This would be something you'd probably have to do manually in a program like Photoshop or GIMP. The trick is the joints. So, you can't just slice the figure up, because obviously then people will see the slices. Instead, consider the knee. What you probably need to do is create one image that's the shin and another that's the thigh, but in the knee area what you probably need is to have some of the knee in the shin and some of the knee also in the thigh, possibly with some transparency in one or the other where the two overlap (whichever one is "on top"), so that as the knee bends the two images (overlapping one another) give the impression of the knee bending, as opposed to a "ball-joint doll."
The other challenge would be areas of overlap that hide parts underneath. Let's say that the arm overlaps the trunk. (Think of a side view of a person.) In this case, you'd obviously have to cut out the arm into a separate image (again, possibly with some transparency at the shoulder) so that it can rotate, but then have to deal with the area that the arm overlaps. That might mean a render (assuming you're using 3D software to generate the figure) without the arm, or with the arm repositioned so that you have all of the trunk, including the part that the arm overlaps when it's in its resting position.
This obviously sounds tedious, but it's definitely doable. "Back when" there were a bunch of games released at
You must be registered to see the links
that used this kind of animation in Flash-based games. If you can still get them to run, you might want to study one, with particular attention to the graphics. If you squint carefully, you can see how he cut up the figures to do exactly what you're talking about. His games also tended to use "progressive animation," where, as you moved the mouse around, certain motions would gradually increase in magnitude over time (um, think "insertion" LOL), or would otherwise gradually change, and so what would happen is that you'd reach a screen where you'd have to execute the motions a number of times until a certain goal was achieved, at which point you'd then be able to progress to the next screen. It was an interesting game mechanic. Not necessarily for everyone, but creative.
The other entirely different approach to this whole thing, of course, is to just create an image series, and to index back and forth through the image series based on the user interaction. A lot depends on exactly what kind of animation you're trying to achieve. If it's a simple linear motion (grin) then the image series approach would probably work. (Although it could be memory-intensive if you have too many images). If you're really going to be animating in multiple directions (up/down/left/right) simultaneously, then the first approach is more likely to give you something "good looking."
