Reading a histogram

LBPhotog

TPF Noob!
Joined
Mar 3, 2010
Messages
372
Reaction score
0
Location
New Hampshire
Can others edit my Photos
Photos NOT OK to edit
Ok ... so this is a more than a basic type question - I hope ...

I know that with a histogram shadow detail is on the left side of the scale and that highlight detail is on the right. I also understand that you are looking for a nice "bell curve" type display (no loss of details in the shadows, no clipping/blown out highlights). Is that all there really is too it? Or, is there more to it than that? I feel as this is an overly simplified explanation of the histogram; but, is it really THAT simple?

I've also noticed that when shooting on black or white drops this sways the grid on the histogram in my camera one way or another (depending on the color of the drop) ... when shooting on drops should I put more emphasis on my flash meter and the exposure I am getting from that and kinda disregard the histogram ...

I'm a little confused - I THINK ... can someone set me straight on this; or, am I completely ok with my understanding and just think I am confused - which is always quite possible. :lol:
 
I think an entire book could be written about the histogram. It gives you so much information about a scene. As you've already stated, one of the most obvious uses is ensuring you're not clipping shadows or highlights. The supposed perfect bell curve histogram is largely a myth. From a purely technical standpoint, a bell curve will roguhly estimate an average of 18% gray through your scene. The problem is, this isn't always what you want.

I'm not very good at explaining things, but here is a good place to look for information on the histogram. The author does a pretty good job explaining it, what it tells you and what it doesn't tell you. One thing you have to understand is, there's no way to tell if an image is properly exposed just by looking at the histogram. It can give you some information not readily available just by looking at the image, but it's fairly useless without the image to look at.
 
I think you have the right idea....except that you don't always want a nice even bell curve. For example...as you mentioned with white or black backdrops, your histogram will most likely be skewed to the left or right...and that's OK because it likely means that your background is dark/bright.

Keep in mind that the histogram is just a graphical representation of the pixel in the image. If there are a lot of dark pixels in the image, the graph will have more weight on the left side. If you want your image to have a large portion of dark areas, then you would be looking for the histogram to be left heavy.

I think the more important thing to look for, is clipping. When the graph is clipped on either side, it likely means that you have lost detail in either the shadows or the highlights. You typically don't want a lot of lost detail in an image...but if the representation of your subject falls within the graph, then it may not matter if parts of the background, for example, are clipped...because those area are less important to the image.

In other word, it can be helpful if you can look at the graph and recognize where your subject is represented.

Understanding Histograms
Expose Right

*edit*
Also, keep in mind that not all histograms are created the same way. It might be based on overall luminance, or it might be based on one, two or three color channels. For example, it's not uncommon to see a histogram that is actually three graphs overlaid...red, green & blue. Depending on the image, one of the colors may be way out of whack compared to the others...and that may or may not be important to the image.
 
Something I just thought about, and the reason you don't always want the bell curve is this. Let's say you're shooting a black background. You look at your histogram and you see it skewed to the left. Well, if you adjust to put that information in the middle, you're going to render your black background as gray. Same thing with a snowy scene. If will be skewed right, and if you adjust to be in the middle, the snow will be rendered gray. This is why most people will say to shoot with +1 to your exposure compensation when shooting in the snow, because the camera wants to render the snow as 18% grey.
 
ok, so when shooting under strobes my flash meter should trump my histogram ... I understand the concept here; but, it seems to me, from what I've read here from what you guys have posted it seems as though the histogram is just another tool in our arsenal to expose properly ...
 
ok, so when shooting under strobes my flash meter should trump my histogram ... I understand the concept here; but, it seems to me, from what I've read here from what you guys have posted it seems as though the histogram is just another tool in our arsenal to expose properly ...

I wouldn't say it should trump the histogram. They are both tools that have their own uses. Experiment and try things. You will likely come up with a method of using all your tools together to determine proper exposure.

For example, when I shoot landscapes, I'll usually start in manual mode, but my first shot is taken exactly how the meter tells me to. I look at the histo and I can now decide how I want to adjust my exposure to get my desired shot. It usually requires 3 or 4 frame for me to get my exposure dialed where I want it.

One thing that wasn't mentioned is that the histo on your camera is based on the jpeg preview. So there is data that was thrown out during the compression. I learned this the hard way when I was first learning how to "expose right." I adjusted my exposure as far right as I could without blowing out highlights. However, even though the highlights weren't blown on preview, when I opened the RAW files in ACR when I got home, I had blown highlights where I didn't want them. Now when I expose right, I go to the point where highlights aren't blown on the preview, and then dial another 1/3 of a stop back. Haven't blown highlights since (except where I intended to, anyway).
 
ok, so when shooting under strobes my flash meter should trump my histogram
The flash meter is actually measuring your lights, the histogram is giving you a representation of the resulting image. The flash meter should be used to find your exposure and to find & set your ratios. The histogram can tell you if the photo falls within the dynamic range of you camera...it's more of a 'check', while the light meter is the better tool for finding your exposure.

One thing that wasn't mentioned is that the histo on your camera is based on the jpeg preview. So there is data that was thrown out during the compression. I learned this the hard way when I was first learning how to "expose right." I adjusted my exposure as far right as I could without blowing out highlights. However, even though the highlights weren't blown on preview, when I opened the RAW files in ACR when I got home, I had blown highlights where I didn't want them. Now when I expose right, I go to the point where highlights aren't blown on the preview, and then dial another 1/3 of a stop back. Haven't blown highlights since (except where I intended to, anyway).
That sounds like tho opposite of what I would think would happen. The RAW file should have more information and thus more highlight detail than the JPEG.

The one example that I remember was when I shot a large moth in Costa Rica. I had a flash set on manual, and unexpectedly had a close up shot of this moth. The flash completely blew out the both...clipped the histogram and it was flashing on the review image. Back home editing the RAW file, I was able to pull most of the texture detail back into the wings...although much of the color was gone.
 
ok, so when shooting under strobes my flash meter should trump my histogram
The flash meter is actually measuring your lights, the histogram is giving you a representation of the resulting image. The flash meter should be used to find your exposure and to find & set your ratios. The histogram can tell you if the photo falls within the dynamic range of you camera...it's more of a 'check', while the light meter is the better tool for finding your exposure.

One thing that wasn't mentioned is that the histo on your camera is based on the jpeg preview. So there is data that was thrown out during the compression. I learned this the hard way when I was first learning how to "expose right." I adjusted my exposure as far right as I could without blowing out highlights. However, even though the highlights weren't blown on preview, when I opened the RAW files in ACR when I got home, I had blown highlights where I didn't want them. Now when I expose right, I go to the point where highlights aren't blown on the preview, and then dial another 1/3 of a stop back. Haven't blown highlights since (except where I intended to, anyway).
That sounds like tho opposite of what I would think would happen. The RAW file should have more information and thus more highlight detail than the JPEG.

The one example that I remember was when I shot a large moth in Costa Rica. I had a flash set on manual, and unexpectedly had a close up shot of this moth. The flash completely blew out the both...clipped the histogram and it was flashing on the review image. Back home editing the RAW file, I was able to pull most of the texture detail back into the wings...although much of the color was gone.

Interesting. All I know is I had no "zebra stripes" on my preview, but when I got home, there were whole sections of the image flashing. Highlight recovery got some back, but most were gone. I'm thinking I need to test this out again for myself. I might have made a mistake or something originally. It'd be nice to have that extra 1/3 stop to the right again if I am, in fact, wrong.

The more I think about it, the more I realize that what I saw shouldn't have happened. The jpeg will clip long before the RAW file would. I must have done something wrong, though I can't imagine what. Back to the drawing board. :)

This still doesn't change my original advice. It's important to keep in mind that the histo on the preview is of the jpeg, not the RAW, so they will end up a bit differently.
 
Also, as I hinted at earlier, the histogram you see on the camera, may not be telling the whole story. It maybe be just the green channel, for example (I don't know if they actually use only one channel). So while you may look at the histogram and thing that you are not clipping...the other two channels may actually be blown out.

The same scenario might happen if the displayed histogram is an average of the three channels. One or two of them may still be clipped, even though the graph doesn't show that.
I believe that some cameras can be set to show you a three channel histogram.
 
Also, as I hinted at earlier, the histogram you see on the camera, may not be telling the whole story. It maybe be just the green channel, for example (I don't know if they actually use only one channel). So while you may look at the histogram and thing that you are not clipping...the other two channels may actually be blown out.

The same scenario might happen if the displayed histogram is an average of the three channels. One or two of them may still be clipped, even though the graph doesn't show that.
I believe that some cameras can be set to show you a three channel histogram.

Certainly something to think about. I have an XTi and I think I can either have the luminance histo showing or the color channels, but not both at the same time (not totally sure on that, will have to play with it when I get home). Thanks for the info there. It shows even veterans have a lot to learn about their chosen craft. :blushing:
 
If you have your histogram set to luminosity (sic?) what you see is the highest level of all the channels, not an average. Some cameras only have this setting. If you set it to RGB, you get a color histogram with the three channels laid on top of each other. Generally, you can see if one of the channels are clipping. Some cameras will also show a histogram for each channel and this is the easiest way to see if one of the channels is clipping.
 
Ok ... so this is a more than a basic type question - I hope ...

I know that with a histogram shadow detail is on the left side of the scale and that highlight detail is on the right. I also understand that you are looking for a nice "bell curve" type display (no loss of details in the shadows, no clipping/blown out highlights). Is that all there really is too it? Or, is there more to it than that? I feel as this is an overly simplified explanation of the histogram; but, is it really THAT simple?

I've also noticed that when shooting on black or white drops this sways the grid on the histogram in my camera one way or another (depending on the color of the drop) ... when shooting on drops should I put more emphasis on my flash meter and the exposure I am getting from that and kinda disregard the histogram ...

I'm a little confused - I THINK ... can someone set me straight on this; or, am I completely ok with my understanding and just think I am confused - which is always quite possible. :lol:
Part 1: Understanding Digital Camera Histograms: Tones and Contrast

Part 2: Understanding Digital Camera Histograms: Luminance and Color

It's my understanding that if you capture images as RAW data files, the histogram the camera shows you is actually from the JPEG/Basic that is embedded in the RAW file.
 
If you have your histogram set to luminosity (sic?) what you see is the highest level of all the channels, not an average.

No true. When looking at a luminosity histogram you see a luminance function. The channel detail gets weighted according to the way the eye perceives the colour's brightness. I.e. two identically pure colours (green and blue) light with exactly the same light source would show up not as two spikes right in the middle (assuming they are lit to be around the middle of the histogram). The blue and green channel may show this, but the luminance channel would show a spike down low for the blue and a spike up high for the green given how much brighter an identically perceived value of green is.

I also understand that you are looking for a nice "bell curve" type display./QUOTE]

NO NO NO NO NO!

You are never looking for a nice "bell curve" type display. A histogram just is. It just shows what the image is doing. For instance there is no way you can ever take a photo of a sunset, or the moon, or the beach or a heck of a bloody lot of things that fits in a bell curve.

You may expect a certain histogram look, but you should never go out trying to make your images fit the bell curve. The histogram is a debugging tool, not a planning tool. Use it to check to see if you've clipped the highlights when you don't want to, or use it on a bright day when you can't tell if the image you took was exposed right.
 
Interesting. I wasn't aware of the weighting factor. Makes since it would be done that way though.
 
It is the difference between luminosity vs brightness value. Try it in photoshop. Make an image with two very pure yet very different colours (yellow and purple are a great choice) and then on the histogram compare the "RGB" setting with "Luminance"

"RGB" is the system you were talking about.
 

Most reactions

New Topics

Back
Top