Metering

[oldgreygary]

I have had a Canon G2 for about 3 months now. I have been using the evaluative metering mode. There are also 2 other options, centre weighted and spot. Bearing in mind that the camera is about 10 years old and in technology 6 months is a long time. As I understand it the evaluative mode relies on the camera having its own onboard database of various scenarios and makes a best choice based on that. Centre and spot metering seem to be similar but spot allow you to be more specific on the area that you are metering.

Anyway, there are several questions in there. Presumably, the technical aspects behind metering have moved on in cameras, so how much of a difference does that make? Another question is do you always rely on evaluative mode or do you adopt a method based on what the subject is? Is a standalone meter a better alternative to the one on the camera?

I think I read somewhere about metering for the darkest area where you want detail to show. So does that mean either centre or spot metering would be more effective? Or can you use evaluative then use options like exposure compensation or automatic exposure bracketing (AEB) to give you options?

I guess it's about using the technology what's available to your best advantage. I will be very interested to hear your thoughts.

Cheers for now

Gary

[Colin Southern]

As I understand it the evaluative mode relies on the camera having its own onboard database of various scenarios and makes a best choice based on that.

Hi Gary,

Not really. Basically evaluative -v- centre-weighted - spot are just a way to tell the camera what part of the scene is important exposure wise. Evaluative looks at the entire scene (typically used for a landscape) - centre-weighted says to the camera "every thing is important - but what's in the middle is MORE important (typically a person perhaps?) - whilst spot-metering says "only the small bit I'm pointing at is important (in terms of exposure).

What a lot of folks don't seem to realise though is that the camera can only look at the absolute value of the light hitting it - and the range of brightnesses hitting it (cameras have more than one sensor) - it has no knowledge of whether it's about to shoot a black cat on a black rug - or a white polar bear in the snow ... so what it does is aim for the middle - and expose for middle grey. So if you shoot the proverbial black cat - you'll get a gray cat - shoot the white polar bear and you'll get a gray polar bear (unless in both cases you add or subtract what's called EC - Exposure Compensation).

So if all that it true, then how come the camera often gets it right? It come down to the dynamic range of the scene; reflective objects only have about a 4 stop variation between black and white - so for a typical landscape scene (with a typical 4 stop dynamic range) it'll aim for the middle so that whites expose 2 stops above middle gray (which is correct) and blacks 2 stops below middle gray (which is also correct).

Although metering has improved over the years, the basics remain the same; so I'd be inclined to just use it and see what kind of results you get. Even with more modern cameras, one still has to modify what the camera suggests depending on the circumstances.

With regards to spot metering - basically - whatever you spot-meter is whatever the camera is going to expose as a middle gray unless you dial in EC. If you spot-meter something dark then those dark areas will be exposed as midtones - which may or may not be a good thing -- because although you'll now have visible detail in dark regions - you'll also be risking blown highlights.

I've written a bit about this in the past if it helps ...

Custom Settings for 1D Mark III

What is the best method for learning how to expose photos correctly?

What does it mean to "meter off of ..."

[oldgreygary]

Thanks Colin, the links are clear and understandable.

I re-read the section in the G2 manual about manual exposure. In there it talks about the standard exposure which I think translates to the 18% grey setting. It then goes on to say that when you set the exposure it displays on the LCD how the setting is related to that. i.e. EV0 = standard and + or - figures are either over or under the standard upto EV-/+2.

I think as I understand things if you either meter for either a white or black subject. Then the camera in the case of white underexposes to give you grey. While in the case of black it overexposes to also give you grey. Therefore you need to feed in some exposure compensation to adjust them to get white or black? But if you have white and black together and meter them both they effectively cancel each other out and you get a correct exposure?

I think that this also adds some light (terrible pun) to metering the sky for a landscape(mentioned in another post) I think that by accident the other day I found this out. I set AEB to +/-2. I then metered off the sky and locked the exposure. I then refocused on the whole scene and took the image. While the EV-2 wasn't much use the +EV2 gave enough exposure to work with in PP. I think what is going on is that by metering on the sky that gave me an exposure to the standard or 18% grey in effect underexposure of the sky. After I re-composed and took the shot with the locked exposure the +2EV just about hangs onto the sky while bringing out the blacks and shadows in the whole scene which gives enough to work with in PP. I guess that by getting into manual mode this can be refined further by exposure adjustments.

Starting to understand this fills in some of the big holes in my knowledge. But it does give the feel of taking steps forward. Which is great!

Cheers for now

Gary

[xpatUSA]

Although metering has improved over the years, the basics remain the same; so I'd be inclined to just use it and see what kind of results you get. Even with more modern cameras, one still has to modify what the camera suggests depending on the circumstances.

Hello Gary,

If you have a point-and-shoot camera with at least evaluative and center-weighted metering, you can get a really good feel for and see "what the camera suggests" in each metering mode by holding down the shutter button 1/2-way and moving the camera slowly around a scene while watching the LCD.

That way, you wouldn't have take a load of shots, make notes, upload them, review, scratch your head and repeat, ad naus.

I use spot metering normally for everything, center-weighted if spot is not available and scene evaluation for snapshots in auto mode. I'm a bit old-fashioned about anything automatic for serious work :-)

[oldgreygary]

Ted,

The G2 does have spot metering as well as evaluative and centre-weighted. Considering it was 10 years old when I bought it at the beginning of the year it has a relatively high spec. I must admit that I am going to try spot metering and see how I get along with it. Most of my efforts have been accidental and I want to be able to work towards a method were I have more confidence in the way I am setting exposures.

Cheers for now

Gary

[Gene Bjerke]

Something else you might think about: many cameras can be set to display a live histogram in the viewfinder. You can play with the exposure adjust control and see how it shifts the balance of highlight and shadow areas. What you want to avoid if you can is a line that goes to the top of the histogram. At the same time, you can often get a sense of how the different parts of the picture will reproduce by how it looks in the finder.

Gene

[Colin Southern]

What you want to avoid if you can is a line that goes to the top of the histogram.

Hi Gene,

Usually yes, for a purely reflective scene, but scenes with specular reflections and/or backlighting (think of someone standing in front of a window where you want the edge lighting and blown background - but a correctly exposed face) then there will be a big spike at the end of the histogram, which is both expected and normal.

[Inkanyezi]

What you want to avoid if you can is a line that goes to the top of the histogram.

Usually yes

No, forget about the top, concentrate only on the lower right corner.

I think a bit of understanding the histogram is needed here. The histogram is only a statistical image of what pixel values exist in the scene, but says nothing about how they are distributed. "Reaching the top of the histogram" only indicates a pixel value that has many more pixels than any other value. A histogram cannot overflow at the top, the top is in fact completely irrelevant to the histogram per se, it is only a limit of what amount of a single value that is possible to represent in the diagram itself.

So the height of any portion of the histogram but the very ends can be disregarded. Height is irrelevant, except at the extreme right end of the histogram. Any height AT ALL at the right end means that there are blown out pixels. The actual height tells us how many pixels that are blown. When there are specular reflections or light sources in the image, we expect some pixels to be blown out, and that is normal. So the histogram should then reach some height at the extreme right. With a pure white or blown-out background, there will be a spike with substantial height at the right end, but mostly a histogram should end in the lower right corner.

The left end of the histogram can sometimes reach substantial height, and it does not indicate under-exposure, but only that there are many totally black pixels in the image. If we want more shadow detail, we may choose the largest possible dynamic range by using lowest ISO, lowest amplification of the signal (and noise), with more exposure. That will lower the peak at the left side if it peaked at a high ISO setting.

In some images, there may be important highlights that we do not want blown out, but which are too small to give any substantial amount of pixels to show up in the histogram. The histogram for such an image may appear to end a bit before it reaches the right end. Histograms are not fine grained enough to show these minor parts of images, so we have to visually check them. Many RAW converters can show such details by highlighting pixels that reach the highest value in any colour channel.

[Colin Southern]

No, forget about the top, concentrate only on the lower right corner.

Not sure if we're talking about the same thing here or not. I took Gene's "top of the histogram" to mean "avoiding a spike at the right-hand end".

[drjuice]

For wildflowers (of which I've taken a LOT of images lately), I spot meter on the piece of the image (usually less than 5mm at the widest spot) that I really want to emphasize in the result. Then, I keep the button pressed halfway down because it will then use the metering result even though I often move the object out of the center of the image. Finally, when I have the image framed in the way I want, I push the button the rest of the way down. Fortunately, when I use the Remote Commander (what Sony calls their remote control device), I can freeze the meter result so it's easy to maintain even for way extended exposure.

HTH.

v

[oldgreygary]

From further reading as I understand it, the highest % of data in a digital camera is stored in the highlight region. The phrase I think that was used was to expose for the highlights and develop for the shadows. The other thing that is well not exactly confusing me but which doesn't help. Is that again, from reading is that a cameras histogram is based on the JPEG and is not totally reliable and can give a false impression of over exposure. I am beginning to wonder whether I should permanently have a positive exposure compensation setup? As my camera seems to be a bit conservative in the highlight region. I have switched to centre weighted/spot metering. I am finding that a bit more reliable than the evaluative mode.
As in one of the above posts I meter then either lock that in and re-compose or switch to manual and use the readings as a starting point.

Cheers for now

Gary

[Colin Southern]

From further reading as I understand it, the highest % of data in a digital camera is stored in the highlight region. The phrase I think that was used was to expose for the highlights and develop for the shadows. The other thing that is well not exactly confusing me but which doesn't help. Is that again, from reading is that a cameras histogram is based on the JPEG and is not totally reliable and can give a false impression of over exposure. I am beginning to wonder whether I should permanently have a positive exposure compensation setup? As my camera seems to be a bit conservative in the highlight region. I have switched to centre weighted/spot metering. I am finding that a bit more reliable than the evaluative mode.
As in one of the above posts I meter then either lock that in and re-compose or switch to manual and use the readings as a starting point.

Cheers for now

Gary

Hi Gary,

1/2 the information captured resides in the top 1 stop that the sensor can handle - then 1/2 of the 1/2 that remains resides in the next stop down etc, and so on and so forth. This leads some folks to theorise that to capture the most information, one should always ETTR (Expose to the Right). It's a good-sounding theory, but in practice, it's not always appropriate (due to sensor characteristics that become apparent around the satuation point).

In reality, it comes down to the dynamic range of the scene that you're trying to capture ... if it's just a normal reflective scene then there's little advantage in exposing to the right; there's more than enough clean data captured with a normal exposure, and additionally, ETTR can make it difficult to get the shadow and midtones back to where they should be without the image looking a bit "off". On the other hand, if the scene has a high dynamic range then yes - you need to expose to the right to avoid shadow noise. Exposing to the right is also more and more important the higher the ISO mode you use (due to the decreased dynamic range capability of the sensor).

In terms of the in-camera histogram, yes, it is based on the in-camera jpeg. And yes, it does tend to be conservative - but - "conservative" also equates to "safety margin" - it all comes down to how much safety margin you can afford - and that in turn comes down to the dynamic range of the scene -v- the dynamic range capability of the camera you're using in it's current configuration.

With regards to always having a positive exposure compensation set ... it's a bit like "always answering C in a multi-choice exam" ... sometimes it'll give you the right answer, but not always. Shoot a black cat on a black rug and you'll get even more over-exposure as an example.

Best solution (in my opinion) is to just learn metering so that you can make the best decision at the time. Spot metering will always expose the bit you spot meter as a middle grey - if it's NOT a middle grey then you need to add or subtract EC to get a correct exposure (ie spot-meter a white sheet and it'll under-expose 2 stops).

[Inkanyezi]

Maybe it is worth mentioning also, that "evaluative" metering is a buzzword, only a buzzword. Did any manufacturer really explain what it does?

In fact there may be cameras that really do evaluate the image, but to that end, the light metering sensors must be aware of the image. Only EVIL cameras always work in a mode where this might be possible, but Nikon as well has a system that at least theoretically can do that. Whether it does and what it does though - search me. No other manufacturer afaik has a system that can evaluate the image. Then comes "center weighted". So what does it do? How much "emphasis" does it put on the center, and what is emphasis in this sense? Can it be expressed as percent, and how can I visualise such a "percentage"? Just like evaluative, center weighted is mostly moot.

There is one measuring method that we may be pretty sure of what it does, spot metering. And to use it, you must understand it, but it is fairly simple. When you meter off something that shall be white, but not burnt out, it's a +2 compensation to make it white. When you measure something black, it is mostly a -2 compensation to get it in the right zone. The easiest to measure in fact is the brightest highlight that you want to save without having it flood over. Any scene that does not have such highlights may be measured with any of the other methods; they are not useless, it's only that their designations are a bit - non-informative. Once one has learned the camera, they can often be used, taking care of how we want to represent the scene. For example a snow scene often needs +1 compensation in any of those modes, just to tell the camera that we want it brighter than "normal". The tricky ones are only high contrast scenes, those that are best metered with spot. The only problem is that sometimes the highlights that you wished you had a reading of are too small for the meter spot. That's where chimping with highlight warning comes in. The daisies in the lawn will not influence any of the meter reading systems enough to save them from being burned out, so either you guess that you need -2/3 to get them right, or you get to that decision after taking an average shot and realise that they are burned out. You may leave them like that, noone will notice. But if you really want to see detail, make out the petals, then you have to decrease exposure if it's averaged, evaluative or center weighted. And spot does not help you there.

So let's not delve too much in the camera salesman terminology. We have essentially two kinds of metering in the camera, where one is split up in three terms, but all three do just about the same thing, meter the whole scene. Then we have one method with a fairly limited measuring area. All of these can be used creatively to achieve the results that we want. The three variants of "average" over the whole area become uncertain when contrast is high, and it's then we may need to spot meter and place the highlight where the highlight shall be. In my camera it's at +2, but if you have a better sensor with more DR it might be higher. You must find it out for yourself.

[oldgreygary]

Colin, Urban, Thanks both for very informative posts. From what you have said its about keeping all options open dependant on situation? I think I am being unrealistic in thinking that one method will solve all problems. As you say I need to find out myself. By shutting off options then that limits what is possible. Note to self, don't become fixed in one routine, try options available!

Cheers for now

Gary

[Colin Southern]

I think I am being unrealistic in thinking that one method will solve all problems.

Hi Gary,

Yeah. But with digital, "who cares"

[Inkanyezi]

There is another method that works very well for PowerShot cameras. With CHDK, you can see in real time, before the image is taken, which parts of it will be blown out and which parts will hold no detail, by using "Zebra Mode", which is akin to "Highlights" in chimping, but shows the highlights and shadow areas real time. The only problem I have encountered with that mode is that it chews more battery power than the standard modes without the firmware extension. CHDK is a very useful tool that makes the PowerShot models more versatile than other compact cameras.