hello all,

i am reading vertex data from a game file that stores a character model in triangle strips (ps2 game, hence why triangle strips are used). i did concider converting the triangle strips into vertex indexing to help ease the pain of calculating my normals at runtime, but still would like some help calculating normals for triangle strips.

i recently read on here that triangle strips flip cw/ccw culling every triangle if backface culling is on. below is my code for calculating normals at the moment;

[source lang="cpp"]
float vu[3];
float vv[3];
float cp[3];
float result;

for (int k = 0; k < nvertices; k++) {
vu[0] = vertices[k + 1][0] - vertices[k][0];
vu[1] = vertices[k + 1][1] - vertices[k][1];
vu[2] = vertices[k + 1][2] - vertices[k][2];

vv[0] = vertices[k + 1][0] - vertices[k + 2][0];
vv[1] = vertices[k + 1][1] - vertices[k + 2][1];
vv[2] = vertices[k + 1][2] - vertices[k + 2][2];

cp[0] = (vu[2] * vv[1]) - (vu[1] * vv[2]);
cp[1] = (vu[0] * vv[2]) - (vu[2] * vv[0]);
cp[2] = (vu[1] * vv[0]) - (vu[0] * vv[1]);

result = (float)sqrt((cp[0]*cp[0]) + (cp[1]*cp[1]) + (cp[2]*cp[2]));

if (result == 0.0f)
result = 1.0f;

normals[k][0] = cp[0] / result;
normals[k][1] = cp[1] / result;
normals[k][2] = cp[2] / result;
}
[/source]

when i run my program with the above code, i see the following;



&&



you can see on this side shot that i can see through the model (less than desireable.) at the moment i am iterating through every vertex, do i need an offset in my loop to calculate the next triangle or next vertex? it looks like the normals are facing away from each other...

and this is the drawing code...

[source lang="cpp"]
glPushMatrix();
glTranslatef(posx_, posy_, posz_);
glRotatef(90.0f, 0, 0, 1);
glRotatef(turnrate_, 1, 0, 0);
glBegin(GL_TRIANGLE_STRIP);
for (int i = 0; i < model_.nvertices; i++) {
glNormal3f(model_.normals[0], model_.normals[1], model_.normals[2]);
glVertex3f(model_.vertices[0], model_.vertices[1], model_.vertices[2]);
}
glEnd();
glPopMatrix();
[/source]


thanks for your time,


best regards,

justin

thank you both for the replys. i think i know what i need to do, just not sure what order i need to read my vu and vv variables when alternating?

so to clarify for myself... for the first alternating read, what vertices do i want vu and vv to be? vu is 2-3? vv is 2-0? what about the second alternating read? vu is ?-? and vv is ?-? does the cross product also change on each alternating read? or just the vertices i read into vu and vv? i thought i had this in concrete, but the more sample code on calculating normals i find, the more confused i become. every source code calculating normals i find reads the vertices in a different order... i've now changed my vu vv and cross product dozens of times...!

correct me if i am wrong, going by the diagram the first read is vertex 1, 2 then 3. and the second read is 3, 4 then 2? cw then alternate ccw?

[source lang="cpp"]
if (flip > 0) {
flip = 0;

vu[0] = vertices[1][x] - vertices[2][x];
vu[1] = vertices[1][y] - vertices[2][y];
vu[2] = vertices[1][z] - vertices[2][z];

vv[0] = vertices[1][x] - vertices[0][x];
vv[1] = vertices[1][y] - vertices[0][y];
vv[2] = vertices[1][z] - vertices[0][z];
}
else {
flip = 1;

vu[0] = vertices[?][x] - vertices[?][x];
vu[1] = vertices[?][y] - vertices[?][y];
vu[2] = vertices[?][z] - vertices[?][z];

vv[0] = vertices[?][x] - vertices[?][x];
vv[1] = vertices[?][y] - vertices[?][y];
vv[2] = vertices[?][z] - vertices[?][z];
}
[/source]

thank you Lauris Kaplinski for the great reply!

i did as you mentioned and have a better result, just not the right one. i tried changing odd to even and (k & 1) is the only one that gives near results. i've taken a screenshot to show what i see...



&&



i can see through the mesh which i hope is a result of incorrect normals, or else something in my rendering section is broken!

[source lang="cpp"]
float vu[3];
float vv[3];
float cp[3];
float result;

for (int k = 0; k < nvertices - 2; k++) {
vu[0] = vertices[k + 1][0] - vertices[k][0];
vu[1] = vertices[k + 1][1] - vertices[k][1];
vu[2] = vertices[k + 1][2] - vertices[k][2];

vv[0] = vertices[k + 2][0] - vertices[k][0];
vv[1] = vertices[k + 2][1] - vertices[k][1];
vv[2] = vertices[k + 2][2] - vertices[k][2];

cp[0] = (vu[1] * vv[2]) - (vu[2] * vv[1]);
cp[1] = (vu[2] * vv[0]) - (vu[0] * vv[2]);
cp[2] = (vu[0] * vv[1]) - (vu[1] * vv[0]);

result = (float)sqrt((cp[0]*cp[0]) + (cp[1]*cp[1]) + (cp[2]*cp[2]));

if (result == 0.0f)
result = 1.0f;

normals[k][0] = cp[0] / result;
normals[k][1] = cp[1] / result;
normals[k][2] = cp[2] / result;

if (k &amp;amp; 1) {
normals[k][0] = -normals[k][0];
normals[k][1] = -normals[k][1];
normals[k][2] = -normals[k][2];
}
}
[/source]

above i changed the cross product from what i had originally. a few articles i read on calculating normals seem to use the same cross product. i hope its not incorrect. changing it gives me a completely dark shaded model.

*scratches head*