Help with Program Structure

I structure it like this. There is a single class that handles rendering. It makes all the GL calls and it has all the information it needs (camera, projection, etc)

Other people structure it like you do. They have various class objects that render themselves.

Thanks V-man,

yea I think I might re-structure as your way makes a bit more sense to me,

so I'll have a rendering class and then how do you actually draw things? Do you have a rendering list or something similar and loop through drawing each object in the list? Do you store all the drawing info such as verts, normals, texCoords stuff in each object or do you also store all this in your renderer? Can you give me a bit more info as to how you do it or a link to an example of program structuring[color="#1c2837"] please.

Thanks
x

Ok so I found this -

http://stackoverflow.com/questions/166356/what-are-some-best-practices-for-opengl-coding-esp-w-r-t-object-orientation


which gives a nice explanation I think. I think what I'm going to do is have a mesh class which will deal with my VBOs etc for each object. A material class which will deal with all my needed shaders and properties for them for each object and then a renderer class which will be used to actually draw using all the information provided inside these two classes.
So each 'gameObject' in my world will not have any openGL functionality in it but will store the information inside these two classes.

Is this a good way of structuring my programs? Can anyone suggest improvements or pitfalls I have not foreseen?

Thanks
x

It is a good idea not to directly call OpenGL functions from within your models.
You should have a graphics module which handles all of the OpenGL functions and makes sure that, for example, redundant states are not set.
The rest of your modules rely solely on the graphics module for rendering and never directly touch OpenGL.

The graphics module also provides a few helper features such as render queues which it can sort for faster rendering.

The graphics module should provide wrappers for vertex and index buffers that make them easy to create and use. Same goes for shaders.
Your models can and should have their own vertex and index buffers and shaders. The models themselves should decide how they should be drawn. They submit the graphics settings needed for rendering such as lighting on/off and which shader will be used, the graphics module sorts based on these settings, sets these states, and then makes a call back to the model to allow it to perform the actual render on the vertex/index buffers.
Some people prefer to submit the index buffers and vertex buffers along with the rest of the data and let the graphics module perform the entire render. I have found that having the models activate and render their own buffers is both faster and more flexible.


L. Spiro