🔒 Locked

Bounce vector, please help!

Started by dmitbcpitew Dec 4, 2001 at 8:17 AM 7 replies 1k views

Original Post

Dec 4, 2001 at 8:17 AM

Hi all, I've got two normalized vectors; one corresponding to a direction of movement (A), the other to the face normal (N) of a polygon (P) with which the moving object has collided. I just want to derive from these two vectors a third, corresponding to the vector with which the object should bounce off of the collided polygon (X); \A.....|N....X/ .\.....|...../. ..\....|..../.. ...\...|.../... ....\..|../.... .....\.|./..... ......\|/.....P --------------- At the moment I'm just using N, which is OK, but doesn't result in really believeable collisions. Please help! I guess this should be easy if only I knew my maths a bit better! thanks, dm. Edited by - dmitbcpitew on December 4, 2001 9:19:46 AM Edited by - dmitbcpitew on December 4, 2001 9:22:09 AM

henryx

Dec 4, 2001 at 10:49 AM

X = 2.N - A

Fruny

Dec 4, 2001 at 1:16 PM

X = 2(N.A)N - A, assuming N is normalized.
If A points towards the surface, so will X.
If A points away from the surface, so will X.

dmitbcpitew

Dec 4, 2001 at 1:34 PM

Thanks guys, I''ll try this out!

Vlion

Dec 5, 2001 at 12:31 AM

angle of incoming to normal is the angle of the outgoing to the normal.

I came, I saw, I got programmers block.
~V''''lion

Fruny

Dec 5, 2001 at 1:00 PM

Yeah, but with the dot product equation, you don''t even need to know the angle (nor use a trig function!)

dmitbcpitew

Dec 5, 2001 at 4:51 PM

Thanks for the feedback; I''ve tried the following

Fruny, I tried:

D3DXVECTOR3 vNew ;
float fDot ;
fDot = D3DXVec3Dot( &vnAvg, &m_vDirectionOfTravel );
vNew = vnAvg * fDot * 2.0f ;
vNew -= m_vDirectionOfTravel ;
//vNew *= -1.0f ;

Doesn''t produce the expected rebound. Have I got the precendence right? I''m assuming

2(N.A)N - A == (2.0f * (N.A) * N) - A

I''ve tried with and without vNew*=-1 (trying to reverse the direction of the vector)... doesn''t seem to work.

henryx: 2.N? Doesn''t dot product take two vectors, not one vector and a number?! Confused!

Thanks for the help guys,

dm.

henryx

Dec 7, 2001 at 10:24 AM

.......|\......
.......|.\.....
.......|..\....
.......|...\...
.......|....\..
.......|.....\.
.......|......\
\A.....|2N...X/
.\.....|...../.
..\....|..../..
...\...|.../...
....\..|../....
.....\.|./.....
......\|/.....P
---------------

Sorry I took so long to get back but the diagram above is what I meant by X=2.N - A. i.e 2 times the vestor N - which as Fruny points out N = (A.n)n where n is the *normalised* surface normal.

This only works correctly if the A vector is pointing away from the surface then the result X is also pointing away from the surface.

If A is pointing towards the surface then X=2.N + A and X is pointing away from the surface - which is probably the most real situation to model.

henryx

dmitbcpitew

Dec 7, 2001 at 11:25 AM

Thanks v. much! I''ll try this out at once!

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.