I have just finished reading Diana Gruber''s artical on whether or not we need quaternions. I was wondering what other people thought about it? Is she right, in your opinions, or is she maybe missing some benefit to quarternions that justify their use? What are you people using in your programs? Should a newbie bother with quarternions or stick with vectors? Any (constructive) responses would be appreciated... <(o)>

I was under the impression that vector methods lost accuracy with time (at least according to some article I read somewhere ages ago ). Does anyone know if this is correct?

http://www.geocities.com/ben32768

Edited by - benjamin bunny on September 12, 2000 1:33:16 PM

In my opinion, Mrs. Gruber is completely off-base. I'm not going to repost the whole thing here, but you can read my reply to her article over on the Graphics Prorgramming and Theory board (or just check out the pointer to that article here.


Edited by - Shaterri on September 12, 2000 5:10:58 PM

bb: I think that''s because of floats losing accuracy -- not because of any inherint problems with vectors.

----------------------------------------
"Before criticizing someone, walk a mile in their shoes.
Then, when you do criticize them, you will be a mile away and have their shoes." -- Deep Thoughts

yes it is the fault of (float) that vectors lose accuracy. I had a problem with this when I kept concatonating rotations on top of each other on all 3 axes. The wonderful thing about quaterions is that you can fix this lose of accuracy by normalizing every so often.