7 replies to this topic

[arithma]

I'm having trouble sending both normals and a u,v pair to my shaders. If I remove the normal, things work as expected.

It appears the v_normal is receiving the values that are intended for v_coord. I still have no idea though.
I've spent a lot of time trying to figure out, but I really am not able to pick a sane set of actions to debug/experiment anymore.
This is my vertex:

[source lang="cpp"]struct Vertex{ Vertex(vec3 const & v) : pos(v) {} vec3 pos; vec3 normal; real u, v;};[/source]
[source lang="cpp"] const int VERTEX_POS_INDX = 0; const int VERTEX_NORMAL_INDX = 1; const int VERTEX_TEXCOORD_INDX = 2; const int VERTEX_POS_SIZE = 3; const int VERTEX_NORMAL_SIZE = 3; const int VERTEX_TEXCOORD_SIZE = 2; GLuint vbo, ibo; glGenBuffers(1, &vbo); glBindBuffer(GL_ARRAY_BUFFER, vbo); glBufferData(GL_ARRAY_BUFFER, sphere->vertices.size()*sizeof(Vertex), &sphere->vertices[0], GL_STATIC_DRAW); glGenBuffers(1, &ibo); glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, ibo); glBufferData(GL_ELEMENT_ARRAY_BUFFER, sphere->indices.size()*sizeof(unsigned short), &sphere->indices[0], GL_STATIC_DRAW); glEnableVertexAttribArray ( VERTEX_POS_INDX ); glEnableVertexAttribArray ( VERTEX_NORMAL_INDX ); glEnableVertexAttribArray ( VERTEX_TEXCOORD_INDX ); int offset = 0; glVertexAttribPointer ( VERTEX_POS_INDX, VERTEX_POS_SIZE, GL_FLOAT, GL_FALSE, sizeof(Vertex), (void*)offset ); offset += VERTEX_POS_SIZE * sizeof(real); glVertexAttribPointer ( VERTEX_NORMAL_INDX, VERTEX_NORMAL_SIZE, GL_FLOAT, GL_FALSE, sizeof(Vertex), (void*)offset ); offset += VERTEX_NORMAL_SIZE * sizeof(real); glVertexAttribPointer ( VERTEX_TEXCOORD_INDX, VERTEX_TEXCOORD_INDX, GL_FLOAT, GL_FALSE, sizeof(Vertex), (void*)offset ); glBindAttribLocation ( programObject, VERTEX_POS_INDX, "a_position" ); glBindAttribLocation ( programObject, VERTEX_NORMAL_INDX, "a_normal" ); glBindAttribLocation ( programObject, VERTEX_TEXCOORD_INDX, "a_coord" );[/source]
[source lang="cpp"]precision highp float;uniform mat4 u_mv;uniform mat4 u_mvp;uniform vec3 u_light;uniform vec3 u_up;attribute vec3 a_position;attribute vec2 a_coord;attribute vec3 a_normal;varying vec2 v_coord;varying vec3 v_normal;void main() { v_coord = a_coord; v_normal = a_normal; gl_Position = u_mvp * vec4(a_position, 1);}[/source]
[source lang="cpp"]precision highp float;uniform vec3 u_up;varying vec3 v_normal;varying vec2 v_coord;precision highp float;uniform vec3 u_up;varying vec3 v_normal;varying vec2 v_coord;void main(){ vec2 coord = v_coord; vec3 normal = v_normal; coord.x = mod(v_coord.x * 5.0, 1.0); coord.y = mod(v_coord.y * 5.0, 1.0); gl_FragColor = vec4 ( mod(coord.x*1.0,1.0), mod(coord.y*1.0,1.0), mod(normal.z*5.0,1.0)*0.0, 1.0 );}[/source]

[szecs]

Your Vertex struct is not pure data type. So you can't assume that
offset += VERTEX_POS_SIZE * sizeof(real); and the other will be the offset you expect.

[larspensjo]

A check: You are doing the linking (glLinkProgram) of the shader program after the call of glBindAttribLocation(), don't you?


I usually do the glVertexAttribPointer offset as follows:

struct Vertex *p = 0;
glVertexAttribPointer ( VERTEX_POS_INDX, VERTEX_POS_SIZE, GL_FLOAT, GL_FALSE, sizeof(Vertex), &p->pos);
glVertexAttribPointer ( VERTEX_NORMAL_INDX, VERTEX_NORMAL_SIZE, GL_FLOAT, GL_FALSE, sizeof(Vertex), &p->normal );
glVertexAttribPointer ( VERTEX_TEXCOORD_INDX, VERTEX_TEXCOORD_INDX, GL_FLOAT, GL_FALSE, sizeof(Vertex), &p->u );

Edited by larspensjo, 26 June 2012 - 06:53 AM.

[mhagain]

Better again to use the GL3.x+ syntax (if available in your implementation) for this in your shader:

layout(location = 0) in vec4 position;

Edited by mhagain, 26 June 2012 - 07:14 AM.

[arithma]

@szecs: sizeof(Vertex) returns 32, which is 8*sizeof(float).
@larspensjo: I tried both tricks (which are really nifty) but to no avail.
@mhagain: I am targeting opengl es 2 on iPhone, so OGL3 is unfortunately not an option.

[arithma]

Got it solved!
I just had to use glGetAttributeLocation to get the identifiers from the linked shader program.

[larspensjo]

Got it solved!
I just had to use glGetAttributeLocation to get the identifiers from the linked shader program.

Nice to see! But that means that the call to

glBindAttribLocation() wasn't done before the linkage phase, or the identifiers would have been the one you specified.

If you are using the same VAO for more than one program, you can't use

glGetAttributeLocation() as the results may be different.

[arithma]

@larspensjo: Thanks, got it!

A check: You are doing the linking (glLinkProgram) of the shader program after the call of glBindAttribLocation(), don't you?

I thought you were warning me of a pitfall rather than reminding me to do the right thing.