Understanding VAOs and adding different arrays to VAOs

Question

I'm really confused on what you do. I can do them, however I got this problem. Say you have several squares, say 1,000 squares. Now I can make a VAO for each 1,000 squares and then do some for loop to render all of the squares. I can also use shaders to move the squares.

However, is it possible to put all 1,000 squares in one VAO.

The problem is I'm going to create a voxel game. So I need to render about 10,000 cubes. However, the game will lag a lot even if I render 200 squares how I'm currently doing it.

float points[] = {
    0.0f, 0.0f, 0.0f,
    0.3f, 0.0f, 0.0f,
    0.3f, 0.3f, 0.0f,
    0.0f, 0.0f, 0.0f,
    0.0f, 0.3f, 0.0f,
    0.3f, 0.3f, 0.0f
};

float colours[] = {
    1.0f, 0.0f,  0.0f,
    1.0f, 0.0f,  0.0f,
    1.0f, 0.0f,  0.0f,
    1.0f, 0.0f,  0.0f,
    1.0f, 0.0f,  0.0f,
    1.0f, 0.0f,  0.0f
};

unsigned int points_vbo = 0;
glGenBuffers (1, &points_vbo);
glBindBuffer (GL_ARRAY_BUFFER, points_vbo);
glBufferData (GL_ARRAY_BUFFER, 18 * sizeof (float), &points, GL_STATIC_DRAW);

unsigned int colours_vbo = 0;
glGenBuffers (1, &colours_vbo);
glBindBuffer (GL_ARRAY_BUFFER, colours_vbo);
glBufferData (GL_ARRAY_BUFFER, 18 * sizeof (float), &colours, GL_STATIC_DRAW);

unsigned int vao = 0;
glGenVertexArrays (1, &vao);
glBindVertexArray (vao);
glBindBuffer (GL_ARRAY_BUFFER, points_vbo);
glVertexAttribPointer (0, 3, GL_FLOAT, GL_FALSE, 0, (GLubyte*)NULL);
glBindBuffer (GL_ARRAY_BUFFER, colours_vbo);
glVertexAttribPointer (1, 3, GL_FLOAT, GL_FALSE, 0, (GLubyte*)NULL);

glEnableVertexAttribArray (0);
glEnableVertexAttribArray (1);

//Setup shaders
std::string vertex_shader = loadshaders("test_vs.txt");
std::string fragment_shader = loadshaders("test_fs.txt");

unsigned int vs = glCreateShader (GL_VERTEX_SHADER);
const char* str = vertex_shader.c_str ();
glShaderSource (vs, 1, &str, NULL);
glCompileShader (vs);
unsigned int fs = glCreateShader (GL_FRAGMENT_SHADER);
const char* strb = fragment_shader.c_str ();
glShaderSource (fs, 1, &strb, NULL);
glCompileShader (fs);

unsigned int shader_programme = glCreateProgram ();
glAttachShader (shader_programme, fs);
glAttachShader (shader_programme, vs);
glLinkProgram (shader_programme);

// main loop
while (!glfwWindowShouldClose(window))
{
    float ratio;
    int width, height;
    glfwGetFramebufferSize(window, &width, &height);

    glClear (GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
    glClearColor (0.6f, 0.6f, 0.8f, 1.0f);
    glUseProgram (shader_programme);
    glBindVertexArray (vao);
    glDrawArrays (GL_TRIANGLES, 0, 6);

    glfwSwapBuffers(window);
    glfwPollEvents();
}

So how would I say add points2[], points3[] and put them into the same VAO. How can I design my shader so I can then do transformation on the points in the arrays say translate points2[] array 4 blocks right, without effecting every other point.

Answer 1

You can do it in the shader but it makes things really complicated.

So the alternative i think of is to bake your geometry into one VBO/VAO for say... 16x16x16 boxes/voxels. If you need to change that data just change it and rebake the VBO/VAO for it.

You need also some abstractions for the Shader Objects, a thing i call Buffer (Manages the VBO/VAO for the gometry) and in Buffer you need to dynamically allocate/reallocate/free memory, don't do it with static arrays.