View Full Version : blinn half vector

03-14-2003, 02:09 PM
i still dont get it, everywhere i go i see H defined as (V+L)/2 , with V and L normalized vecs
but, shouldnt H be normalized as well?
if so, what's the use of dividing it by 2?
thanks for your help

03-14-2003, 02:27 PM
H is an approximation of an apporximation.

H' ~ R = 2(N.L)N - L

H' = (L + V)/ | L + V |

H = (L +V)/2 ~ H'

1/2 ~ | L + V |

|L| = |V| = 1

I think the equations are right. Double check in your favorite copy of Foley et al.

[This message has been edited by PK (edited 03-14-2003).]

03-14-2003, 02:33 PM
You are perfectly right. H should be normalized. The equation is usually written that way, to state that H is the average of L and V, and after that you should normalize H. Of course, you can skip the division, since in that case it's mathematically equivalent.

03-14-2003, 09:54 PM
oh thanks it finally makes some sense
pk: i think the approximation only works in the viewable range of R since i just dont see how | L + V | could be ~ 2 even if |L|=|V|=1, and i wouldnt say that H (or H') ~ R because H is usually far away from R and it's being dotted with N, while R has to be dotted with V
anyway thanks again, this H thing was just driving me crazy