It'd be easiest to have some kind of quad/sprite manager, that keeps track of a VB, and adds quads to itself. How you do this is up to you, but it should be fairly straightforward. There's nothing saying that one VB needs to be associated with one C++ class instance either.
If it were me, I'd make a CVertexBuffer class, that holds the IDirect3DVertexBuffer9 pointer, the size of the buffer, and the next offset to add vertices to. Then you could pass a CVertexBuffer pointer to your iQuad class to render into.

You can have as many vertex buffers as you like, but there's a performance penalty involved in switching vertex buffers (With SetStreamSource()), so it's best to have as few as possible. If you're using the fixed-function pipeline, then you'll really need one vertex buffer for each vertex type.

In a rendering engine I made (which I lost the source code of, unfortunately), I could call CRender::CreateVertexBuffer(), which would return a class representing a section of a IDirect3DVertexBuffer9. This way I could cram several CVertexBuffers into one IDirect3DVertexBuffer9 more or less transparently.

Hope this helps.

Quote:Original post by Krisc
How would I go about this? Do I keep track of the vertices added in a seperate array and then add them all to the buffer at once? When I create the buffer, all I see is a function SetData and nothing along the lines of Add() which will allows me to add on to the end of the buffer data.

The way ID3DXSprite works (you might be better using that actually) is when you call ID3DXSprite::Begin(), it locks its internal vertex buffer. Then, for every Draw() call you make, it adds the quad to the vertex buffer. If it runs out of space in the vertex buffer, or you call call ID3DXSprite::End(), it'll unlock the VB, and draw all of the sprites.

So, you could either just add all of the vertices to an array, then lock your vertex buffer, copy the vertices, unlock and render, or you could lock the VB, and add vertices to it, then when you come to render you unlock the VB and render.

When I say "add vertices to the VB", I mean lock the vetex buffer and copy vertices into it. Or just lock the whole buffer at the start, and copy vertices into the next unused part of the buffer when you need.