Yesterday I made some progress again in reading and understanding the animation code. I can now succesfully read the scenegraph information as well. This information determines how the different transformations and animations relate to each other (child and peer relations). This makes sure they are added together in the correct way, to get the desired result.
If I now read in most of the default objects they look quite good. For example the default Elephant looks like one now, while before it more looked like a lump of polygons when I read in that model. The skin and bone animations are not supported yet in ModelConverterX though, so the animation still looks a bit funny. Also it seems there are still some minor bugs in the rotations I have to check. But overall I am quite happy with the progress. I will try to post some more pictures this evening.