Another book I recommend is Dan Margulis’s Professional Photoshop Classic Guide to Color Correction, which also has a detailed chapter on sharpening. He actually prefers USM to Smart Sharpen, and his general philosophy is to sharpen as much as you can get away with, remembering that what you print–especially on inkjet–looks a lot softer in print than it does on screen.
This works on so many things. Back in the eighties, when I was starting out as a drum scanner op, the best three words my boss could say were “See what happens.”
I learned more just by ‘seeing what happened’ than I did in three weeks of intensive training.
If you’re trying to make selections where you’re eliminating a solid background, always be sure to check your channels.
If you’ve put an object in front of a green screen, or some such solid color, and snapped it’s photo that way, it should be fairly easy to get rid of that background by looking through the channels for the one that has the right/most contrast. If you load that channel as a selection, it may make a complex selection quickly and easily. (This assumes that the foreground object does not itself contain any of the same color as the background.) You load the channel as a selection by Command-click on the channel’s icon in the Channels palette. You may need to clean up the selection, depending on the image; but it can get most of your desired selection done quickly and easily. Then you just delete the background.
You want to do that to a copy of the layer, of course; always duplicate a layer before you edit it, including the “Background” layer; you always want to be able to go back to the original, if you really fuck up.
If you insist on using a tool to select an object to crop out the background, I suggest surrounding the item with a path using the pen tool. The path stays there, infinitely modifiable, until you’re ready to make the selection. You can be very precise with the pen tool, if you’re willing to work at it. Just keep zooming in as tight as needs be to get around tricky corners and such. Once you have it perfect, just Command-click on the path icon in the Paths palette, invert the selection (assumes you surrounded what you want to keep, not what you want to delete), and then just fire away with the Delete key.
I’ve got little to say about sharpening. For most work I’ve done, just a touch of Unsharp Mask is enough. I never get crazy with it.
I don’t use Curves much either. If I want to boost contrast, or correct the color, I usually go with Levels. I read once that Curves should be used when you want to make a tonal shift on the image as a whole–that is, you change tones one way or the other, but the pixels generally stay the same relative to each other. Levels takes your tones and moves them around relative to each other, which is generally more useful, especially for contrast.
As was noted above, there’s no right or wrong answer for this stuff. You just have to play around with the tools and see what gets you the best results.
By the way, what I’ve seen of the newest Photoshop–CS5 Photoshop–seems to show some pretty amazing new features for knocking out backgrounds, and making fine selections, and deleting visual objects from images and such. It might be worth the upgrade, if you can afford it. (I can’t right now, sadly.)
That’s long-winded enough for now. I have to go to bed. Keep editing.
I’m not exactly sure what this means. Levels is basically a crude version of curves. It’s a curve with a black point, a white point, and a midpoint (gamma.) Curves gives you a lot more control. If I want to increase contrast of an image, say, I will usually have four points: the black point and white point (which is already there), and then two more points, one for my shadow or dark midtones, and one for my light midtones/highlights. I move the shadow point down, and the highlight point up (forming what’s known as an “S-curve”), and voila, I have increased contrast without having to clip my black or white point. Plus I can keep adding points (if I wanted to) to continue making more localized contrast adjustments (although I usually work by adding only two, sometimes three, points.)
There is a little bit of a learning curve (hehe) to use Curves, but stick with it–it’s far, far more capable than Levels. And, when you get good at it, you’ll find yourself adjusting curves not in the master RGB (or CMYK or LAB) channel, but individually in each color channel. It is an incredible tool for color and contrast correction.
I started experimenting with curves (again) last night. The basic S curve with 3 or 4 points.
I think one of my mistakes in the past was trying to use too man points. I was trying to fix too much.
For awhile, I’ll keep it simple until I better understand the process.
I bought Dan Margulis’s color correction book about 8 years ago. I didn’t get far because he focused so much on CMYK. I barely understood RGB. Changing color spaces got me very confused. I’ve heard the newest version of Dan Margulis’s book has more RGB examples. That should help a lot.
I still use 7.0. Does CS or CS2 offer any tools that makes updating worth it?
I started with ver 5. Then upgraded to 7. At that point I was satisfied with what I had. Sometimes perfection can’t be improved on. 
duplicate
I don’t want to lose Image Ready either. It’s an excellent tool. It’s a shame Photoshop took it away.
I may buy a used copy of Photoshop CS2 that still has image ready. They snatched it away in 2007.
Yeah, for most stuff, you don’t really to add more than one or two more points. If I just want to lift the midtones (increase brightness without disturbing the white and black points), I add one point. This is similar to using Levels, except that you can put your midpoint anywhere you want. I usually put mine somewhere between the true midpoint and the white point. If I want to add additional contrast, I put it another point nearer the black point, and finish off an S-curve. It’s rare that I add a third or fourth point, but it happens sometimes if I want to eke out some details in the shadow areas. Still, I only end up using one or two points 95% of the time.
It’s a great book, but it is quite advanced. I have the fifth edition of Professional Photoshop and, flipping through it, there’s only one chapter really on color and contrast, and the examples are equally split between RGB and CMYK, with perhaps a little more RGB examples. “By the numbers” skin correction values are given in CMYK values, but you can get those values in an RGB space by using the color picker and selecting CMYK in the drop-down menu under the picker icon in the Info pallete. I do almost all my post-processing in RGB (I occasionally venture into LAB, but that’s it.)
Honestly, I can’t remember what upgrades are in CS or CS2. I have CS5, and there are a slew of features I have in CS5 that didn’t exist in the pre-CS editions of Photoshop (Smart Sharpen, Highlight/Shadow, Bridge file browser, Context-Aware fill, Context-Aware scale, Healing brush, Photomerge, etc. And those are just the ones I use with regularity. For me, the healing brush is one of my most used tools–it’s like a smart clone. I’m sure there are a million other features that other people use–those are just the most important ones to me. I think all the above, except for the context-aware stuff, is available in CS2.)
I don’t even remember what Image Ready was.
Was all set to say, “I don’t even remember Image Ready.” And then, i got to the very last line. Was it a Bridge precursor?
I’m in CS5 also, and love lots of things about it. Bridge and Camera Raw are probably worth an upgrade. I shoot RAW, and can make lots of minor adjustments in Camera Raw before i ever take the image into PS. Exposure can be adjusted up to +/- 2 stops without really losing quality. That’s built in to CS5, now, though i haven’t used it.
It will be pricey to upgrade to the current version, and before you do that, you might want to check out PS Elements. I haven’t used it since about version 5, but i understand it now has mask capability, and probably has everything you need to learn on, at under $100.
To get an understanding of what Curves is doing, open an image, and set up a Curves adjustment layer. Add lots of points–the idea is to stabilize that line so that you are only affecting one tonal area. Now, pick a point, and move it up and down. Go to one of your color channels, and do the same. Parts of the image are going to look solarized, but it should give you an idea of what is happening.
Curves is great for general contrast, as has been said, but one of my favorite uses is when i need just a little change of value in a certain area. For instance, that left ear is a bit too dark and distracting. Open a Curves adjustment, Command (Ctrl) click on the problem area to set a point there, and make the adjustment in Curves. Fill the mask with black and paint in the adjustment in white with a very soft brush at lowered opacity for a subtle effect that is only visible where YOU want it to show up.
For images that i’ve invested a lot of time in, i can easily end up with scads of adjustment layers–Curves for left ear, Hue/Sat for under eye bags, Selective Color for hair, etc. It doesn’t happen often, but it’s great to have that flexibility and power.
Don’t want to get into lecture length here, but just a couple more notes:
I use USM for tonal adjustment only. 25/50/0 on a dup of the background adds a nice pop, and intensity can be adjusted by changing the opacity of the dup. I like High Pass for sharpening, again on a dup with a black layer mask and desired sharpened details painted in. I might use USM for first-pass sharpening on a landscape or still life, but generally i’m pretty fussy about how sharp i want what, and i almost never want the image sharpened overall.
MASKS ROCK! Great not only for specific areas to be adjusted, but for vignettes, painting effects, layer compositing, and nearly everything else. I rarely make a selection to cut out a background–i mask it out, and add a new background underneath. Painting with 10-30% white at the edges of the subject can really help to make it look naturally a part of the inserted background. You can get into the mask with a 1-pixel brush to adjust details, and you can always select it and turn it into a path if you need that.
Top PS Tip–always work on a copy of the original.
Fun “Did you noobs know PS could do this?” project. Open an image. Double-click the background layer and click OK to make it a regular layer. Add a layer underneath (Command/Ctrl click the new layer icon in the layers palette. Or just make a new layer and drag it underneath the image layer). Hold Option/Alt and click the line between the layers. The cursor will turn into a double circle when you’re in the right area. Now, take your paintbrush at 100% opacity, and paint some squiggles on the lower layer. Wheeeee!
This is a clipping mask. The layer above (in this case, your image) is clipped to the layer below (the squiggles.) The image layer will only show where there are pixels in the underlying layer. Try it with text in the lower layer, or with a cutout of your dog’s face, or with a shape.
Well, I was pretty damned close with this guess. According to Image Sharpening with Adobe Photoshop CS2 that I linked to above:
Also, re: sharpening for print:
This book advocates a two-stage sharpening workflow: at the source (from scanner/camera/etc) and a second sweep at the very end, after resampling to final output resolution. It also calls Unsharp Mask “by far the most important item in the sharpening toolbox.”
You know, in all the years I’ve been working as a graphic artist, I’ve never seen Curves explained quite as well as these two examples. I think I’m going to play around with it using both examples above when I get home today. Thanks for that; I think you’ve raised my skill level a bit.
(Fortunately, most of my work has been more about page layouts, with quick and dirty image editing for not especially high-end printing. I’ve been able to get away with just using Levels for the most part.)
For those who don’t remember, ImageReady was more like Adobe’s version of Fireworks. More oriented for web images and animation. Some of those features were merged into Photoshop a few versions back. Adobe also still has its own version of Fireworks in some of the CS editions. Personally, I never really liked Fireworks.
Thanks again for teaching an old cat new tricks.
I love masks. And clipping masks.
It was all about optimised compression, animated gifs, and image slices. All things that are not really a big deal anymore.
I’ve never understood the obsession over image slicing. Seemed to have a very limited and specific use that was superseded by judicious CSS.
Didn’t someone upthread mention shooting green screen? If so, try changing your layer blending options, rather than cutting out your green screen background. Double click the layer, and at the bottom of the pop-up, in the Blend If section, choose Green from the drop-down menu, and move the This Layer slider 'til your background drops out. You can option click to separate the point and fine tune the color range.
You can do that, but remember that just changing the blending makes those pixels transparent, but doesn’t eliminate them. If you save that file in a flattened format, the pixels will still be there and opaque; you won’t get the transparency. I’ve done that sort of thing myself, where I only saw the visual transparency, worked on other parts of the image, and forgot the pixels were still there and opaque. You still have make those pixels permanently transparent if you want to really get rid of that background color. (Just insert an empty layer above or below the layer with the blending change and then merge those two layers; that will set the transparency.)
Wow, i didn’t know that it did that. I usually create a new comp layer with all the changes, then fine tune that before saving in a flat format. Thanks!
I think any way in which you merge the layer (with the blending change for transparency) with any other layer, it will set that transparency permanently. So your comp layer probably was fine. I just know that there have been occasions where I changed the blending like that, then placed the PS file (in native format) into a layout, and the transparency survive the trip. Just make sure to save a test image in your preferred flattened format, if you’re placing that image somewhere else. It’s always good to keep an eye on hidden pixels, so to speak.
Another quick and dirty way to achieve transparency for certain colors–if you don’t care too much about color fidelity (I know, I know…)–is to use the Save for Web feature. Choose Save for Web from File, then select the GIF option, and use the color palette selectors to selectively apply transparency to certain tints. If you have a relatively monochrome background, you can quickly drop out those colors and get transparent pixels in their place. Save the GIF and then open the file back up in Photoshop and presto! you have a clipped object with certain colors knocked out (albeit with limited color range, and you may get complaints from the program if you’re working on a big, high rez file.) Good for quick and dirty knockouts of relatively flat color fields.