I'd appreciate feedback, especially if I'm unclear on something, as this will eventually be archived on a high-traffic site and I'd rather contribute something that makes sense. Also, I'm going to be writing a few coloring tutorials for the same site and they will reference this with the shorthand, "prepare your base." Thanks!

This is a tutorial for achieving a clear, crisp base to make into an icon. The degree of clarity needed depends on the subject and crop. It's most critical in a close-up and less so in scenery icons. I have a personal "tried & true" method for making bases of people and several tricks for others.

Part I. The extreme close-up
From =>

I first learned about base preparation from this excellent tutorial written by icon_ascension. I highly recommend it. Over time, I've adapted my own version that allows me to control every single pixel. Quite probably there's a faster and easier method out there that can get reasonably close results but this gives me the quality base that I want.

On 99% of my people icons, I follow the same work flow:
- Crop: I make several crops and then pick the one I think works best.
- Image Clarity: I perfect the image, getting it as sharp and clear as I can. This creates my base.
- Color/Texture/Whatever: Manipulate the image to change colors, lightings, add textures, whathaveyou.
- Image Contrast/Reinforce Details: Soft Light Layer of the Base to add snap and reinforce details of the image.

1. Crop to 100x100 pixels. I set my cropping tool to crop 100x100 with a resolution of 100 pixels/inch. Internet images are only 72 ppi so I've got a little resolution to burn.


2. Copy Layer (Ctrl+J). Sharpen the copied layer. Filter -> Sharpen -> Sharpen.


3. I have a hard time seeing what's happening in this cap so I temporarily brighten it using a curves layer. Adjustment Layer -> Curves. I go to the RGB curve, click the middle of the line and pull it upward until I can see well enough to work on the image.


4. Enlarge to 500% so that you can see each pixel and polish the skin using the Smudge tool (5 pixels, 11%). If an area is smaller than 5 px, don't mess with it. Avoid eyes, nose, and mouth. Just smudge the smooth expanses of skin. I mostly use circular motions and follow light patterns. Don't smudge across a high contrast shadow/light area or it will smear the shadow out of shape and look weird.


5. Add a layer mask to the copied & smudged layer to remove obvious oversharpening. Generally, it's the pixels that go white that make something look oversharpened. These are most noticeable in the whites of the eyes, eyelids, right next to eyebrows, on the nose around the nostrils, along the edge of a face, around lips, and in hair. These are the pixels that make something look oversharpened and must be removed. Have a look:
(Normal-sized after smudging)
(Click to see the above image at 500%)

Using a pencil (black, 1 px, 100%) on the mask, hide all of the over sharpened pixels. Because you have the original image below, this produces a seamless blending of the two images. Here's the sharpened & smoothed layer minus the erasures:
Highlighting what I removed with the pencil:

Janet's hair shows signs of oversharpening but I don't want to get rid of the sharpening entirely, so I use a brush set at 50% opacity to go over her hair. Lately, I've taken to fully erasing the background as well. This is a really subtle effect but I think it makes the subject of the icon just that much more "in focus" and stand out from the background.

6. Either hide the curves layer or delete it because you're done with it. Stamp visible onto a new layer.
<= This is now your BASE.

7. Copy the stamped layer and set to Screen 100%. Every image is different, so copy the screen layer until you're happy with how light the image is. For this image, I needed 3 screened layers.


8. Do whatever coloring or texturing or whatever you want to the image.


9. I always top off whatever I'm doing with a copy of the BASE on soft light. If it's too dark, I play with the opacity, lowering it to lighten the image. If I can't lighten it enough, I go back to the screened layers and add another copy to lighten the image. I use the soft light layer to restore a little contrast and depth to the image as well as to emphasize details that might have gotten blurred or otherwise obscured in the coloring process.


Here's my palette:
(click to see full size)
(final icon)

Part II. The scenic icon
Examples:

Far less critical to sharpen everything in order to end up with something that looks in focus but the greater variety in scenery means I can't say, "here's a technique I use that will help on 99% of your icons." It's far more useful to adapt your technique to the image. Typically, I start by:

- Crop the image. I think of this as the Base Image.
- Copy the Base Image, set to Screen 100%.
- Copy the Base Image, set it to Soft Light 100%.

This will look good on - what? - 50% of images? At most? I experiment by altering the opacity of the layers or duplicating the screened layer so that I can tell how much clarity my image has. Once I've got the image lit so I can tell what details are and aren't there, I sharpen the Screened layer. If I started with a reasonably good image - and really? that's the #1 secret to achieving clarity - this might be enough. Too much? I fade the effect. Not enough? I sharpen the Soft Light layer. Again, I fade the filter if it's too much.

So, the first image above, of the Stargate, I sharpened just the screened layer and was happy with the definition that gave me. I wanted more contrast so I used my standard black/white gradient map on multiply, colored the chevrons a brighter red, fiddled just a tiny bit with curves, and applied a blue/white, radial gradient over the top to finish off. Very simple coloring.

The second icon, with the team from Stargate, I was pickier about because it had people; I combined the two approaches. I copied my base, sharpened it, and faded the effect a little. I didn't think the team stood out enough so I masked out the sky on the sharpened layer (that's the same technique I showed above: masking out the background as a subtle method of focusing the eye on the subject). It also kept the sky soft. I liked that clarity so I tweaked it a little with curves for lighting/color and used that resulting image as my base.

(I'm not covering the coloring method here because A) that's outside the scope of this tutorial, B) it has a load of steps, and C)it was inspired by the gorgeous work of oxoniensis and she's the true master. If anyone should write the tutorial on the coloring method, it is she.)

Hope some of that was helpful. Feel free to ask any questions you want. I can't guarantee that I know the answer but I'll happily answer any that I do!