Hi Wangyang -- As for definition of 2.5-D see my first posting above (way back there!). I notice the article I posted is no longer there, don't know if I did something wrong or what. If you send me an email, I will reply with the article attached if you like.
Also incidentally, one thing that is the same about 3-D arbitrary and 3-D planar (what some people like to call 2.5-D even though the term has problems in this use, as described in my paper) is that they all fail to analyze real world objects, and they all give the wrong answer.
(Wow! Did Rautio really say that?) Yes I did. They all give the wrong answer. The only question we can hope to answer is how much wrong the answer is. All computer modeling is an abstraction from reality. It is not and never will be reality. It can only give us a hint as to what will really happen when we really build it.
To answer the question of how wrong is it, we can use handwaving discussion to indicate possibilities in a most general and uncertain fashion. If one stops there for a real project, one is not doing engineering and does not deserve the title of engineer. If you want to do engineering, when things like "How wrong is it?" are important, you have to do numerical experiments and quantify it. Engineers (real engineers) want numbers. But quantifing error is another topic and I have quite a few papers on that if people are interested, feel free to start a new topic. -- Jim Rautio