Basic PhotoShop Boot Camp
I didn't always use PhotoShop. In fact, I started out with a program called PaintShop Pro. For most beginners, I would recommend against starting with full-on PhotoShop. The program is powerful, but not the easiest thing in the world, and one can quickly get lost in a blizzard of odd jargon, unfamiliar shortcuts, and weird user interface. Someday I'll write a brief treatise on all the UI changes I would make if I was driving the SS PhotoShop. For now, I get by with my "Workflow Cheat Sheet". PhotoShop is so complex that most people who use it have actually had to create a standardized process or workflow that they use to process photos. This ensures consistency, convenience, and through the magic of PhotoShop Actions (like what many products call macros or scripts), you can automate your workflow to a greater or lesser extent. So far, the cheat sheet has done well enough.
If you are new to image processing, read my PaintShop Pro article first to get some orientation and see what's happening to photos step by step. If you're ready for the PhotoShop fire hose, here is my workflow cheat sheet. I break down each step below the chart and try to help you understand the whys and wherefores of it.
My Workflow Cheat Sheet
I'm still using a USB-based gizmo that takes a Compact Flash card. Mine is called a ZiO! It was cheap, and makes the card look like a Windows drive. Very convenient. I dump the images into my photo archives unprocessed.
A word about file formats. I use the highest quality JPEG setting on my camera, and I'm pretty happy with it. However, if you want the absolute best possible results and don't care what they cost, use RAW. There are a lot of reasons for this, but they boil down to two simple issues. First, conversion to JPEG will lose some information from the original image. This comes about because of the JPEG format, and because of the camera presets. The colors, sharpening, white balance, and overall dynamic range are pretty heavily molested by the algorithms inside the camera, and this cannot be undone once it happens. The second issue is that however good the algorithms are inside the camera to do this stuff, they are not as good as what can be done outside. This is because the CPU in the camera isn't as powerful as in your PC, there isn't as much memory, the software isn't as powerful as PhotoShop, and you have very little opportunity to tune the parameters to your eye, the way you would in PhotoShop.
Despite all that, I am still not a RAW lover at this time. My issue is it takes a lot more processing to make a RAW image look good. They come out pretty nasty, and this is why the camera does all that processing. Perhaps one day, if I fully automate my workflow using PhotoShop Actions, I can automate what the camera does well enough that it is very little more work for me. That is what the pros do, so you can hardly blame them for liking RAW.
You need to give some thought to your archiving strategy. You never know what you might want to do with the images in the future. The same image that looks great on a web site needs completely different processing if you print it, for example. My solution is to archive the raw (not RAW) images as they come from the camera. I never touch those files, instead I make copies and work on the copies. Be suspicious of even something as simple as rotating an image. Rotating a JPEG image with no loss of quality is not something every program that will rotate does. Leave the raw images alone. You'll also want to organize them in some way. In my case, I just use my web site's organization, but there are many other possibilities that may work better for you. Another possibility is to archive the PhotoShop file format. See my comments on Layers below, because if you're going that route, it will make sense to understand those trade offs.
So, before I work on my images, the Archive step is to copy them to wherever I plan to work on them. For me, this is usually my local Dreamweaver directory that'll get uploaded to this web site.
Secure in the knowledge I have the original, untouched image, if I screw up, I load the file into PhotoShop and proceed with the munging process. My first step is to rotate the image to proper perspective. I frankly can't work on an image that's rotated too badly. There are two kinds of rotation. First is major, as in, I took it with the camera held vertically so it needs a 90 degree rotation. Do this one immediately in any case.
The second rotation comes because we rarely hold the camera perfectly level. It's off by 1 to 3 degrees. Sometimes this is not so noticeable, but if there is a horizon, or vertical and horizontal lines associated with manmade objects, it can be very distracting. There is a cool trick to fix this easily. Switch to PhotoShop's ruler tool, which is under the eye dropper. Now find a line on the photo you want to make square. Horizons are great for this, but vertical lines work too. Drag over the line in the photo as it sits, not how you want it to be. Now bring up the Arbitrary Rotation dialog after drawing the ruler line and the correct angle is filled in for you!
I think cropping has done more to improve the quality of my photographs after I take them than anything else. I find it very hard to get the composition just right in a viewfinder. For one thing, I'm usually not able to devote a lot of time to it. For another, I can hardly see the darned thing because the image is small. Cropping in PhotoShop let's you leisurely deal with it in the comfort and privacy of your own home. You may even find one picture turning into two or three variations that use different cropping.
Warning: Hard Stuff Ahead
Everything until now has been pretty straightforward, and frankly, hard to screw up. Get ready to enter the painful part of the exercise. Everything from here on is fraught with subjectivity, complexity, and the opportunity to make your image worse instead of better. Nevertheless, you do still have to deal with it, so get a grip, and let's dive in.
You'll note that I list three columns in the workflow. The columns denote increasing complexity to the right, and demonstrate that there are a lot of ways to do things in PhotoShop. Believe me when I tell you I am barely scratching the surface and all 3 of my columns go in the "Beginner" column for most experts. What's also true is that every image does not need to most advanced possible processing. Frankly, I just don't have time for it, and a lot of images come out greatly improved with just the simple stuff. Some images are more challenging, and will need the advanced techniques to really work out. As we go through these harder steps, I'll break out each column and try to explain more about when and how to use each option.
Keep in mind that it's hard to impossible to make a bad picture into a great picture with PhotoShop. You can make a bad picture into a tolerable picture, or a good picture into a great picture pretty easily, so take heart.
Before we consider those options, let's understand what we're trying to do with levels. Unlike the human eye (which has even better algorithms, more cpu, more memory, and automatic "tweaking" compared to PhotoShop), cameras are pretty dumb about how they record light. They measure the absolute brightness of each pixel and call it good enough. They have a limited number of "buckets" (bits) they can fit this information into. An 8-bit color can only have 256 different values, for example.
The trouble is, digital differences in brightness don't map well to human perception. There can be shadowy areas that just look featureless, but that contain enough hidden information to bring out details. Likewise, bright areas that look featureless can hide detail too. Also, the details may not be hidden,but the image may call attention to the wrong details (i.e. the bright parts may not be your intended focus), or may not offer as much contrast and detail as you'd like. Our eyes manage to extract a lot of detail everywhere, no matter how bright or how dark, even if they have to make it up because they know what the thing SHOULD look like! We can't do quite as well in PhotoShop, but we can beat the camera's defaults every time. Enter the world of PhotoShop Adjustments and Filters.
I always try PhotoShop's automatic setting. Try this Beginner thing first and see if it's good enough--it's a quick key sequence. Follow that sequence with Ctrl+Z (Undo) if you didn't like the automatic result. Move on immediately to Color Balance (Ctrl+Shift+B) if it looks pretty close.
For a lot of my pictures this is good enough and I'm off on the next adjustment. Someday, my eye will just recognize which images it isn't even worth trying this on, but for now, the cost is minimal to try, and the results are often very close. Also, you don't have to undo if the image is substantially better but still needs more work, just go on to a more advanced step with the default automatic adjustment as your new starting point.
Ctrl+L and Manual Adjustment
If the automated defaults didn't work, it's time to fiddle it manually. With Levels, we are trying to make sure the image uses all the available "buckets" for brightness without wasting any. The Levels dialog contains a histogram that is simply a measure of how many pixels in the image there are for each brightness level. If there are none at the bottom (dark, on the left in the histogram) end, we've wasted some of the buckets. Likewise, none at the top (bright, on the right in the histogram) means we're wasting those buckets. Moving the little arrows on the histogram will cause PhotoShop to stretch out things to use all available brightness levels. This will usually make the image look better, but wait, there's more.
If this is all you wanted, the automatic default would probably have worked. Instead, the photo may have places that are too under or over exposed for the default to be good enough. It's time to make some trade offs, and as the human side of this equation, you get to decide which ones. Start with the dark and light arrows. As you drag them into live histogram curve, you are destroying information by making all the data between the histogram edge and the arrow be treated as one brightness value. That may be good, because it opens up some room in the middle to let the middle ones breath. You gotta play with it to see if you like the result. Maybe it's fine that the overexposed areas stay overexposed or the dark areas stay dark. Maybe it is the middle you want to get looking better.
Towards that end, you can also fiddle the middle arrow and punch up those middle values. This is cruder than Curves (see below), but can bring out a lot of shadowy details. If you are at all comfortable with Curves, I'd encourage you to consider leaving the gray arrow alone.
Bottom line is you have to work those little arrows, flip the Preview on and off to see "Before" and "After", and see how much progress you can make.
Sometimes, bringing out the details requires not just the levels, but also requires some color fiddling. I find this is most true of my underwater photos and less true on dry land. For those cases, I have had some success with the Variations command. Play with it and try it out. Sometimes its exactly the right beast, but it is limited.
Ctrl+M, Eye Dropper Black, White, & Gray
Things are getting pretty difficult if you are still not happy. The next step up in your level setting weapons would be the eye droppers. If you can find some place in your photo that you are absolutely sure should be white, another that should be black, and hopefully a third that is gray, PhotoShop will figure out the optimal Levels settings. You just have to click on those places with that eye dropper. For some experts, this is so automatic and easy, they always go right here. Even if you don't have all three, setting even one can be helpful. I have a memo to self to try to do this more often, but for now, it is in my advanced category.
Some programs will do this with colors too, for example, if you are certain what a skin tone should be, or hair color, or some primary colored object. I'm sure Photoshop does that, but I've never explored it.
What if it's hard to find black, white, and gray? I don't have a clever answer for gray, just keep clicking near misses until the color pleases you. However, for black and white, think "lightest" and "darkest". Select the color-sampler-eyedropper tool from the palette, and then select the threshold tool, Image | Adjustments | Threshold. Drag the arrow left until a very small target remains--that will be your darkest spot. Shift click the eyedropper on it (best to use the 3x3 average mode). Now drag the arrow to the right to find the lightest spot and shift click there too. You've now marked the brightest and darkest spots. Cancel thresholds and go back to the Curves tool (Ctrl+M). Click the darkest and lightest eyedroppers on the appropriate spots. This works pretty well, though more often with black than white. If it's a little off, tweak the curves manually on the individual Red, Green, and Blue curves.
Consider Blended Layers
PhotoShop mavens just love Layers. They can't get enough of them. Recipes for PhotoShop spend half their time and effort getting the right Layers geared up to do the task. Maybe this is automatic for the mavens, and maybe it's worth it, but I confess, I don't like Layers. They seem like a lot of extra work for minimal payoff to me.
The deal with Layers is they let you change your mind later. A real Jedi PhotoShop guy can blast through keystrokes to create a Layer, set the blend mode to Multiply or Screen depending on whether they want to lighten or darken, and have all that happen nearly as fast as you or I could trigger the Automatic Defaults. I just don't fiddle with PhotoShop that much, and so far, I have never wanted to go back to an image I've saved and change my mind later. I don't even keep the PhotoShop format file that would have the Layers in it. I just use Undo or the History Window, and don't move on to the next step in the workflow until I'm happy with the result. That's all the more attention I'll ever have time to give my photo anyway!
Nevertheless, this subversive talk against Layers is part of a Vast Right Wing Conspiracy, and you should ignore it, and get right back to work using Layers!
Consider Selecting Different Areas and Treating Them Differently
Sometimes things are so badly screwed up that there is no single thing you can do to the whole image that makes all parts of the image happy. When that happens, you need to process different areas, well, differently. I like to do this by selecting the areas using some combination of the Magic Wand, Lasso, and other tools with Shift, Alt, and Control to add or subtract regions until I have approximately the area I want to process. Hopefully I only need to process two selections, one bright, one dark. For example, sunlit windows in a room shot get one treatment, and what's in the room gets another. Or, above the horizon is bright sky, below the horizon is dark stuff needing more detail. Feather the selection or this technique will look painfully unnatural. Use a pretty small feature selection if sharp edges are visible (1 to 3) and a bit more (say 5) if things are less distinct.
Process the selected region using the techniques described above, and then use Invert Selection to deal with the other half separately.
Consider Dodge and Burn
If there are more than 2 areas to be selected, and I can't use invert selection to effectively only have to select one region, I generally fall back to Dodge and Burn tools. These tools are really nifty, and have a much better UI than all the Layering and Funky Selection approaches to the problem. Their sole drawback so far as I can see, is that PhotoShop isn't very good at letting you change your mind about them much later. As I mentioned, I don't care about that much, so I am happy to dive in and try these tools. They do take a bit of touch, so Save before starting in on them, apply them sparingly, stop frequently to see if you like what you are getting, and Save whenever you do.
I believe the mavens get
the benefit of this style of UI using things like History brushes with
their layers, but being not a maven, I haven't explored that yet.
This is just evil voodoo, particularly if you want to be a total stickler about accuracy. I have watched an expert friend turn painful contortions to get color balance exact from screen to printer, and even he says he won't futz with it much beyond that. Color is too subjective, and most of us are too inexperienced at manipulating it to even know what we want to change. Darkness and lightness from levels is MUCH easier.
With that said, let's plunge right in, shall we?
To be honest, I try Ctrl-Shift-B and move on if I don't like what I get from it. I don't care about color accuracy, and I don't print, so I get off easy. All I care about is that it pleases my eye. In fact, as we'll see a couple of steps down, I frequently kick up saturation quite a bit because I like my colors to pop. Sometimes I will go a little further and experiment with the Variations tool. Frankly, if I'm doing that, it's to alter an overall color cast to bring out more contrast on details, and it usually only works for my underwater photos which have a rather obvious blue cast. I have also seen it work for sunset photos that also have a strong color cast.
However, if you want to do better than me, use those eye dropper tools in Levels. They'll get you spot on. Calibrate your monitor. Heck, my pro buddy shoots a special gray card in every scene (not necessarily every photo, the camera on a tripod takes the same scene unless you move it) and he knows exactly how that card's colors should come out the other end and will have a pretty easy time making PhotoShop deal with it. If this is your cup of tea, you go for it. I have also read about some folks using their camera's functions to calibrate color balance for a particular shot. Some swear you can never go without this. Sounds a worthwhile thing to look into, but I haven't tried it.
In theory, you can do everything in Curves you could in Levels, and then some. There are PS Mavens who don't bother in Levels because they like Curves and can make it sing and dance. In practice, I see them as two different user interfaces optimized for two different purposes. Levels will do the yoeman's task of grossly manhandling the brightness levels into approximately the right range so as to make maximum use of all the brightness buckets available. It gets things into the right ballpark. Curves then comes along with the fine tuning touch to make it even better.
Ctrl+M and "2" and "S" Curves
With Curves, you can come along and second guess that gross approximation Levels makes at various detail points on the curve. I find a subtle "2" curve, or an "S" curve can really help the contrast of an image. A "2" curve slightly de-emphasizes the brightest buckets and emphasizes the darker ones. An "S" curve gives you the opposite. Play with dragging those curves, but start with subtle almost hard to see (on the Curve) settings, and use Preview to flip back and forth before and after to see the impact. Look all over the image for the best overall effect.
Ctrl+M with Ctrl+Click on Points of Interest
The next step up after the manual curves is to precisely locate inflection points on the curve based on real points on the photo. If you Ctrl+Click on an area that is too bright or dark, that area will turn into an adjustment point on the curve. Play with that to achieve individual adjustments tuned for different areas of the photo. This can almost be easier than figuring out a good "S" curve in many cases.
Okay, we'll take a little break from the hard stuff. Things will be easy until we get to Sharpen.
For my workflow, Saturation is almost more in the nature of a decorative filagree than an adjustment towards real life accuracy. My camera gives good saturation right out of the box, or at least it looks pretty life like. I use this Ctrl+U Saturation command to "Kick it up a notch." I want my colors to pop. I want my skin tones to be warm and vibrant, my plants to be green and verdant, and my waters to be blue. Basically, I try setting Saturation to 20 on every image to see if I like the result.
Important Safety Tip: Save Resize almost until last. You'll be losing important information, so get all the other processing done first. With that said, this is another easy operation. I find my web site looks better and my life is easier if I stick to a couple of standard sizes for most images, which I've listed in the cheat sheet above.
Whew! If you've made it this far, you are almost done. There is one last opportunity to make your image worse rather than better, and it's one that a lot of folks overdo. It's called sharpening. Most images can do with a little bit of sharpening, but unless you're very used to looking at images and playing with sharpening, most people will choose to much sharpening on the quick glance test. This is also something to do last, because it too will destroy information that is useful to other steps.
Drop back to reasonable defaults, which I've listed above. See how well that came out. Crank it slightly, but try to avoid the super hard edge etched look unless that's really what you want to go for. The best sharpening tool is Unsharp Mask. You can buy plug-ins that are even better, but I haven't played with them.
You can get better sharpening results if you alternate sharpening and resizing over smaller steps. For example, to get to a sharp 640 x 480 image for the web site, I could do Sharpen the full size image, resize it partly down, Sharpen, resize more, Sharpen, and so on, taking as many as four steps to get the image all the way down to 640 x 480. It's laborious, but will produce a better result.
Now for the bad news. You can't really fix a horrendously blurred picture with sharpening any more than you can fix a bad exposure with Levels. I know, I have tried many times. However, you can help things along. If you have a really badly blurred picture, you need some heavier artillery than the basic Unsharp Mask. Cranking it up just gets results that are too ugly. Sometimes you even want the reverse. Images that are out of focus can look better if you can blur out a lot of the background clutter and work on your foreground subject.
If you deal with this a lot, I would experiment with sharpening plug-ins. I don't own any, so can't recommend. I do have this crazy action called the "Deadman's Custom Sharpening Action." I don't know where I got it, but I bet you can find some similar stuff on the Internet to download for free.
It works by selecting only the lines that most benefit from massive sharpening, and leaving all else out of it. It does this with a complex series of layers and masks and by asking you questions. Using this thing, you can get up to 400% Unsharp settings to work without too many artifacts. It's utitity is really limited to fairly simple images. I used it to process some jellyfish photos I took, for example.
You're past Basic PhotoShop Bootcamp, and while you are not yet a Maven, you are a functioning well-trained Private in the PhotoShop cadre. You know just enough to be dangerous to your friends and photos!
"Fixing" underwater photos is a little more challenging because they tend to have a lot of blue green color cast. I've tried several method with varying degrees of success.
Method #1: Use Hue/Saturation
In this method I use the Image | Adjust | Hue Saturation tool. Basically, I crank up the red channel's saturation (up to 60%) and then I crank down Green and Blue (20% or so). I finish with an Image | Adjust | Auto Color. The results are not bad, but not perfect either. Sometimes it all comes together though:
Method #2: Use Color Mixer
This is a more interesting approach than the brute force of cranking red up and turning cyan down. Here, we're actually going to convert some of the blues and greens into reds using the color mixer. This seems to produce a more natural result in some cases. Consider the before and after shots:
After: Hey that guy is actually pink, not blue!!!
To perform this magic select Image | Adjust | Channel Mixer. Pick the Red channel. Set green to mix about 60% to red and Blue to mix about 40%. Now go back and hand tweak it with Curves or Hue/Saturation.
Method #3: Curves Gray Setting
If you can pick a really good spot on the photo that's supposed to be gray and then use the gray eyedropper from the Image | Adjust | Curves tool, you can get great results too. These will tend to need more hand adjustment afterward than the two methods shown above. I'm seriously considering bringing a gray card along on the dives to help calibrate the colors every so often. Here's a shot that was processed with the gray eyedropper approach:
The trick is finding something that is "reference" gray. The highlights on that tank were about right...
I've found that you can almost always get a useful result with the black eyedropper too. See the technique above using the Threshold tool to find the darkest and lightest spots.
Method #4: Warming Filter
A little judicious use of the Warming Filter (Image | Adjustments | Photo Filters) can often be helpful, though it seldom does the whole job. I find using Method #3 together with #4 and adding back a little Saturation is what I do fastest and most often.
All material © 2001-2006, Robert W. Warfield.