As I already said, I don't want to use a full-blown library for such a small and simple skinned area with just a few buttons. And there's also the matter of personal interest; I would really like to know how it's done. Using a library just because you don't know how to do something, is a wrong reflex.

OK, I've done some more searching, but unfortunately, I couldn't find any examples of how to use such a multi-colored mask. However, I think that I should take the following approach: for each color in the multi-colored mask, create a 2-color mask with the current color replaced by black, and all other colors replaced by white (by using something like ReplaceColor at How to replace a color in a HBITMAP[^]). Then, those 2-color masks can be used to BitBlt the individual segments (that might come from different state images) into one single image. Is this the right approach? I want to be sure that I haven't missed anything before starting to implementing this.

Of course it must be possible, I'm pretty sure that no library was used in the WMP desk band either. SkinCrafter looks like it's hugely overkill for my (relatively simple) problem.

Thanks for the link, but that tutorial only shows how to create non-rectangular windows with a bitmap background. I don't need a non-rectangular window, but a normal window with non-rectangular buttons on it, by using a colored mask (the top image in http://img130.imageshack.us/img130/5195/wmpbandimages.png[^]).

Hi, I would like to create something similar to the Windows Media Player miniplayer in the task bar. I don't expect many problems with coding the desk band (plenty of documentation about that), but I'm still wondering how to start with coding the skinned UI. When looking into Windows Media Player's resources, I have found the following images: http://img130.imageshack.us/img130/5195/wmpbandimages.png[^] The first image is obviously a mask, and then there are images for normal, disabled, hottracked and pressed states. I guess that I should take the different parts of the 'state' images (according to the state of each part), and merge these into a single image that shows the current UI. But how can I do this? I would prefer not to use any libraries, just C++ and the plain Windows API. Thanks.