If you are interested in creating work such as this, there are a number of technical/logistical things to be aware of when shooting. Here is a list that I gave to my students:
Must shoot RAW.
Must set exposure once, use same exact setting for each of the two (or more) shots.
Must focus once, don't refocus between shots. This will require you to leave your camera on manual focus after initial focus. Think through where you need to focus at. Consider depth of field issues.
If you have a more compact camera that doesn't offer manual focus, see if you can set your focus by at predetermined distance.
White balance can be synced in Camera Raw, but must be the same for each image.
Must use a tripod. Take care not to bump the camera between exposures. Use a remote release if you have one.
Between shots, you need to change the appearance of your model as much as possible. Consider the following:
clothes
hair
jewelry including ear rings, piercings, watches, bracelets, etc.
glasses
facial hair?
make up - some then none, do it differently, etc.
shoes, you can change someones apparent height by changing their shoes...
tight fitting and then baggy clothes?
general style of clothing.. street than dress clothes?
Remember, you want to create a series of three in which the relationship of the people is ambiguous. All the rules of design still apply. Avoid flat looking images in whcih he subject is in the same plane each time.
Domestic scenes may work well?
Then you need to edit the RAW images exactly the same. Then you need to combine them.
Here is a link to an tutorial on how to Combine multiple images into a group portrait.
Although the intent of the tutorial is different, it is exactly the same thing. Here is a link to a video on Aligning Layers by Content.
Avoid the obvious things like a person playing cards with themselves.
You may want to take three images or more to bring back down to two. Here is Ohio we get a fair amount of snow. I was thinking of an image where the person is walking through the snow with one another. Of course you would create footprints as you leave the scene to go change. If you took a photo before you had your model walk into the scene, then you would have a photo to mask back the snow from. You could do it with just two shots, but it seems like the third one would come in handy.
