dudx

I have been looking over this per pixel lighting presentation again and i just cant figure out when they talk about solving dudx, dudy, etc how they are coming up with the solution. Under the slide that says “Generating the Partial Derivatives” they show some math that gets dudx from the plane equation Ax + Bu + Cv + D = 0. This is what they do:
Assume: Cv and D = 0
Ax + Bu = 0
Ax = -Bu
x = -Bu/A
dudx = -Bu/A

Ok where was a derivative done? I see how the Cv and D were set as constants to they would go away since we are doing a partial derivative. But they just solve for x and then say that dudx is the same as what we got for x. What gives? Could someone explain this? Its just annoying to me that i dont completly understand it, i dont like just using code im not sure about.

-SirKnight

I had the same problem before, so I sat down and thought about it and this is what I got :
http://members.rogers.com/deseric/tangentspace.htm

I hope it will help you.

The derivative are actually the same has the value you found because it is simply the slope of the plane of the line in (x,s) or (x,t) space.

[This message has been edited by Gorg (edited 01-07-2002).]

Well i just scaned the document and it looks like this will be a big help. I think this doc will do it. If i have any further questions ill post back. Thanks a lot.

-SirKnight