PDA

View Full Version : Need emergency help with setting Normal vectors!



rovolution
06-28-2010, 11:28 PM
Hello all,


I have a huge question about setting normal vectors for coned surfaces.



So I have a piece of code where i want to draw a tree by approximating it with 4 triangles


glBegin( GL_TRIANGLES);
glColor3f(0, 1, 0);


double yval = 0.5;



//One triangle side
glTexCoord2f( leafRep/2,leafRep);
glVertex3f(0,1,0);

glNormal3f(-1, yval, -1);
glTexCoord2f(leafRep,0.0);
glVertex3f(-1,0,-1);

glNormal3f(1, yval,-1);
glTexCoord2f(0.0 ,0.0);
glVertex3f(1,0,-1);


glEnd();


glBegin( GL_TRIANGLES);

//another side


glNormal3f(0, 1, 0);
glTexCoord2f( leafRep/2,leafRep);
glVertex3f(0,1,0);


glNormal3f(1, yval,-1);
glTexCoord2f(leafRep,0.0);
glVertex3f(1,0,-1);



glNormal3f(1, yval, 1);
glTexCoord2f(0.0 ,0.0);
glVertex3f(1,0,1);


glEnd();


glBegin( GL_TRIANGLES);


//another side




glNormal3f(0, 1, 0);
glTexCoord2f( leafRep/2,leafRep);
glVertex3f(0,1,0);

glNormal3f(1, yval, 1);
glTexCoord2f(leafRep,0.0);
glVertex3f(1,0,1);

glNormal3f(-1, yval, 1);
glTexCoord2f(0.0 ,0.0);
glVertex3f(-1,0,1);

glEnd();




glBegin( GL_TRIANGLES);


//another side


glNormal3f(0, 1, 0);
glTexCoord2f( leafRep/2,leafRep);
glVertex3f(0,1,0);

glNormal3f(-1, yval, -1);
glTexCoord2f(leafRep,0.0);
glVertex3f(-1,0,-1);

glNormal3f(-1,yval,1);
glTexCoord2f(0.0 ,0.0);
glVertex3f(-1,0,1);


glEnd();


I have been told that i am setting the normals incorrectly.

So i went on the internet and found a number of tutorials on Gourad Shading and how I should calculate the normals for each triangular surface and then calculate the average of the surface normals that intersect at a particular vertice to get the proper normal vector for that vertice.

But then I was told this:


This technique is only necessary if all you know is the what the triangles
are. When an object is defined ONLY in terms of the triangles, then
Gouraud requires that you calculate the normals that way.

However, when drawing a cylinder or cone or some object which is
analytically defined, you don't have to do that because you KNOW what the
normal is. Using symmetry, it is very easy to show that the answer you
get when averaging the inidividual normals is in fact the true normal.

So the bottom line is if you apply the technique you outline, you will
laboriously calculate a normal which you can determine by inspection.



I do not understand how that applies to my code because i did not use any kind of parametic equation involving sins or cosines to generate my shape.

How would i properly calculate the normals for the vertices to show a "curved effect" without going through the whole averaging the surface normals technique?

Is there even such a way? Do i have to calculate the slope from the top of my tree to one of the edge vertices and then figure something out. I am really confused :(