Light Metering

B

Briggsy

Suspended / Banned
Messages
117
Edit My Images
Yes
Can anyone explain or point me somewhere that explains what this is and how it's useful?

I don't really get what it is!
 
the meter works out how dark or light the scene is and then sets the aperture and shutter speed to allow the light through to the film or sensor depending on what asa/iso you have loaded/set.

Most meters are calibrated to meter to 18 per cent grey which is similar to light tarmac on the road

stew
 
The camera has an array of light metering sensors which are read and averaged together (based on the "metering mode" you have selected). This gives the camera a single value for the amount of light in your scene.

Based on which mode your camera is in, it will then work with the parameters it can change to get a good exposure in that light. If you are shooting full auto with auto ISO, then the camera will find a good balance of ISO, aperture size and exposure time all on it's own.

Of course, sometimes the metering algorithm fails to give the result you want, like with a person standing next to a window with a sunny day outside. So you have to nudge it with "exposure compensation". This tells the camera to adjust the magically derived light level number (EV).

Are you seeing the light now? :) I got my first handheld dedicated light meter today in the post, so I've refreshed my brain cells a bit..
 
Briggsy said:
Can anyone explain or point me somewhere that explains what this is and how it's useful?

I don't really get what it is!
Click to expand...

Very basically, the sensor needs a certain amount of light for the best result - the correct exposure. Metering is the business of measuring the light, and the camera then converts the reading into a combination of camera settings to deliver the right amount of light to the sensor.

The light level varies a great deal throughout the day. Indoors under artificial light, it's actually relatively dark. You don't notice it much as your eye is extremely good at adjusting itself, much better than any camera can.

While the camera's meter makes a good guess at what the correct exposure settings might be, what you see on the LCD and histogram is what the exposure actually is - they are often slighly different. Follow the LCD as that's the one that matters, and learn to read the histogram, it's very easy.
 
From what I've read (very briefly) on the histogram. If it touches the right hand side the pic is overexposed?
 
Briggsy said:
From what I've read (very briefly) on the histogram. If it touches the right hand side the pic is overexposed?
Click to expand...

If it only touches the right hand side, then it's okay. Basically, so long as all the important tones are somewhere in the middle, preferably a bit to the right (as there's much more data recorded on the right) then you'll have a workable exposure.

You can make it much more complicated than that, in the quest to extract the last drop of image quality, but it won't be far off that anyway. That will be quite 'workable' even if it's not absolutely 'optimum' - and if you go for optimum and the expose-to-the-right technique, then you need to shoot Raw and post process the files.

Enabling blinkies is perhaps the single most helpful thing there is for getting correct exposure. Blinkies is the highlight over exposure warning that flashes on the LCD (see handbook). Use that with the histogram and you have everything you need to know.

Good guide to histograms here http://www.sekonic.com/images/files/HistogramsLightmetersWorkTogether.pdf
 
Would somebody be so kind as to explain the 18% gray thing I keep reading about? [/HIJACK]
 
When the camera meters the scene it tries to make an average of all the bright, dark, and indeed all the tones to an average 18% grey tone.

You can confirm this yourself by getting a white sheet of paper, a grey sheet of paper, and a (you guessed it) black sheet of paper, taking a picture of each in one of the semi automatic modes of each sheet so that the paper fills the viewfinder. You should get images which look pretty similar to each other, ie a grey image. The camera reduces the exposure (either by shorter shutter speeds or smaller apertures) to make the white paper grey, and lengthens the exposure (either by longer time or wider aperture) to make the black paper grey.

Why is it good to know this? Well, if you know that the camera tries to get an average of 18% grey, then you will get an idea why some pictures of snowy scenes may come out a bit dull. After awhile you would hopefully get to a place where you know to add a bit of plus Exposure Compensation in that situation, or over expose if in Manual mode. This is what a 'Snowy Scene' preset does on cameras which have it.

The same is true for predominantly dark scenes, the camera will try to brighten the scene (to make it average out to grey), so if you want the dark to stay dark, you may need to add some minus Exposure Compensation, or underexpose in Manual mode.

Hope that helps.:)
 
benhaha85 said:
Would somebody be so kind as to explain the 18% gray thing I keep reading about? [/HIJACK]
Click to expand...

Do a google for Zone Metering. You'll find some charts for 5 different percentages of grey. Read up about the different shades and how exposure compensation can affect your exposure.
 
benhaha85 said:
Would somebody be so kind as to explain the 18% gray thing I keep reading about? [/HIJACK]
Click to expand...

18% grey is a mid-grey tone used for calibrating exposure meters. It sits roughly in the middle of the histogram but really relates to film. It's actually a bit of an irrelevance with digital and anyway there is debate as to whether modern cameras are calibrated for 18%, or 12%, or 15%. By the same token, manufacturers are also coy about what their ISO ratings actually mean which is a kind of flipside of the same debate.

However, 18% grey is universally used. You can buy 18% grey cards for exposure reference and if nothing else it is a constant, which is the important thing.

If you want to get the most out of a digital sensor, then expose-to-the-right (of the histogram) technique is the way to do it. This turns the 18% grey principle on its head for exposure purposes, deliberately shifting 18% grey away from the middle of the histogram, then pulling it back down again in post processing.

I think it's important to bear in mind that the business of exposure metering is about making a best guess at what the best exposure settings should be, before you take the picture. That is obviously very important, and vital with film, but there is room for lots of errors.

If you want optimum exposure, the advantage of digital is that if you take a test shot, then the actual exposure is shown on the LCD (no guesswork there) and a detailed analysis is available on the histogram. Using the LCD image, the histogram, and with the wonderful invention of blinkies turned on (highlight over exposure warning, enabled in the menu) which flashes black and white on the LCD, using all those things in combination will get you as close to optimum exposure as is possible. It is by far the most accurate method.
 
Cheers all. I already tend to expose to the right and then pulling it back down, but now I understand the reason why. I guess you can call that my one new thing I learnt today!
 
As a newbie to Manual mode, I find the ETTR discussion and the posts for and against its application fascinating but quite messy because everyone is second guessing camera manufacturer engineering and real-world photos with tangible differences are few and far in between. Color swatch charts have been shot and shots shot down and reshot and the cycle goes on :|

Also my understanding is that the histogram displayed on the camera's screen is based on "what-would-an-in-camera-jpeg-look-like" and as such is compressed to an 8-bit space with curves & styles applied. I might be wrong but if it is this way, it means that there probably is headroom to find data beyond the histogram and blinkies in your raw conversion. Bring on the raw histograms and a whole new debate! :lol:

When I first got a Canon DSLR, everyone told me to underexpose, to make sure I don't blow the highlights and that Canon tends to overexpose. Now everyone tells me to overexpose just short of blowing them.. I'm too lazy to intentionally add more PP work on my photos so I'll (for now) shoot them with exposures pleasing to my eye :)

Now Mr. Canon, why not make me a sensor with at least 14EV of dynamic range ;)

But I digress.. a lot.
 
With regard to the image and Histogram displayed on the LCD screen, it is based on a Jpeg with all the settings set in camera as vaizki said.

The consequence of this is that any picture styles or in camera settings such as sharpening contrast etc are being applied, even if you are shooting RAW. This may result in an inaccurate RAW file being recorded. This could be a problem if you are shooting RAW, but shooting Jpeg or Tiff however lets you see what you are recording accurately. I shoot RAW all of the time so have set my camera to the Neutral Picture Control in the hope that the Histogram is a little bit more accurate.

Just another thing to consider. ;) :lol:

Apparently the camera manufacturers only allow the Jpeg to be viewed because dull RAW files wouldn't look as attractive on the screen in shops. :bonk:
 
Briggsy said:
From what I've read (very briefly) on the histogram. If it touches the right hand side the pic is overexposed?
Click to expand...

if it touches the right hand side of the graph, some parts of your scene were brighter than your sensor was capable of rendering. Whether or not that's important depends on what you're trying to achieve; you may not care. Or you may.
 
Some misunderstandings here.

The Raw file is just a pile of data. You can't see it until it is converted into an image file, usually a JPEG. What you see on the LCD is a JPEG, even if you only shoot Raw, and that has all the camera's pre-sets applied, picture styles, white balance, noise reduction etc. That JPEG is always tagged to the file, but underneath the Raw is original and untouched. If you make changes, the it's the JPEG that changes, not the Raw.

If you shoot Raw, most post processing programmes will pick up on the JPEG presets and show the Raw conversion with those applied. If you print that straight out without making any modifications, the image will look exactly like the LCD image you saw. It will be just like you had shot as a JPEG in the first place. So in that sense, if you don't make any changes to the image in post processing, shooting Raw is a waste of time. Just let the camera do it automatically - same result, same quality, much easier.

The histogram is also generated off the JPEG and in that sense it differs from the Raw, which holds more data. If you whack up the picture styles etc the JPEG can become substantially changed from the Raw and therefore misleading. In terms of exposure and the histogram, the thing to be aware of is the contrast setting. It stretches the histogram and can suggest overexposure when in fact there's quite a bit of headroom left on the Raw. If you want to use the LCD/histogram/blinkies for expose-to-the-right technique, then turn the contrast down to minimum and it will show the maximum range of tones on the Raw with minimal (if any) change. You'll see the full dynamic range and be able to exploit that by pushing the exposure (to the right) with confidence.

If you work in this way, ie ETTR technique, you must shoot Raw and post process in order to restore the tones you have shifted away from their normal value (so you can exploit the full dynamic range of the sensor) back to where they should be. If you don't, the whole thing will look washed out and over exposed. But if you do, you will get more detail in the mid tones, and much more detail in the shadows.
 
I'll assume you're talking about the in camera histogram - I wonder if it's using 8 bit data for that, I'll have to dig around a bit. A photoshop histogram is taken off whatever it's holding. So your photoshop histogram of a raw image will look very different to that on the camera, if the camera is representing 8 bit data.

<TechSpeak>
A histogram is ''a graphical - as in graph, not graphic - representation of the tonal distribution in an image'. Essentially it's a graph plot of intensity - horizontal - against number of pixels. So for any given value along the bottom, the greater the vertical height of the graph at that point, the more pixels there are at that intensity.

So when we said earlier, 'touches the right', what's also important is the height of the graph where it touches. If the graph touches the right but the pixel count values at the point on the graph leading up to the right are low, you have only a small part of your image is approaching peak white and you may not notice it.

When you're dealing with this stuff, it's important to have an understanding of what the true bit depth of your sensor is. By the time you get your raw into Photoshop, it should be 16 bit but in no way does that mean the captured shot was 16, it could have been up-sampled. In fact probably was. Very few cameras capture at full 16 bit and you can't make up data that isn't there.

here are some useful links:

http://www.normankoren.com/digital_tonality.html
http://www.makeuseof.com/tag/read-cameras-histogram-perfectly-balanced-images/

the first is a bit techy but worth the work.
</TechSpeak>
 
kimaldis said:
I'll assume you're talking about the in camera histogram - I wonder if it's using 8 bit data for that, I'll have to dig around a bit. A photoshop histogram is taken off whatever it's holding. So your photoshop histogram of a raw image will look very different to that on the camera, if the camera is representing 8 bit data.

<TechSpeak>
A histogram is ''a graphical - as in graph, not graphic - representation of the tonal distribution in an image'. Essentially it's a graph plot of intensity - horizontal - against number of pixels. So for any given value along the bottom, the greater the vertical height of the graph at that point, the more pixels there are at that intensity.

So when we said earlier, 'touches the right', what's also important is the height of the graph where it touches. If the graph touches the right but the pixel count values at the point on the graph leading up to the right are low, you have only a small part of your image is approaching peak white and you may not notice it.

When you're dealing with this stuff, it's important to have an understanding of what the true bit depth of your sensor is. By the time you get your raw into Photoshop, it should be 16 bit but in no way does that mean the captured shot was 16, it could have been up-sampled. In fact probably was. Very few cameras capture at full 16 bit and you can't make up data that isn't there.

here are some useful links:

http://www.normankoren.com/digital_tonality.html
http://www.makeuseof.com/tag/read-cameras-histogram-perfectly-balanced-images/

the first is a bit techy but worth the work.
</TechSpeak>
Click to expand...

One way of thinking about bits - 8, 12, 16 or whatever - is to think of a car's speedometer. Say it goes from 0-100mph in 10mph increments. That would be ten bits. If it is only marked at 0, 50 and 100mph, that's only two bits but the car still goes from 0-100 and any speed inbetween.

In practise, I find histograms very easy to use as a rough guide to where the majority of the important tones are, and to shift them left and right accordingly. But when it comes to really fine tuning, that's much harder. To see what I mean, check out the quiz on the Sekonic link on histograms I posted earlier (this one http://www.sekonic.com/images/files/HistogramsLightmetersWorkTogether.pdf ) and try to match the histograms to the images. It's not easy!

When you're pushing the exposure to the right (of the histogram) for maximum quality, you need to know exactly which areas of the image are blowing to pure white, and the histogram often doesn't tell you clearly. The point being, most images will contain areas of blown white and that doesn't matter at all - they're white after all, that's what you want, and trying to contain them is not only unnecessary but will likely ruin the whole point of ETTR excercise.

Looking at the histogram, it will be clear that some areas of the image are blowing to white, but which ones? In a group portrait it could be the bright highlights shining off white cuffs and collars, or bright spots on the sky, whatever. No problem, let them blow. But if it's some important faces in the background, maybe caught in a bit of slightly brighter light, then you need to keep them and will want to moderate the exposure in order to retain some tonal detail there.

That's where blinkies are so invaluable - the highlight exposure warning that flashes black and white on the actual LCD image. Blinkies are coupled to the histogram and give you a very precise visual indicator of exactly which areas of the image relate to those parts of the histogram that are busting up against the right hand edge.
 
HoppyUK said:
When you're pushing the exposure to the right (of the histogram) for maximum quality, you need to know exactly which areas of the image are blowing to pure white, and the histogram often doesn't tell you clearly.
Click to expand...

There's nothing wrong with looking at the picture; if you can't see it on the image, it doesn't matter. So why bother.


To see what I mean, check out the quiz on the Sekonic link on histograms I posted earlier (this one http://www.sekonic.com/images/files/...rkTogether.pdf )
Click to expand...

This is misleading, it's asking you to do something that's much more difficult, but more importantly, something you'll never need to do. You don't use these things to identify images, they're just there to give you some idea of how your tonality is spread. Again, they're no substitute for Looking At Your Picture but they are a useful guide to where you need to go to fix it.
 
kimaldis said:
There's nothing wrong with looking at the picture; if you can't see it on the image, it doesn't matter. So why bother.




This is misleading, it's asking you to do something that's much more difficult, but more importantly, something you'll never need to do. You don't use these things to identify images, they're just there to give you some idea of how your tonality is spread. Again, they're no substitute for Looking At Your Picture but they are a useful guide to where you need to go to fix it.
Click to expand...

It's just showing you that the histogram can be hard to read if you want to relate detailed tonal data to detailed image information. Ie, the histogram that can be misleading.

I agree that 'looking at your picture' is the ultimate conclusion to all this, but at the time of shooting all you have to look at is the LCD picture. Even on the best screens, like yours (and mine :) ) the image is small, often hard to see in bright light, and since it is a JPEG it is also subject to Picture Styles etc.

For critical exposure assessment, you need to look at the histogram and blinkies.
 
Slightly OT, but if using a hand held meter, am I right in thinking you have to choose the ISO to use and either what F stop you want or what shutter speed, the meter then gives you the third factor, either shutter speed or F stop?

Or am I missing something?
 
Can I just say this is one of the most interesting and useful threads I've read for a long time! I think I'll be spending the next few days reading all the articles you lot have links to!
:thumbs:
 
Shutterbox said:
Slightly OT, but if using a hand held meter, am I right in thinking you have to choose the ISO to use and either what F stop you want or what shutter speed, the meter then gives you the third factor, either shutter speed or F stop?

Or am I missing something?
Click to expand...

A hand meter is no different. There are three parameters - shutter speed, f/number and ISO. You mix and match them however you want according to the light level. That's all there is to it.

Traditionally, we tend to set the ISO first. Certainly with film, as you can't change it once it's loaded in the camera. Whichever way of working you prefer.
 
You must log in or register to reply here.
Back
Top