Log in

No account? Create an account
31 January 2009 @ 11:09 pm
Tutorial: Cropping, Leveling, and Sharpening  
A Few Basics: Cropping, Leveling, and Sharpening
(in photoshop)
And Using Them on Less Than Perfect Pictures

Let's start off with a few links to other basic tutorials that might help:

How to Crop an Image in Photoshop

Image Resolution Tips  (helpful to know when cropping)




All of these techniques are well and good if your image is fairly sharp and clear and normally colored to begin with, but . . . Well, how often do you stumble across perfect images when you're working with screencaps or manip source pics?  Not often enough, damnit.  But there are ways to manipulate less-than-perfect images and fix them up using the above techniques, and that's what we'll be discussing today.


Now, obviously cropping is good for a lot of things, for instance, getting rid of those awful station logos that pepper some screencaps (unless they happen to be in the middle of your picture, in which case you'll need a clone brush and some patience, but we'll talk about that later).  However, another thing to know when cropping is a little about composition.  Last month our magnificent katekat1010 explained the rule of thirds as part of her wonderful discussion, and that is an excellent place to start.

We'll use this picture as an example . . .

(click for full-sized image)

We now have a lovely (and scary) screencap from Doctor Who, but there's a lot of stuff in there that's entirely unnecessary.  I'm going to use a 300 px by 300 px, crop so you can see what I'm doing, but it's not too huge, however these principles apply to any size crop.  Now, we could just crop it with the statue in the center, which would give us . . .

This crop does have some strengths, but it has weaknesses, too.  Just about anything will, the trick is to figure out exactly what you're trying to convey with the picture.  In this one, the sharp upper teeth on the statue lay along the top horizontal line and the claws occupy the middle squares, which gives them a sort of reaching toward the viewer look.  However, there's nothing at the lines' junctions and so there's no real direction.  The viewer has to take the picture in all at once and you'd have to use other elements of the design to define the direction of the composition.  (Direction is, according to Wiki: "visual routes which take vertical, horizontal or diagonal paths."  Which means that that's where the viewer's eye begins and where it then moves to, which we can manipulate by creating focuses.)

So, let's apply the rule of thirds and focus on those vicious looking teeth, as an example.  I like to crop an image more than once so that I can see the final products side by side and figure out which one I like best.  (To do this, crop the image, then go to Image --> Duplicate.  This way you can go back to the original screencap, undo the crop, and crop it again.)

Focusing on the teeth, here are some possibilities . . .

Strengths: The teeth and the clawed hand both lay at line junctions; the light behind the head and the hand will also draw the eye to those areas; the position of the hand, partially in the middle line of horizontal sections, gives the feeling of reaching movement; and the statues eyes are bisected by one of the lines, which draws the eyes up to them.

Weaknesses: The lower half of the picture is considerably less interesting than the top (remember there are two focal points there, too); we don't really get to see the wing at all; and the statue takes up the whole picture (vertically) for the middle strip and that makes it look a little more elongated than it should.

Neutral, but good to note: the picture concentrates mostly on the darker side of the statue (Which would matter if you were, say, planning to use this in a wallpaper with textures.  Many of the layer settings (darken, multiply, exclusion, etc.) will effect lighter or darker bits of the picture differently.)

Strengths: Again, the teeth are at the junction of two lines; the clawed hand is reaching forward, too; this time we have more of the wing; one of the bottom focal points pulls the eye down to the wing;  there's still light behind the head and the hand; and we still have the reaching effect; the right vertical line still bisects the eyes.

Weaknesses:  The figure still seems elongated because we don't have anything above its head; the brightness of the wing makes the image a little disconnected, especially with that dark shadow over the statue's neck.

Neutral, but good to note: This background has more color than the previous one; this crop focuses more on the lighter half of the statue.

Strengths: Here the teeth are literally the central point of the image, framed exactly in the middle of the lines; the eyes are close to, and parallel to, the upper horizontal line; there is light behind the focal points; and the hands still have that reaching effect; the hand on our right is framed in the middle of the right middle box; there is a better balance of light and dark; the figure doesn't look so elongated with space above its head.

Weaknesses: None of the focal points exactly hit any given spot; it's very straightforward, and so looks rather posed (which I think takes away from the scary/creepy); the viewer is forced to take it all in at once.

Neutral, but good to note: One of the hands is a bit clipped, which will make blending the image into a large background a bit awkward.

Which of these is the best crop is largely personal opinion, and it depends on what you want from the image to begin with.  The rules of composition are just guidelines, but in knowing them consciously we begin to use them to their best effect for our purposes.  Here's a link (to Wiki!) for more on the basics of composition.


Now, what we're going to do here is something that's touched upon in the tutorial I linked to above, but the tutorial doesn't really give you an idea of just what you can do with it, so I'm going to try.  If you're a heroes fan, you're probably familiar with the ubiquitous blue screencap.  They look a lot like this screencap of Isaac. . .

(click for full-sized image)

. . . and unless you've got a yen to make some very blue icons, they're a little useless.  There are various ways of dealing with this.  You could use a filter of the opposite color (create a new layer, fill it with the opposite color--in this case orange--and set it to Color Dodge) to counter the blue, which sorta works.  It gives us this . . .

It still needs adjusting, though, and you'd have to mess around with selective colors until you got it looking more natural, and really who wants to put in that much work?  You could also use Variations (Image --> Adjustments --> Variations) which will let you add more or less of six colors (green, yellow, red, magenta, blue, and cyan) and to make it lighter or darker.  You can get some pretty good results, something like this . . .

. . . But I don't like that method, for a few reasons.  Sometimes it produces a lot of large grains of color that you'll have to smooth out later.  (Though, you may have to do that anyway, since we're talking about pictures that are less than perfect, but we'll cover that later, too.)  And I don't think Variations is very versatile.  You can't do it as an adjustment layer (which means it makes permanent changes to the original picture), and it can't be adjusted as you go along (which, given that I love textures, I usually have to do.)  Plus, it's likely that you'll still have to fine tune the color with a selective color adjustment layer (Layer --> New Adjustment Layer --> Selective Color) and why make more work for yourself?

You can get very good results simply by adding a color balance adjustment layer ( Layer --> New Adjustment Layer --> Color Balance), which is very adjustable, and will definitely give you more natural coloring, like this . . .

. . . but it doesn't let you adjust the contrast or brightness of the original image.  For that you'll still need to add a level adjustment layer, (or adjust the brightness some other way) so . . . Why not just do it all in the same place?  I prefer just to use levels right off the bat.

You can make a level adjustment layer ( Layer --> New Adjustment Layer --> Levels), which makes it a more flexible method.  It splits things into each of the basic colors (red, blue, and green), and it's easy to tweak later on, because it's all right there in the same layer.  You don't have to go searching for each layer used to make the color changes.  (It gets confusing enough in there as it is, right?  *G*)

The way you would do this would be to open the blue screencap (or any screencap in which one color predominates) and then create a level adjustment layer (you can use the menu commands given above, or at the bottom of your Layers tab there's a circle that's half white and half black, if you click on that it will give you a menu of possible adjustment layers you can add). 

What you do with levels is going to be different for each picture, but here's what I did with this one, to give you an idea.

First, I went to the Channel menu at the top of the Levels box.  You then have the option of working in RGB (all three color channels) or working in just the red, green, or blue channels.

Because it's blue we want to counter, we'd go first to the red and green channels, because those colors need to be added in.  Don't go to the blue first because all you would be able to do there (before working with the red and green channels) would be to darken all the blues to blacks or lighten all the blues to whites.  Since there are no other colors to show through, that will mean turning the whole picture black or white.

I'm going to use the white point slider because the black point slider will only turn what reds and greens are already in the picture into black, and this cap is so dark that they're already pretty close.  There's no room for them to become more vivid, visible colors.  However, if you were working with a picture that was particularly washed out, you'd want to use the black point slider and bring the colors down in order to darken them.  It helps if you think of it all in terms of adding or subtracting brightness from the colors.  Since all the colors together add up to white (when they're bright) or black (when they're dark), working in RGB mode will change the brightness and contrast of the image, and working in an individual color channel will change the brightness of that color.

So, I went first to the red channel and moved the white point slider up to 151.  That number really doesn't mean anything.  I simply looked at the preview, and put it where I thought there would be enough red in the picture.  And, since I can come back and change it whenever I like, there's no need to be absolutely precise.

I then went to the green channel and moved the white point slider up to 190.  Again, same thing with that number.  I then went back to the RGB channel and slid the white point slider up to 124, because there's where I was happy with the brightness, but I still thought the colors looked a little washed out, so I went to the blue channel and moved the gray point slider down to 0.85.  The colors you've already added in on the red and green channels now show through more because you're lightening the blue channel (and only the blue channel).  Then I went back to the RGB channel one more time to tweak the contrast (moving the gray point slider down a bit to darken the shadows and the white point slider up a little to brighten the light.)  Done!

Using levels, we can take that first, very blue screencap and get this . . .

There's still some grainy-ness, and it's not perfect, but it's a lot better than what we started with, and now it's a wonderful base for further fiddling.  And, if it needs cropped again, (which changes the histogram and therefore means you might need to adjust further) you can just go into the adjustment layer you've already created and make any further adjustments that are required.  You only need one layer to make the change (unlike the other methods, which all required a layer in addition to the method itself) and it means that, if you need to tweak it as you add textures, gradients, brushes, etc. it's all in one place.


The High Pass filter is particularly useful to anyone working with less than perfect images.  It helps address one of the main problems you so often run into with screencaps, blurriness, and it does it in a way that doesn't damage the original image while allowing you to make a wide range of changes.  You can mask a high pass filter, soften bits that are too sharp (like skin texture, hair, and random bits of background) without taking the sharpness out of the rest of the image.  It's also, I think, more easily adjustable than the unsharpen mask option, since you can see exactly which lines you're going to be sharpening.

So, for example, we could start with a screencap of Penelope Garcia from Criminal Minds . . .

It's a lovely image, but a little blurry.  So, we duplicate the base image, go to Filter --> Other --> High Pass and we get this . . .

This one is set to five, and the lines it shows in the grass preview are the ones that will be given more contrast to sharpen them.  I wouldn't normally use such a high setting, but I want you to actually be able to see what's happening here and the more subtle sharpening would be harder to spot.  (Usually I keep it low, say . . . 2.  You can always increase the effect by duplicating the layer, so a little goes a long way.)  After you've hit okay and applied the change, you then need to change the layer from normal to soft light.

You get something like this . . .

Much sharper, but there are elements that I don't really like: the background is sharpened, too, and I liked her hair on the blurry image better.  Easily fixed.  All you have to do is create a masking layer.  You do this either by going to Layer --> Layer Mask --> Reveal All or by using the shortcut icon at the bottom of your layer menu.  It looks like a gray box with a white circle in the middle.

Once you've got the mask up, all you need to do is use the brush tool (preferably a smallish, soft brush) set to black and (after making sure you have the masking layer selected) go in and brush over anything you want blurred.  (Our lovely katekat1010 is going to tell us more about masking layers later, so I'll just say how freakin' awesome they are and move on.  *G*)

Here, I used a brush at 100% opacity to re-blur the background (to reinforce Penelope as the focus of the picture) and a brush at 75% opacity to soften her hair.  Getting a final result of this . . .

Or, if there's more than one character in a picture, but you want to focus on someone specific, you can use the high pass filter and then mask it so that only that character is sharpened.  In the following example, the first one is the unsharpened image and the second is the sharpened image.

Current Mood: accomplished
littlespank: BSG-FamilyMomentlittlespank on February 1st, 2009 03:33 pm (UTC)
I really liked the Levels stuff. I always used levels, but I didn't have much of a clue to what they where ACTUALLY doing to the picture, so I would mess with them, not really knowing what was going on and unless it came out looking NOT weird, I would keep it. heh.

but now I get it. I make a lot of Battlestar Galactica icons, and like Heros the caps are dark and muddy and sometimes very blue. So I appreciate that very much.

thanks a bunch.
Lostgirllostgirlslair on February 3rd, 2009 02:18 am (UTC)
I'm glad it was helpful!
my monkied brain: _nihon_prayerskatekat1010 on February 1st, 2009 06:22 pm (UTC)
beautiful tutorial lady!! love the "good / bad / middling" of the crop comparisons!! And I've never used high-pass to sharpen - what a fantastic idea.
Lostgirllostgirlslair on February 3rd, 2009 02:22 am (UTC)
Thank you, sweetie! I like the high pass, sometimes a little too much. *G*
xjubesxxjubesx on March 9th, 2009 09:00 pm (UTC)
Great tutorial on sharpening! This helps a lot! Thank you. Adding to mems.
Lostgirllostgirlslair on March 21st, 2009 05:17 pm (UTC)
I'm so glad it was helpful!