I recently watched a video on selection techniques and one of the examples was for selecting a clear glass with liquid in it. They're technique to preserve the transparency went like this:
1. go to the channels palette
2. Cmd (Ctrl on a PC) click on the RGB layer in the channels palette to select the light parts
3. Invert the selection (Select/Inverse)
4. Go back to the layers palette and Cmd+J to put the selection on its own layer
5. Hide the background layer
6. Duplicate the selection layer 1-6 times to build up density of selection
7. Select all the selection layers and merge them into one layer
Would this method be okay as an alternative to the one you suggested on a previous post? Do you see any flaws in it?