CODEWITHSUNDEEP
c++opengl
greedsin
greedsin

Reputation: 1272

OpenGL Cube Structure

Iam new to OpenGL and also new to C++ . I created a main.cpp where i do everything .I Create an GLfloat data[] where i store my Cube's Vertices and also one GLfloat color[] for the Color. I create a VAO and a VBO and i have method, where i compile a simple Vertex and Fragment Shader.

The problem is,that my Code is very long,so i thought about creating a Cube Class,Triangle Class and so on.

My Question is now, can i use the same VAO for several Objects or do i use a new VAO for each Object?.

The other thing is, that i dont exactly know how to program this.I'am missing the structure(I used Java before).

My thoughts were :

class Cube {
   public:
     Cube();
     Cube(GLfloat position[],GLfloat color[]);
     //But i need the same Shaders for each Object , so do i initialize it here or somewhere else?
     compileShader();
     drawCube();

}

Note that this are only my thoughts about it..Maybe someone can help me out with this :)

Upvotes: 2

Views: 2050

Answers (1)

lassombra
lassombra

Reputation: 427

VAO and VBO can be reused but for performance only should be if they aren't changing (for example, if you're keeping a 1x1x1 cube but using transformations to draw additional ones).

For a simple cube in a small application Your ideal structure would be something like

class Cube{
  private static int vbo;
  private static int vao;
  public Cube(){
    //If vbo and vao are 0 initialize them here
  }
  public void render(mat4* transform){
  }
}

The shader would be a separate class, which can either be part of the rendering call for this object public void render (mat4* transform, shader* shader) or can be called by the main pipeline.

Now, I would recommend reading about header files in c++. In c++ headers are the right way to develop especially if you are going to be sharing your class around your program. Other classes need to only inherit the header, they don't have to care about the implementation, and that way your code only gets compiled once and you're using the same compiled assembly code throughout your class. (In c++ if you include actual code, it gets compiled, and even though modern linkers are good about merging duplicate code, it's still bad practice)

Cube.h:

class Cube{
  static int vbo;
  static int vao;
public:
  Cube();
  void render(mat4* transform);
}

cube.cpp:

Cube::Cube(){
  //Initialize Here
}
Cube::render(mat4* transform){
  //Render here
}

Then you'd also want a pair for shader shader.h:

class Shader{
  int ProgramID;
public:
  Shader();
  void activate ();
}

shader.cpp:

Shader::Shader(){
  //compile shader here
}
Shader::activate(){
  //make this shader program active here
}

The mat4 data for the transform that I mentioned came from http://glm.g-truc.net/0.9.6/index.html which is a great library for handling the math of 3d graphics.

If you want to refactor this further, you can define the render function of cube in an interface that all of your graphic objects will inherit, allowing you to more dynamically instantiate and configure the objects, and not care about what types they are, just that they're of the interface required.

Upvotes: 2

Related Questions