If the scene surrounding the bed and comforter is important to the overall image, then you can use the traditional CG compositing technique of shooting an HDR spherical panoramic image within the scene and then drop in the CG bed/comforter and illuminate the CG element with the spherical HDR data from the scene. This is often referred to as Image Based Lighting (IBL) and, at this point, all CG applications support it.
As stated above, your exercise will be much easier by UV mapping your graphic designs onto a CG comforter. Because your CG object (especially the comforter) does not contain any reflective surfaces, you would likely not even need to use IBL techniques to get convincing lighting, because you do not need HDR reflections of the scene on your CG object.
There are probably stock 3D models of beds and comforters that you could use, eliminating the need to have a model built from scratch, and a clever CG artist could actually use your background plate and model just the comforter, with some skillful lighting and compositing to make a convincing composite. Many CG applications have utilities for deriving the virtual camera position and focal length from the actual camera data and background image, as well as soft body (cloth) dynamics that will permit the modeler to create a CG comforter that drapes and folds convincingly over some stand-in geometry (a box representing the bed).
Lots of ways, but pure PS manipulation would take a lot more time with less than optimal results - and that would have to be repeated for each comforter graphic. You would likely get variation in the overall appearance of the composite between designs, where you really want uniformity in lighting, etc., so that the attention is on the design itself.
Good luck!
kirk