Join Taz Tally for an in-depth discussion in this video Using histograms, part of Photoshop Color Correction: Fundamentals.
In this video I'd like to dig in a lot deeper, on how to really read and evaluate histograms. Because this is going to be, really one of our key tools for guiding us as to how we're going to, correct our images. So we've got a variety of images here. We've got five images to take a look at, that have a wide range of different histogram distributions. Let's go right over to Photoshop. And dig in, and take a look at each one. Let's start with this lighthouse image here. And, we've got the histogram palette set up here. And I've got this set up to look at the composite RG, but just look at the tonal data in the image.
And we're not looking at all the channels, just the composite, expanded view and large view, of just the overall tonal data in the image. First, we're going to look at just tone, and then we're going to look at the distribution of tone on all three channels. So this is a composite of all the tones. When we look at this image, we see just visually, this is a fairly dark, or low-key image and it's got low contrast. And, when we look at this, you can see well, there really aren't many highlight pixels in here at all. When we look at the histogram, it shows us exactly where the data is in this image. And the histogram coincides perfectly with the image itself.
Now, when we read a histogram. Starting from the right hand side, this is pure white, alright? This is pure black. So this is highlight, shadow. Darkest shadow, lightest highlight. Midtone is right here in the middle. Quarter tone, and then three-quarter tone. When we look at this histogram, we see there's almost no data from highlight to mid-tone. All the data in this image is from mid-tone to shadow. And that's why it looks dark. And that's why we call this a low key image. And that's why the contrast is low. Because you don't have a lot of difference in tonal data. There's only, you know, fifty shades of gray on a scale of zero to one hundred.
Or hundred and twenty eight shades of gray on a scale of zero to two fifty five. Only half the potential tonal value is there, so it's a relatively low-contrast, low-key dark image and histogram tells you why. And you can see in the histogram, boy, it starts right at about 45%, doesn't it? It ends just about at 99%. So we can get a semi quantitative sense. We're not actually looking at numbers, but by looking at this graph and knowing this is zero and this is 100% grey, we can get a real sense for where the values are. All right, let's put that one away and let's go to the next image. And this is an image of some clouds.
And when we look at the histogram of this, we see, well, it's almost not quite the exact opposite but this is a high key image, right? It's not a low key image and we see we've got some very bright portions of the image. We've got some mid-tones. Probably not super dark, even in the shadow portion. You know, there's not a lot of dark when you look at the histogram. Again, the histogram agrees and provide us with a semi quantitative data distribution of where the data is in this image. This starts just shy of the lightest highlight, and goes to, oh, about 60, 65%, and then from there to the shadow, there's nothing.
So this is a relatively high key image. And we can see the data is distributed from about 10%, grey down to maybe 60 or 65% grey, OK. Lets take a look at the moose. And look at the distribution of data in this image. When we look at this image, just visually, we see this got a little bit more contrast in it, but still not a super contrasty image. And we can see detail throughout this image, but there aren't really any super dark areas, or any super light areas.
And we look at the histogram, the distribution of data, well we see again, the histogram really tells the story. It really gives us confidence to understand where the data is in this image. Most of this data is piled you know between forty and 60% grey. It tails off to about 70% or so. Maybe up here to maybe about 25 or 30%, but these long flat lines here that you see, is usually noise in the images not actually data in the image. It's usually noise, and by the way this is a very typical histogram that you see in images that are shot in low light. Early in the morning, late in the afternoon, you don't have a lot of bright highlight or deep shadows.
And you end up with a relatively you know medium contrast image with no highlights and no shadows in it. We're getting the sense of, how important and how useful histograms can be in evaluating images. All right? This next one is of shot, I made on the Harding Ice Field. Up at about forty four, forty five hundred feet. Looking eyeballed, eyeball with the beautiful Harding Ice Field in Alaska. With some hikers that were just on a ridge, just in the foreground. And we look at this image visually you see lots of bright, you know, white highlight areas.
And then we've got some deep, dark shadow areas. And not a whole lot in between. We've got some darker, you know mid-tone areas here, but most of the data is going to be on the high light to mid tone end and then probably some pretty deep shadows. And sure enough, when we look at our histogram we see there's a very bimodal distribution of data. We have a relatively fat peak of data from the white highlight which is in the lightest areas of the image, and down into the mid-tones which is kind of like in here. And then we kind of have a flat area with not too much data. And then we've got a couple of bumps with the big area down here, and of course this is the shadow portions of the image.
When you look at a histogram and you knew it's like this means, you start comparing them back and forth, you begin to see how you can match up areas on the histogram with areas in the image, where these bumps down here are the darkest areas of the image. These bumps on the shadow rend, and this broad highlight area that you see highlight to, to just about mid-tone. Mostly it's the highlight in quarter tone. Or all the very light clouds, and down into the mid-tone area of the snow that's in the shadow. So it makes perfect sense, doesn't it? When you look at the image and then look at the historgram. Alright.
And then finally, let's take a look at the, Shark. And when we look at the shark, we see, this is a whole different animal, pardon the pun. From the other images. And yet, this one has a very, very obvious color caste to it, doesn't it? And when we look at the histogram, just the tonal histogram of this image. We see that there's data, oh, from about 10%. Alright, there may be there's a little bit of data on the pure highlight, then kind of mostly noise, and then the data picks up at about 10%. And then we've got peaks here, here, and here.
We're not sure what that's about. And then the data stops at about hm, 75%. When you look here, there's really nothing it's truly black, right? All right? So, we've got a pretty good distribution of tonal data, but really, nothing from the three-quarter tone to shadow. But the real story in this image is told by, let's look at all the channels and then let's look at this, master histogram and channels. now we got, really start to see the big story that these histograms can tell us. When we see an image like this, it has a very obvious.
No matter what monitor you're looking at this on, it's going to look blue, green, even if it's not calibrated at all. It's going to look blue, green. And so far we've been talking about tonal distribution in the image. Right now, let's talk about colors. And what we use histograms for is not only to look at where the total distribution of tone is. But most importantly in color correction is, where are the color channels? Where is the tone on the color correct channels? And notice we have a huge offset of blue, which is offset way to the right of green, which in turn is offset well to the right of red. And that's what creates this very strong blue, green color cast.
And we tend to look at the highlight areas of the image, right? We look at the highlight end to see what's the closest color? Boom. And that's very often going to be the color cast. In this case, blue and then green and then red. So we can very clearly match up the histogram, the offset of the blue to the green to the red to account for the very strong color cast that we see in the shark. We're going to take a lot of advantage of this offsetting of histograms to help guide us in our color corrections. Alright, now what I'd like to do is just take a quick trip backwards, here now that we've got all the histograms shown and let's look at these individual images again.
The lighthouse. When we look at the lighthouse now that we've talked about offsetting of histograms can you look at this multi-channel and the display of colors in the individual channels and say. Hey, not only is this a low-key image with no data from highlight to mid-tone, do the offset of any of these histograms indicate a color cast to you? Well, you can look here and you can also look here and see that the blue is indeed offset to the right. So it might not have been apparent when you first looked at that image, but you bet, this has a blue color cast to it. When we look at the cloud image. And again, we can see data that is offset not so much in the highlight end.
The highlight end is hm, pretty consistent. So there's really doesn't have much of a color cast in the highlight end. compared to the other images. Let's look at the moose. Moose is not too bad. A little bit of offset of red data, compared to the green and the blue as we would expect in an, shot that shot in late afternoon, we expect it to be a little bit warmer. So a bit of a color cast here, but is it a problem? Is it something that needs to be corrected? Perhaps not. We look at the Harding Ice Field image, and again when we look at the offset of histograms we see, oop there's a pretty strong blue offset here.
And when we first looked at this image, did you think that this had a blue cast to it? You perhaps didn't, but when we look at it now and we look at the histograms we say you know there is an overall blue cast to this. And this is why we need to use quantitative or semi-quantitative methods for evaluating our color. We're not using numbers in this course, that's for a more advanced course, but here, we can use our histograms and that very clearly shows us in a semi-quantitative graphical way that we have a blue color cast. So even if our eyes are lying to us. See, our mind and our eye wants to see that as white, so without something to compare it to, and we'll get back to that, we really don't know what the color cast is.
And then finally, just to return to the shark image again, and we see the huge color cast that's in this image. So, we can see the total distribution of tone and image, and then by looking at the offset and the position of relative histograms on the individual channels. We can see color casts in these images and they're going to guide us as to how we should correct them.
- What is color correction?
- Comparing RGB and CMYK color modes
- Using grayscales and neutrals for color correction
- Understanding pixels and bit depth
- Evaluating and correcting images with histograms
- Using nondestructive editing tools
- Removing a color cast
- Performing curve corrections in Camera Raw
- Affecting creative adjustments
- Retouching an image
- Sharpening images
- Preparing for print and web use