned to approximate the exposure compensation to gain correct values.

Caucasian faces should be about +1
Anything which you want to come out almost white without loss of detail: +2
Concrete is approximately 18% gray, so no EC compensation is required.
Palm of your hand is always EC +1

I only meter for shadows when I shoot film, so I don't use EC with digital.

[Robert_Lay]

I'm most comfortable in "M". I find that when I switch over to AV or TV, I keep on accidentally adjusting the EV, when I think I'm adjusting something else. :LOL: And the histogram is my friend! (Again, the prof doesn't like using the histogram. He also prefers to underexpose EVERYTHING then adjust exposure in raw. Not sure if I am getting my $$s worth. :( )

I hope you misunderstood your prof. No one - neither in film nor in digital, should be underexposing and trying to make up for it in processing, except for one case that I can think of. When shooting an exceptionally high contrast scene with film, where you have an extended range of values, and you have to not only underexpose to some extent, you have to under-develop, as well.

The general rule for B&W film has always to expose for the shadows and develop for the highlights. With digital, you "expose to the right". If your prof doesn't understand that, refer him to Bruce Fraser's book on Adoble Camera RAW, or the more recent book by Fraser and Blatner on PSCS2.

[Sageg]

You tell me:

The third line of defense in protecting blown out highlights is to habitually underexpose your pictures...I underexpose almost all my images by –2/3 f/stops.

[DocFrankenstein]

You tell me:
Maybe he'd underexposes intelligently? (using his brain too) I do something similar to prevent blowing out a color channel.

[Sageg]

He is an amazing photographer, so his technique works for him.

[Robert_Lay]

You tell me:

Could you explain from which reference and what page your quote is from? I'd like to see the context, because I don't remember reading that.

My reference from Fraser's Camera RAW with Adobe Photoshop CS is in the last paragraph of the sidebar on page 32, where he says, " ...preferable to err on the side of slight overexposure rather than underexposure, because underexposing to hold the highlights will make your shadows noisier than they need be."

In the book by Blatner and Fraser, he says essentially the same thing on page 580.

[Sageg]

Sorry...rushed too quickly. The quote is from my prof. And, as you say, is the exact opposite of what Fraser saw in his book.

Anyhow, I don't think I misinterpreted the prof.

[DocFrankenstein]

Blowing out of an individual color channel will result in a flat image, which you see so often. Especially the red channel in faces.