Histogram and exposure

R

RL355

Messages
80
Edit My Images
Yes
Sorry newbie question…:) I am fairly new to photography and have been experimenting with my digital camera. I have a question regarding exposure and the histogram. I hope someone could explain the following….

If I take a picture of a white card filling the whole of my frame. I then look at the histogram I notice that the details are centrally plotted on the histogram.

If I take a picture of a black card filling the whole of my frame. I then look at the histogram I notice that this is ALSO centrally plotted on the histogram

My understanding of the above is due to the fact that the camera is programmed to expose a scene as 18 percent middle tone as that is what the camera thinks is an ideally exposed picture.

If I take a picture now with the frame filled with half white and half black card. The histogram shows 2 distinct peaks. One in the highlight end of the histogram and one in the Shadow end of the histogram. How come there is no peak in the middle of the histogram like you do when you took the black and white cards individually ?

thanks in advance
 
You have already answered your question.

The half black/white will show as 2 peaks on the histogram because that is what it actually is and in that case even if the camera exposed to get 18% tones the 18% would actually be made up by 2 distinct shades - which would show as the 2 peaks.

You will also see this on histograms taken of proper scenes - on a sunny day for example.

The histogram is an excellent way of determining that you have the correct exposure - if the histogram sits right in the middle then you have the correct exposure.

If you find that the histogram extends all the way from the left hand to the right hand side then the dynamic range of the scene may exceed the dynamic range which your sensor is capable of registering.

In this case you have the choice of exposing for the highlights or the shadows - or alternatively of creating an HDR picture.
 
Thanks for the quick reponse...Iam still lost as to why there is two peaks in the histogram when we have half black and half white filling up the frame. I would have thought it would be still centrally plotted on the histogram.
 
think of what the histogram is telling you: number of pixels at a certain brightness

the average of the 2 peaks will be in the middle

one peak is for the pixels registering dark values, the other peak is the number of pixels at the bright values
 
RL355 said:
If I take a picture now with the frame filled with half white and half black card. The histogram shows 2 distinct peaks. One in the highlight end of the histogram and one in the Shadow end of the histogram. How come there is no peak in the middle of the histogram like you do when you took the black and white cards individually ?
Click to expand...

Welcome to TP :)

If you scambled the two peaks, you would get one peak at the 18% grey mid-tone.


petersmart said:
You have already answered your question.

The half black/white will show as 2 peaks on the histogram because that is what it actually is and in that case even if the camera exposed to get 18% tones the 18% would actually be made up by 2 distinct shades - which would show as the 2 peaks.

You will also see this on histograms taken of proper scenes - on a sunny day for example.

The histogram is an excellent way of determining that you have the correct exposure - if the histogram sits right in the middle then you have the correct exposure.

If you find that the histogram extends all the way from the left hand to the right hand side then the dynamic range of the scene may exceed the dynamic range which your sensor is capable of registering.

In this case you have the choice of exposing for the highlights or the shadows - or alternatively of creating an HDR picture.
Click to expand...

While the histogram is indeed an excellent way of determining optimum exposure, having all tones lumped in the middle is rarely optimum.

(Bear in mind that the histogram is generated from a JPEG off the Raw file, so you it can be changed a bit according to the parameters you have pre-set in the camera [Picture Styles] but it's usually a very accurate guide - certainly the best you can get.)

Wookie said:
think of what the histogram is telling you: number of pixels at a certain brightness

the average of the 2 peaks will be in the middle

one peak is for the pixels registering dark values, the other peak is the number of pixels at the bright values
Click to expand...

That's what the histogram shows - the volume and distribution of dark and light tones. Where they sit on the scale depends on the level of exposure. The shape of the histogram does not change much with exposure level as it is determined by the subject (although it is not on a linear scale) so what happens with different exposure levels is that you can slide the graph right and left so that you get the important subject tones where you want them.

This is where personal interpretation comes in; there is no such thing as 'correct' exposure, only optimum exposure according to interpretation, hence some people prefer to 'expose to the right' (there is much more image data recorded on the right (bright area) than on the left (shadows).

This is a good link from Sekonic, exposure meter folks http://www.sekonic.com/images/files/HistogramsLightmetersWorkTogether.pdf
 
You can use the histogram another way too.

There is no correct exposure for any picture - it depends on what you want to get from the scene in front of you.

However, the h/gram can show you how much of one tone there is. Provided it is still on the scale you should be OK - if it goes off the scale (peaks too high) they you are in trouble because that part of the scene has blown out or blocked up.
The H/gram shows a graph, basically, of the spread of light captured by your sensor. One picture where the graph is heaped up to the left, could be a lovely silhouette, but by your interpretation it is garbage, because it isn't in the middle....
 
HoppyUK said:
Welcome to TP :)

While the histogram is indeed an excellent way of determining optimum exposure, having all tones lumped in the middle is rarely optimum.
Click to expand...

Well there are many occasions when you will get almost nothing else - think of taking pics on a dull day for instance - the contrast will be reduced which will result in a restricted histogram.

If shooting in Jpegs then you can increase the contrast in the camera but I always prefer do any processing on the computer not in the camera.

For that reason I always shoot in RAW with everything (contrast, sharpness etc) turned to 0.
 
RL355 said:
Sorry newbie question…:) I am fairly new to photography and have been experimenting with my digital camera. I have a question regarding exposure and the histogram. I hope someone could explain the following….

If I take a picture of a white card filling the whole of my frame. I then look at the histogram I notice that the details are centrally plotted on the histogram.

If I take a picture of a black card filling the whole of my frame. I then look at the histogram I notice that this is ALSO centrally plotted on the histogram

My understanding of the above is due to the fact that the camera is programmed to expose a scene as 18 percent middle tone as that is what the camera thinks is an ideally exposed picture.

If I take a picture now with the frame filled with half white and half black card. The histogram shows 2 distinct peaks. One in the highlight end of the histogram and one in the Shadow end of the histogram. How come there is no peak in the middle of the histogram like you do when you took the black and white cards individually ?

thanks in advance
Click to expand...

OK just to complicate matters I am going to introduce a grey card :)

If you take a picture of a grey card the camera will think "that is an average grey card which is what I am calibrated for, therefore I will give it an average exposure". The pixels are therefore half full and the histogram peak will be in the middle which is where the half full pixels are.

You now take a picture of a white card and the camera thinks "that is an average grey card which is what I am calibrated for but someone has turned the lights up, therefore I need to give it less exposure". The pixels are therefore half full and the histogram peak will be in the middle which is where the half full pixels are.

You now take a picture of a black card and the camera thinks "that is an average grey card which is what I am calibrated for but someone has turned the lights down, therefore I need to give it more exposure". The pixels are therefore half full and the histogram peak will be in the middle which is where the half full pixels are.

Now take a picture of all 3 cards, the camera averages the scene out and as white and black and grey average to grey it thinks "that is an average grey card which is what I am calibrated for, therefore I will give it an average exposure".
In this case the pixels recording the grey card will be half full and will give a histogram peak in the middle - the same as the first example.
The pixels recording the white card will receive more reflected light and will be full, giving a histogram peak on the right.
The pixels recording the black card will receive less reflected light and will be almost empty, giving a histogram peak on the left.
So you get 3 peaks on the histogram.

If you now take a picture of just the white and black cards the camera will still average the scene and as white and black average to grey it thinks "that is an average grey card which is what I am calibrated for, therefore I will give it an average exposure".
The pixels recording the white card will receive more reflected light and will be full, giving a histogram peak on the right.
The pixels recording the black card will receive less reflected light and will be almost empty, giving a histogram peak on the left.
So you get 2 peaks on the histogram.

This tutorial by EdBray explains exposure very well http://www.talkphotography.co.uk/forums/showthread.php?t=158332
 
petersmart said:
Well there are many occasions when you will get almost nothing else -
Click to expand...

Sorry Peter, I'm with the others because I think it's a gross oversimplification to say that histograms should be in the middle. It depends what you are shooting surely? If I'm shooting a bride then she is usually dressed in white and guess what that does to the histogram? If I made sure it was in the middle I would have a lot of very underexposed images.

Same with snow scenes, the camera will try to pull white down to 18% grey so the camera's metering will be about 2 stops out.

If anyone wants experience of reading histograms, pull up your image editing software and have the image and the histogram on the same screen. You soon begin to understand that it depends on the tonality of the scene where the histogram lies.
 
petersmart said:
Well there are many occasions when you will get almost nothing else - think of taking pics on a dull day for instance - the contrast will be reduced which will result in a restricted histogram.

If shooting in Jpegs then you can increase the contrast in the camera but I always prefer do any processing on the computer not in the camera.

For that reason I always shoot in RAW with everything (contrast, sharpness etc) turned to 0.
Click to expand...

From what you have described, with a big lump in the middle and no highlight peaks on the right - if you are post processing, then adding maybe a couple of stops extra exposure will give a better result. Expose to the Right theory (good tutorial on http://www.luminous-landscape.com/tutorials/expose-right.shtml ).

However, if you are outputting straight to JPEG, then that will make everything too bright. In fact, if you shoot JPEGs only, the best guide to exposure is the LCD image - when that looks good, shoot at that and you'll get an identical print or on-screen image without any adjustment.

Those are two examples where working method, ie JPEG or Raw, can result in an 'optimum' exposure which is maybe two stops different but actually both correct.
 
AliB said:
Sorry Peter, I'm with the others because I think it's a gross oversimplification to say that histograms should be in the middle. It depends what you are shooting surely? If I'm shooting a bride then she is usually dressed in white and guess what that does to the histogram? If I made sure it was in the middle I would have a lot of very underexposed images.
Click to expand...

I didn't say that it SHOULD be in the middle - I actually said:

"Well there are many occasions when you will get almost nothing else - think of taking pics on a dull day for instance - the contrast will be reduced which will result in a restricted histogram."

I was talking about a dull day when your bride's dress will not lose the highlights and you will get all the tones rendered on the sensor.

I was also trying to answer a newbie's question without getting too technical and getting the histogram in the middle is, for a newbie, far easier to understand than getting bogged down in all the technicalities surrounding exposure and dynamic range.
 
Hi

Thanks to those who have taken the time to reply...I know the answer is right here in front of me but for some reason I still cant get my head round it at the moment. I can understand the results on the histogram when you have the frame completely filled with white and when its completely filled with black and the histogram peaks in the middle. But Iam still not getting my head round the fact of the two peaks when Black and white is in the same frame.

My thinking process is as follows...When the camera takes a scene with the frame filled with an equal amount of black and white. The camera will interpret the whole scene as 18 percent middle tone and will expose it accordingly as an 18 percent middletone as that is what the camera thinks is an ideally exposed picture as its been calibrated this way.

So the blacks will be made to go lighter to be 18 percent middle tone towards the cameras calibrated ideal and the whites will be made to go darker to be 18 percent middle tone towards the cameras calibrated ideal and the histogram will show black and white tones as middle grey in the centre of the histogram. Thats my logic at the moment I know what my thinking is wrong. I cant see how the two peaks arises from this on the Histogram. The two peaks result is the correct answer as thats whats its telling me on the camera histogram :)

I understand tonal range of the histogram. ie any black tones will be plotted on the left hand side and any whites will be plotted on the right hand side anything 18 percent middletone will be centrally plotted.


Suvv from what you've said could you explain the part highlighted in red or anyone else. I know I am very close to understanding this.

If you now take a picture of just the white and black cards the camera will still average the scene and as white and black average to grey it thinks "that is an average grey card which is what I am calibrated for, therefore I will give it an average exposure".
The pixels recording the white card will receive more reflected light and will be full, giving a histogram peak on the right.
The pixels recording the black card will receive less reflected light and will be almost empty, giving a histogram peak on the left.
So you get 2 peaks on the histogram.
Click to expand...

many thanks
 
I think you are making hard work of this!

Your thinking sound right, and as I said earlier if you scrambled everything on the histogram you would get 18% grey. That's how hand-held exposure meters work because they do not have the benefit of a focused and framed image. Histogram is different.

The histogram reads the actual finished image. It knows just how many pixels are exposed at which levels (and colours) and that's what it is presenting. It is simply saying this is what you've got, and it's calibrated to suggest that this is the best exposure according to its parameters. It might not be optimum though, that is your judgement.

Also be aware of the different metering modes, which give different emphasis to different areas of the scene. In particular, evaluative/matrix metering is programmed to recognise certain situations and will not necessarily react in quite the way you expect from your theory. It has a degree of intelligence in that respect and will recognise subjects with extremes of dynamic range. For example, if it sees a lot of very bright light at the top of the frame, it will assume this is sky and largely ignore it, preventing under exposure. It's actually quite clever.
 
RL355 said:
Suvv from what you've said could you explain the part highlighted in red or anyone else. I know I am very close to understanding this.

many thanks
Click to expand...

If you now take a picture of just the white and black cards the camera will still average the scene and as white and black average to grey it thinks "that is an average grey card which is what I am calibrated for, therefore I will give it an average exposure".
The pixels recording the white card will receive more reflected light and will be full, giving a histogram peak on the right.
The pixels recording the black card will receive less reflected light and will be almost empty, giving a histogram peak on the left.
So you get 2 peaks on the histogram.


What I am saying is that with just a white and black card the camera sees an average of grey - I think you understand that.
So the exposure would be correct to give a central peak for any average grey in the picture. However there is no grey only white, which is lighter than grey and black which is darker than grey.

In other words more light is being reflected from the white card than a grey card would have reflected and less light is being reflected from the black card than a grey card would have.

In your 3rd para you are still treating the white and black cards separately, you have to consider them together as the camera will average them. The meter does not see a white card and a black card it sees an overall average of grey.

I know what it is like when you get stuck on something like this - once the penny drops you wonder why it seemed so difficult to understand. :)
 
Thanks Suvv.... for your further explanation in a very easy to understand way I think I am making sense now of what is happening and where I was going wrong :)

If there is only ONE tone in a scene the camera will always expose to make it a middle tone (18 percent middle grey tone)as that is what its been calibrated to do by adjusting settings available to it like ISO, Aperature and Shutter speed etc…and that is why the peak is in the middle of the histogram.
So a snowy scene which is predominantly white will be marked down to Middle grey tone hence why the snow is always grey looking. In the same way a Black cat filling the frame will be grey looking as its been marked down to a more middle tone.

When TWO or more tones are involved in the same scene (Black and white card filling the frame). The Scene is AVERAGED to grey for a middle tone exposure. As a result all the settings like exposure, ISO and aperture etc is SET for exposing this middle tone. However like you said there is only Blacks and Whites in reality. The white will be lighter than middle tone (reflecting greater than 18percent light ) will be recorded as Highlights and will show as peaks on the right side of the histogram . Anything darker than middle tone (reflecting less than 18 percent light )will be recorded as Shadows and will show as peaks on the left side of the histogram.

Understanding this is making other things a lot clearer and its a real stepping stone for me.


Thanks and merry Christmas and new Year to all.
 
Yes - glad you have got it.

Have a good Christmas & New Year
 
You must log in or register to reply here.
Back
Top