In learning real-time 3D graphics programming with OpenGL, how can I tell what methods are antequated? For example, I found out that masking is not really used. I also assume the stipple and bitmap stuff probably isn't used that much anymore. What about fog? What about the selection and feedback modes? And frame buffers are the replacement for pixel buffers? What about indexed color modes and associated functions and pixel manipulation? Vertex buffers are preferred over vertex arrays now? Also what are pixel fragments, is this a rasterized primitive in whole or in part? Also is it easier to go OGL -> D3D or D3D -> OGL in terms of which you learned first? Also I can never understand RHW and how W changes as it goes through the pipeline and what it is used for. Is there any way to understand this without writing a complete software renderer?

Of course, answering your questions is influenced heavily by application in mind and partly also by personal preference. The following is my sight.

Quote:Original post by Boder
For example, I found out that masking is not really used. I also assume the stipple and bitmap stuff probably isn't used that much anymore.

Stencilling has still the one or other use case. E.g. recently there was a thread asking about on-the-fly CSG. Bitmaps are sometimes used e.g. for fonts, but there are faster methods. I personally use small bitmaps for displaying markers for vertices, axes, control points, and so on in my editor. Say, the usefullness depends on your application.

Quote:Original post by Boder
What about fog?

The fog you presumably mean is a pure view dependent thing, say there is no volume restriction (like "ground fog" or such). Hence its use cases are restricted. I use the fog only for view depth restriction in my engine currently.

Quote:Original post by Boder
What about the selection and feedback modes?

In the past 30 days I remember of 3 threads dealing with them, so it can't be totally obsolete ;) I personally don't use them atm.

Quote:Original post by Boder
And frame buffers are the replacement for pixel buffers?

No. Frame buffer is the normal color pixel buffer (say the x/y part of the display, extended by the depth buffer into the 3rd dimension): An "on-screen" rendering target. Pixel buffers are a kind of "off-screen" rendering target for algorithms that need to render intermediate steps somewhere.

Quote:Original post by Boder
What about indexed color modes and associated functions and pixel manipulation?

Nowadays I think the application of indexed color modes are seldom.

Quote:Original post by Boder
Vertex buffers are preferred over vertex arrays now?

In general, yes.

Quote:Original post by Boder
Also is it easier to go OGL -> D3D or D3D -> OGL in terms of which you learned first?

Well, D3D already provides help to deal with many details (like vectors and matrices) what OGL doesn't. If you have programmed OGL you were forced to do much stuff yourself, and hence you must have background knowledge then. So IMHO using D3D later may be easier. On the other hand, there're people saying that due to the same reason it is easier to learn D3D first, since it allows a faster way of doing things.

Quote:Original post by Boder
Also I can never understand RHW and how W changes as it goes through the pipeline and what it is used for. Is there any way to understand this without writing a complete software renderer?

Err, now you got me ;) What means RHW? Okay, w is the typical variable name for the homogeneous co-ordinate. Do you mean that? If so, understanding homogeneous co-ordinates is IMHO easy if you have some background of vector math already, and knowing affine transformations may help also but isn't really required (but else you have to learn it besides).

Quote:Original post by haegarr

Quote:Original post by Boder
Also is it easier to go OGL -> D3D or D3D -> OGL in terms of which you learned first?

Well, D3D already provides help to deal with many details (like vectors and matrices) what OGL doesn't. If you have programmed OGL you were forced to do much stuff yourself, and hence you must have background knowledge then. So IMHO using D3D later may be easier. On the other hand, there're people saying that due to the same reason it is easier to learn D3D first, since it allows a faster way of doing things.

Things that speak for OpenGL IMHO:

- immediate mode: For lerning and experimenting it is wonderull

- low setup overhead: Create a window, add a gl context, set up the perspective transformation and start drawing.

- clean and (mostly) consisten interface: It is easy to guess how something is called after a while, which is always a Good Thing.

- well documented: also a Good Thing

As for the math you have to learn them anyway, doing so from start is better, because you will already know them when you will realy need them.

I would sugest OGL -> D3D but I think it doesn't realy matter that much.

Take it all with a grain of salt, since I havent used D3D and don't plan on doing so. My Eyes hurt just by looking at D3D code (and its not supported on my OS anyway).