You could probably monitor the DOS VM to scrape camera/object matrices, render the static world in a huge GL (or what have you) display list, and then the other

You could probably monitor the DOS VM to scrape camera/object matrices, render the static world in a huge GL (or what have you) display list, and then the other objects. The mesh formats were reverse engineered long ago.