Both methods start with isolating the image area using selection. That sounds easy, but it is the most complex part of the process. You may need to use a variety of toolsfrom the selection tools themselves to the drawing tools used to edit the mask/selection.
If you are not sure how to approach selection, probably the most versatile way is to use a mask. To create an editable mask, you can either create a new alpha channel (on the Channels palette popup menu, choose New Channel), or use Quick Mask.
Using Quick Mask is almost the same thing as creating an alpha channel mask (in fact, it does create an alpha, but in a temporary way).
Once the channel is created, be sure that the view for it is on (click the view "eye" on the Channels menu). Essentially, using a channel for creating a selection allows you to make a selection onscreen as a grayscale representation. Use an eraser with a 96% hard brush, and start clearing away the area of the mask that would contain the portion of the image you want to retain. The eraser brush should have slightly soft edges to help smooth out some of the more difficult areas of the selection.
If you are very familiar with masking and selection, you may have other preferences for creating the selection.
To turn the masking channel on and off as you work to view the image below, click the eye for the masking channel in the Channels palette. You may need to switch brushes several times to get some of the tougher spots; and/or use other drawing, painting, and selection tools as you become familiar with them. Take care when clearing away all the red mask over the area of the image you want to keep. When you are done selecting, the hard part is really done.
Grayscale portions of the mask act as a partial selection. For example, 25% gray will select 75% of the tonality of the image. For the sake of the transparency that you are creating, pixels in your selection should be either pure white or pure black.
The white part of the mask is the area of selection by defaultalthough you can change it in the tool options. The black area is the portion outside the selection. Also, by default, the black area of the grayscale is viewed as red when the mask is viewed along with the content; if you look at the channel/mask by itself, it is black (see Figure 1).
This image helps clarify mask views. The first view is the original image with no mask; the next is the image with a the mask viewed at the same time; the third is the mask in grayscale; the fourth is the image area affected by the selection (showing the potential transparent area as a checkerboard) (image courtesy of Photosphere© 2001).
After completing the selection (if you use Quick Mask, you need to convert back to a selection and save the selection), save the image as a PSD (for example, myimage.PSD). The save retains what you have done to this point, and effectively saves the selection with the image. With the selection saved, you can use it to create the image parts you need for creating transparency on the Web and in print.
Using extensions can simplify your image management and stop you from saving over an image that you really want to save. In a later step, when you save the image as a GIF, the naming scheme actually can act like a filing system (that is, myimage.PSD is related to myimage.GIF, myimage.EPS, and myimage.TRG).