NEXT Tutorials

But – what is “local contrast”? I guess you all know about contrast. A loose definition is that contrast is connected to the ratio of the luminance of the brightest point to that of the darkest point in an image. For instance, if I show you a picture like this one, you’d have a hard time stating that it has good contrast. The real question is – can this be salvaged? Well, of course it can – and there are several ways to do it. I’ve placed two samplers in the lightest and darkest significant points of the picture. We are in grayscale here, for the sake of simplicity, but the same holds when we deal with color pictures, as well. One of the most important sources of information for us is the info palette: if you read the values under the samplers, you will discover that the darkest one has a gray density of 84% and the lightest chimes in at 24%. The extreme values are, of course, 100%, full black, and 0%, full white, respectively. This means that in this case, we are missing about a third of the contrast available. As I said, you can fix that in several ways, but my favorite method has a name, and the name is “curves.”

So let’s open the Curves dialog box and move the endpoints of the curve (a straight line, actually) so that the shadow falls in the range of 95% and the highlight around 5%. I won’t go all the way because I do want to spare some margin for further maneuvers that may be needed, but this is not so important right now. OK, done: do you see how much the image has improved? It definitely has more snap and more appeal. And we may go further: if we decide that the light parts of the picture are the most important, we may grab the curve, say, here, and steepen the area of the highlights a bit. The principle, as stated by the great Dan Margulis is simple: “the steeper the curve, the more the  contrast.” So, more improvement: but there is a trade-off… when we steepen the curve in the quarter-tones, we also flatten it near the shadows, so we gain contrast in the quarter-tones at the expense of a contrast loss in other parts of the picture. Well, the best things in life don’t come free, after all.

But what I need to say right now is that this is a global intervention: you haven’t seen me selecting anything, I am not using masks, and so on. This means that I am working on the image as a whole, rather than a section of it. For instance, look at this area here, and let me select it with the rectangular marquee tool. Curves again, and let’s boost the contrast… nothing fancy, I just need a steeper curve of some kind to prove the point. OK – look: I now deselect the area, so you can see the obvious contrast boost where I operated locally. It looks fantastic, but there is a problem. Well, two problems: first, the curve you’ve just seen would have ruined the image completely if I had it applied globally because it would plug the shadows and wash out the highlights – trade-off, remember! Second, even if I were insane enough to select area after area and operate locally painstakingly, the result would be appalling, because there is no serious way to deal with the borders of the selected area. Result? A series of fantastically contrasted tiles looking like a checkerboard – oh my. So, end of the story, no way, basta, fine.

I am not going to show you how NEXT works, right now, but I’ll show you what it can do. This file has two layers. The one at the bottom is identical to the one you’ve just seen – the original picture with a curve applied. On top, you have the result of such layer processed with ALCE. The question is: could you get to a similar result by using traditional methods like overlays, hiraloam unsharp masking and similar techniques? If you do, give me a ring – because I know a few tricks, but I would find it extremely hard to get there. And I definitely couldn’t do it in a few seconds. NEXT can. The purpose of this and the next video is to show you how. But before we get into that, we need to learn about the nuts and bolts of NEXT. So, let’s start from the beginning.

So what is the radius about? Well, when you think “radius” you automatically think “circle.” Not in our case: what we call “radius” is half the side of a square. When we say “radius 50”, for instance, we mean “a square with a side of 100 pixels”. I’ve prepared a file to show you roughly what happens. Do you remember when I was suggesting that curving a certain selection in an image would allow you to maximize the contrast therein? It’s more or less what ALCE does, but it does it in an incredibly sophisticated way. In this file, I’ve created three layers containing checkers which correspond to certain radii. Let’s examine the of the image, first. Image Menu, Image Size… there you are: about 1300 pixels wide and about 850 pixels high. Keep these figures in mind, for now. If we imagine to cover the image with selections of radius 10, that is 20 pixels wide, this is how it would look like. Please notice that each square over the leaves contains just a fraction of them. Encompassing one leaf with one of these squares would be impossible. Here’s what the selections would look like at radius 50, that is 100 pixels wide. Still, each leaf contains more than one selection, but the area of each selection is much bigger. At radius 100, this is what would happen: selection is roughly as large as a leaf.
The difference between a leaf and its veins is mainly one of scale. If your aim is to enhance the veins, then you should realize that they belong to the fine details and that their variations occur over an area of one or a few pixels. If you choose a radius of 100 pixels, you’ll be let down. But – if you choose a radius of 100 pixels you’ll be dealing with a much bigger area, and you’ll actually be able to re-allocate the luminosity structure of the leaves as objects, and the veins will be largely ignored. We call the fine detail “high frequency” and the large structure of luminosity “low frequency.” The larger the radius, the more you’ll ignore the latter, and enhance the former.
Let’s see how we translate in reality. I’ve prepared three layers treated with ALCE at the radiei mentioned above: this is 10, and you’ve already seen it. This is 50. And this is 100. Spot the difference? Again: original… ALCE 10… ALCE 50… ALCE 100.
The two larger radii are not as impressive as radius 10, here, but they actually do something interesting. So, what radius you will use? It depends: you should first think how big is the detail you want to enhance. It will take you a while to be able to guess correctly, but this is roughly similar to the choice of the radius in the unsharp masking technique. And – very important: you may ask yourself why, a few minutes ago the result of radius 50 on this same picture in grayscale was so incredibly impressive, and here it is… well, different. Well, the simple answer is: I cheated. It’s not the same picture. This one was rescaled to one-third on the previous version. So, the relative size of the objects in pixels is different. A radius of 50 there is equivalent to a radius of about 17 here. And, indeed, radius 10 is much closer to that result than radius 50.
So, two rules: first, consider how big your image is. Second, consider how big your areas of interest are on the image.

I would like to address an issue which has been discussed a lot. That is, I would like to reply to the question: does ALCE desaturate colors?
There are two seemingly contradictory answers to this, although they are both correct – and they are: in fact, not; perceptively, sometimes.
Let me show you an example. I’ve chosen this photograph because it isn’t certainly short of colors. This is the original, and on the top layer, I’ve put an ALCE-processed version with radius 50. As you see, there’s a serious impression that a lot of color was lost along the way. That is, the ALCE version looks less vivid than the original. You’ll be very surprised when I show you the middle layer, then: it contains the color of the processed version and the luminosity of the original. As I turn it on, you would expect to see the less colorful version. Alas, not: I can’t perceive any difference, and if there’s any I would consider it completely irrelevant.
So, why does the original ALCE layer looks less saturated than the unprocessed image? The answer is: it’s a side-effect of the re-allocation of luminosity. The saturation undergoes a perceptual change because there are lighter and darker areas which force us to believe we are looking at a different color. But from a practical point of view, this is not true.
Our advice is simple: if you like the result you obtain, don’t do anything. Otherwise, if you feel you need a bit more vivid colors, just open a Hue/Saturation adjustment layer and push up the middle slider, that is saturation, a bit. How much? Very difficult to say: +10 might be a good starting point, yet it will be too much in some cases and too little in others. By all means, let your eyes be the judge.
We also suggest that you do NOT pre-saturate, that is use Hue/Saturation before ALCE, because in very vivid pictures this may kick some of your colors out of gamut, and cause unpleasant artifacts. ALCE should come first, and a little make-up might then come after – if it’s at all needed.