Rav

06-28-2002, 05:45 AM

On a paper on the microsoft developers website, I found a forumula for generating tangentspace that goes like the following (pseudocode-ish):

Vec Vector1 = ( vertex3 - vertex2 )

Vec Vector2 = ( vertex1 - vertex2 )

float delta1 = vertex3.v - vertex2.v

float delta2 = vertex1.v - vertex2.v

Vector1.scale( delta2 );

Vector2.scale( delta1 );

tangentSpaceY.set( Vector1 - Vector2 )

tangentSpaceY.normalize();

tangentSpaceX.Cross3( tangentSpaceY, MainTriangle.normal );

tangentSpaceX.normalize3();

tangentSpaceZ.Cross3( tangentSpaceX, tangentSpaceY );

tangentSpaceZ.normalize();

After this code, tangentSpaceX,Y,Z should contain a matrix of "negative" tangentspace. Anyways, I tried it and it works, however I dont understand the following line:

tangentSpaceY.set( Vector1 - Vector2 )

how does that generate the tangent space vector for Y(or V)

Vec Vector1 = ( vertex3 - vertex2 )

Vec Vector2 = ( vertex1 - vertex2 )

float delta1 = vertex3.v - vertex2.v

float delta2 = vertex1.v - vertex2.v

Vector1.scale( delta2 );

Vector2.scale( delta1 );

tangentSpaceY.set( Vector1 - Vector2 )

tangentSpaceY.normalize();

tangentSpaceX.Cross3( tangentSpaceY, MainTriangle.normal );

tangentSpaceX.normalize3();

tangentSpaceZ.Cross3( tangentSpaceX, tangentSpaceY );

tangentSpaceZ.normalize();

After this code, tangentSpaceX,Y,Z should contain a matrix of "negative" tangentspace. Anyways, I tried it and it works, however I dont understand the following line:

tangentSpaceY.set( Vector1 - Vector2 )

how does that generate the tangent space vector for Y(or V)