Technical metering thread

The people in the "Understanding your camera's built-in metering" thread (here) have asked me to put technical questions and discussion elsewhere.

This thread is that elsewhere.

To start it off, I'll reply to one of the messages in that thread.

RichNY wrote in post #3483907
TMR is correct concerning my post. Shooting in Manual wouldn't be an issue when learning exposure if the Exif data showed you how much above/below the center line you shot your image at. (the amount +/- the center line is the equivalent to EC). The end result is that when you look at your Exif you have no idea if you shot using the camera's suggested metering, above it, below it, and by how much.

This isn't quite true, but it's close. The data is there but it takes a little work to use it.

In the "Maker Notes" section of the EXIF data is an entry for "Measured EV". This is the EV the meter read from the scene. It's present for the 30D and, I expect, for earlier models as well, but I can't verify that. This is present in the jpeg image generated by the camera, and probably the raw image as well (I haven't looked for the latter).

Unfortunately, the "Maker Notes" information isn't guaranteed to be transferred to postprocessed images. Rawshooter Essentials, for instance, seems to drop it on the floor. So you may have to go back to your originals to see it, and it's even possible that you'll have to set your camera to store RAW+jpeg.


Okay, so...how do you use it?

Well...EV is defined as log2(N^2/t) where "log2" means "base 2 logarithm", N is the f-number, and t is the exposure time in seconds. If you can't directly compute a base 2 logarithm, you can do so by computing log(x)/log(2). Source for all this: Wikipedia.

The camera's meter seems to be yielding the EV based on ISO 4. A bit odd, but the math does appear to work out. So if you have the EV from the EXIF and the aperture and you want to compute the shutter speed the camera would have used (if you hadn't been in manual mode), you can use this formula:

s = N^2/(2^(EV + log2(ISO/100) + 4.644))

Similarly, if you have the EV and the shutter and want to compute the aperture, you can use this formula:

N = sqrt(s*2^(EV + log2(ISO/100) + 4.644))

Where did the magic number 4.644 come from? It's log2(100) - log2(4). In other words, it's basically the number of stops between ISO 100 and ISO 4 (which appears to be the ISO the camera's EV measurement is based on). And log2(ISO/100) is the number of stops difference between ISO 100 and the ISO you're using.

It's still probably going to be easier for most people to just use Av or Tv + exposure compensation than to go through the trouble to deal with the above, but it's there for anyone with the determination to use it.

I'll also ask a couple of questions:

1. How exactly does evaluative metering mode work, and how exactly does it differ from center-weighted average?
   
2. What's the weighting distribution used in center-weighted average? In other words, how much emphasis is given to the center relative to the rest of the frame?
   
3. If evaluative mode is, as I believe, a generic term for "automagically figure out the best exposure", then wouldn't it change as technology improves (e.g., by making use of an exposure database or something)? If you're having to use EC when using evaluative mode, wouldn't you be better off using something a bit more stable, like center-weighted average?

kcbrown wrote in post #3486916
The people in the "Understanding your camera's built-in metering" thread (here) have asked me to put technical questions and discussion elsewhere.

We didn't ask you to put technical questions elsewhere, we asked you to stop posting irrelevant comments that detracted from the other thread over there

RichNY wrote in post #3463659

kcbrown wrote in post #3463616
How do you determine ahead of time whether a given scene will exceed the dynamic range of your film, or do you just have to shoot it and look at the results later? I'd think there'd be some way of telling, since the film guys didn't have the option of checking the results right then and there.

And I'd think the method would be to point the spot meter at various places that appear to be the lightest parts of the scene, and then do the same for the darkest parts of the scene, and see how much of a difference there is.


Sorta implies you have to use spot metering for the initial evaluation of most scenes unless you Just Know that the scene fits within the dynamic range of the film/sensor.

That would work. But you make a big jump to thinking this implies that using the other metering modes would not also give you the same difference in range.

The other metering modes can't give you that information, at least not precisely. It'll involve guesswork for anything other than spot metering. They might be able to get you close depending on how much you can zoom in on any given point in the scene, but that's basically the equivalent of spot metering anyway.

The reason is that all the other modes perform some averaging of a part of the scene (technically, so does spot but the area it averages is far smaller than the rest, so it's as close as it gets to "pure" spot metering). In any part of the scene that involves variations in reflected light intensity, the average intensity will by definition lie between the extremes of the light intensity.

But in order to know the dynamic range of the scene, you have to know the two extremes (light and shadow) of the light intensities within the scene. Since averaging by definition gives you a value in between the two for any part of the scene you're looking at, it follows that no averaging mode will give you the endpoints. They can't, by definition.

I could mathematically prove this if you really want, but I think you get my point.


So: the implication is that only spot metering will tell you with accuracy the dynamic range of the scene. Anything else is just a guess. An educated guess based on many years of experience, perhaps, but still a guess.

How much accuracy you get for the dynamic range from an averaging mode depends on how the light is distributed throughout the scene. If there are large, unbroken areas of extreme light and shadow, then you'll get a more accurate idea of the dynamic range from an averaging mode than you would if the areas were small.

An extreme example is: suppose you're taking a picture of the sunlight filtering through the trees. The transitions from darkness to light are harsh and the areas involved are small. No averaging mode will directly tell you what the intensity extremes are in that case, because the areas of the extremes are too small. In other words, you have to guess.

RichNY wrote in post #3487027
We didn't ask you to put technical questions elsewhere, we asked you to stop posting irrelevant comments that detracted from the other thread over there

I've deleted most, if not all, of those comments, so hopefully the thread reads a bit better and is easier to follow. If you have any responses to said deleted messages, please feel free to delete them as well unless you believe they'll be useful to the readers of that thread.

Why do we need to get technical about light metering when we have a simple tool called a light meter to do all the work for us. The simplest way to visualize a light meter is to use the spot meter, set the ISO, put your camera on M and set an aperture. Point the spot meter at a neutral toned subject and set the shutter so the indicater marks the centre of the scale. Now point at the lightest part of the image and note if the reading is more the +2. Next look at the darkest part and note if the reading is less than -2. It the brightest part was much over +2 your exposure will blow out the highlights. If you don't want this to happen you will have to close down or slow down a bit (or both) so that the brightest part of the scene will read not much more than +2. The rest of the image may be dark, but you have saved the highlights. A similar process on not having the darkest part of the image less than -3 can be used to avoid blocking up the shadows. If you can't find an exposure setting that is within this range for both the darkest and brightest parts of the image you will have to compromise and accept some blown highlights or blocked shadows.

All the evaluative metering does is take readings at many places in the image, compare the distribution of light patterns to pre-programmed algorithms, pick an algorithm and set the exposure accordingly. A predominantly dark or light scene may fool the algorithm if it does not recognize that it is not the assumed predominantly neutral toned.

If shooting from a tripod you may want to two or more shots exposed for the entire range and combine them late in Photoshop.

As you learn to use your meter you can start metering off non-neutral toned subjects. For example in a snow covered winter scene you my meter off the snow and set +1.3 to +2 or so exposure to get a properly exposed snow scene.

You don't need to do any math to take a properly exposed picture.

ScottE wrote in post #3487348
Why do we need to get technical about light metering when we have a simple tool called a light meter to do all the work for us. The simplest way to visualize a light meter is to use the spot meter, set the ISO, put your camera on M and set an aperture. Point the spot meter at a neutral toned subject and set the shutter so the indicater marks the centre of the scale. Now point at the lightest part of the image and note if the reading is more the +2. Next look at the darkest part and note if the reading is less than -2. It the brightest part was much over +2 your exposure will blow out the highlights. If you don't want this to happen you will have to close down or slow down a bit (or both) so that the brightest part of the scene will read not much more than +2. The rest of the image may be dark, but you have saved the highlights. A similar process on not having the darkest part of the image less than -3 can be used to avoid blocking up the shadows. If you can't find an exposure setting that is within this range for both the darkest and brightest parts of the image you will have to compromise and accept some blown highlights or blocked shadows.

This works nicely if you don't mind fixing the image in postprocessing so that it appears to be correctly exposed. In certain situations, you might have to do that anyway.

I was under the impression that "correct exposure" was all about avoiding the need to do any postprocessing of the image.

Right. Well, in the thread I linked to, the use of evaluative metering mode plus exposure compensation is exactly the method they recommend using to get correct exposures.

But you can't do that unless you understand how exactly the mode works and, thus, under what situations you will need to apply EC. If you don't actually know how the mode does its thing, then you have to remember the amount of EC needed for any given situation. And as evaluative metering mode changes over time (across camera models, for instance, and possibly within the same camera body if you do a firmware update), you may have to re-learn the EC values.

You don't if you're willing to commit to memory various "rules of thumb" and the like. I'm one of those people who has a really horrible memory, so "rules of thumb" don't work well for me most of the time. I compensate for that by trying to acquire a deep understanding of the subject at hand so that I can derive what I need on the fly from first principles.

I presume I'm not unique in the world in that way, but maybe I am.

Photography, being as much (if not more so) an art than anything else, may require more rote learning than most other disciplines I have experience with, so some of that is almost certainly unavoiadable. But it's in my best interests, at least, to minimize that as much as possible. In other words, make it as much of a science as possible.

kcbrown wrote in post #3486916
....The camera's meter seems to be yielding the EV based on ISO 4. A bit odd, but the math does appear to work out. So if you have the EV from the EXIF and the aperture and you want to compute the shutter speed the camera would have used (if you hadn't been in manual mode), you can use this formula:

s = N^2/(2^(EV + log2(ISO/100) + 4.644))

Similarly, if you have the EV and the shutter and want to compute the aperture, you can use this formula:

N = sqrt(s*2^(EV + log2(ISO/100) + 4.644))

Where did the magic number 4.644 come from? It's log2(100) - log2(4). In other words, it's basically the number of stops between ISO 100 and ISO 4 (which appears to be the ISO the camera's EV measurement is based on). And log2(ISO/100) is the number of stops difference between ISO 100 and the ISO you're using.

It's still probably going to be easier for most people to just use Av or Tv + exposure compensation than to go through the trouble to deal with the above, but it's there for anyone with the determination to use it.

As I said in the other thread, I went through all this some time ago and established to my satisfaction that the for EV is based on ISO 100 - all of which is explained in the Help file for my PhotoTool, which is available here:
http://www.zaffora.com​/W9DMK/Programs/PhotoT​ool.zip

Robert_Lay wrote in post #3487945
As I said in the other thread, I went through all this some time ago and established to my satisfaction that the for EV is based on ISO 100 - all of which is explained in the Help file for my PhotoTool, which is available here:
http://www.zaffora.com​/W9DMK/Programs/PhotoT​ool.zip

That's excellent. I'll redo my calculations to verify which ISO the 30D uses as its reference, and perhaps also attach some images that include that EXIF data so that this can be independently verified.

Your writeup on EV (the "Understanding Exposure Values" topic) in the help file is much more illuminating than the one on Wikipedia (here). I don't suppose you'd be willing to update the one on Wikipedia to include what you've written?

kcbrown wrote in post #3488772
That's excellent. I'll redo my calculations to verify which ISO the 30D uses as its reference, and perhaps also attach some images that include that EXIF data so that this can be independently verified.

OK, attached you'll find two images I shot today which shows the values that I'm using to derive the ISO.

The EXIF of the first image has the following:

ISOSpeedRatings: 100
ApertureValue: 6.00 EV (f/8.0)
ShutterSpeedValue: 7.64 EV (APEX: 14, 1/200 sec.)
Measured EV: 8.75 EV

The EXIF of the second image has the following:

ISOSpeedRatings: 100
ApertureValue: 4.64 EV (f/5.0)
ShutterSpeedValue: 6.97 EV (APEX: 11, 1/125 sec.)
Measured EV: 6.50 EV

Now, for the exposure to be correct, EV + S + A + T = 0, where S is the ISO, A is the aperture, and T is the shutter speed, all in f-stops. The reference values in the above are 100 for the ISO, f/1 for the aperture, and 1 second for the exposure.

The shutter speed and aperture EVs above obviously should be negative, since they are both smaller than the reference.

Now, we're trying to figure out what reference ISO the camera uses for its "Measured EV" field. So, using "Measured EV" as the EV, and Sc as the camera's ISO reference for "Measured EV" we get for the first image:

-Sc + 8.75 + 0 (ISO 100's EV) + -6.00 + -7.64 = 0
Or Sc = -4.89, which corresponds to an ISO of 3.37.

And for the second image:
-Sc + 6.50 + 0 (ISO 100's EV) + -4.64 + -6.97 = 0
Or, Sc = -5.11, which corresponds to an ISO of 2.89.


Hmm...so it appears it's using ISO 3 as the reference for the Measured EV field and not ISO 4...

ISO 3's EV is -5.06, close enough to -5.0, so substitute 5.0 for 4.644 in the original equations, at least if you're using a 30D.

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

kcbrown wrote in post #3488772
That's excellent. I'll redo my calculations to verify which ISO the 30D uses as its reference, and perhaps also attach some images that include that EXIF data so that this can be independently verified.

Your writeup on EV (the "Understanding Exposure Values" topic) in the help file is much more illuminating than the one on Wikipedia (here). I don't suppose you'd be willing to update the one on Wikipedia to include what you've written?

Please note that in the concept of correct exposure there is the presumption of Zone V placement whereas in your calculations you are getting an EV that would give a Zone 0 placement.

Yes, I think that is what is going on. If you were using a hand-held digital spot meter by Pentax, the upper of the two pictures above would be characterized as an EV = 13.75 brightness instead of the EV = 8.75 found in the camera data, and that value would be consistent with your ISO, shutter and Aperture. There is an ANSI standard for all this, which is mentioned in my Help file. However, we do know that Polaroid's standard was different and we also know how to convert to the old Weston scale of candles per square foot.

Robert_Lay wrote in post #3490034
Please note that in the concept of correct exposure there is the presumption of Zone V placement whereas in your calculations you are getting an EV that would give a Zone 0 placement.

Yes, I think that is what is going on. If you were using a hand-held digital spot meter by Pentax, the upper of the two pictures above would be characterized as an EV = 13.75 brightness instead of the EV = 8.75 found in the camera data, and that value would be consistent with your ISO, shutter and Aperture. There is an ANSI standard for all this, which is mentioned in my Help file. However, we do know that Polaroid's standard was different and we also know how to convert to the old Weston scale of candles per square foot.

Ahh...yes, this makes sense.

Hmm...so for the purpose of using the Measured EV in the EXIF to work backwards towards figuring out how much "exposure compensation" you used while in manual mode, it sounds like you'd still have to account for that 5 stop compensation regardless.

kcbrown wrote in post #3487452
This works nicely if you don't mind fixing the image in postprocessing so that it appears to be correctly exposed. In certain situations, you might have to do that anyway.

I was under the impression that "correct exposure" was all about avoiding the need to do any postprocessing of the image.

If the range between the brightest and darkest areas in a photo are beyond the capabilities of your camer to record there is no "correct exposure" that will be able to get all the details in the highlights and the shadows. You have to make a compromise if you want to only take one exposure.

This is not a "rule of thumb" but simply learning from your experience so that you can get the best results within the limitations of your equipment.

Alternatively, you can realize that your camera cannot record everything and shoot two or more images that can be combined later in post processing to make up for the limitations of your equipment.

A third alternative of using a graduated neutral density filter is also available for many situations if you want to avoid post processing.

http://www.unswphotocl​ub.org …c33a68a4a0bf00f​ecaacec79c
Kinda neat tutorial not real tech but gives you an idea .

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

KC since you like partial and spot thought you might like to read this website I found .
I used partial for my moon pic was interested in learning more about it for XT.It has link for 30D
http://www.spotmeterin​g.com/spc30d/