So I have a base class of "actors" that will be drawn to the screen,which includes basic stuff like vertex,index buffers,some data types,information about geometry(index vertex count etc.). So here's my question:

Should I pass ID3D11Device* and ID3D11DeviceContext* to classes derived from base class and simply let this classes handle all rendering by themselves in a method such as virtual void Render( ID3D11DeviceContext* ),

or

Should I create a renderer class which takes this derived classes and their data and implement their rendering logic?

or better suggestions? I am building a huge project and I have never been worried about efficiency so I need your help.

I am with L. Spiro here, the renderer should not have to know anything about how a model will be rendered, instead he will know about how are all the stuff in the world rendered:

1. Order of the models and the type being rendered ( fe: First the non alpha. then alpha models )

2. which effects are being applied and in which order.

3. Post effects applied

It is like a mix of multiplexor, manager and organizer system.

Hope it helps.