visualpython-users

Re: [Visualpython-users] a wish or two

[<Art...@ao...>]

David wrote -

>Adding a primitive in C++ that would behave similarly but more efficiently
>or triangles would be relatively easy.

>Alpha is harder, because to render translucent surfaces accurately, you need
>to draw them from back to front.  At present, Visual doesn't need to sort
>most primitives (because they are drawn opaquely, and can be zbuffered).  If
>you want to attack this, I can give you some pointers - but it will take
>some effort.

An overview of the cvisual's rendering approach might help me get started. 
What
I am seeing in the code, to the extent I follow it, is not what I would 
expect.  If I look
at the code for a glut sphere or cone I see it being built geometrically from 
triangle fans and strips. 

Cvisual seems to look more like ray-tracing code - which I have never been 
able to
follow - though wish I could.    
 
Is there anything available that might help me better understand cvisual at 
this level?

ART

[Visualpython-users] re: a wish or two

[<Art...@ao...>]

>>There are basically 3 kinds of code in Visual primitives .. (snip)

That helps.

I had already mentioned to someone that if I were to try to learn
Python extensions I would probably choose to do it via cvisual -
the general subject matter is one I am already famialiar with, and 
I have confidence it represents a clean methodology. And there is
real impetous - cvisual probably already having 90% of the specifics
to implement my interests, but might have to go it a bit on my own 
after that

So I will move in that direction as time allows.

I will try to cry for help only when at the end of a rope.

Looks like some getting some understanding of CXX is a
logical place to start.  We're off.

ART

RE: [Visualpython-users] a wish or two

[David S. <dsc...@vy...>]

> An overview of the cvisual's rendering approach might help me get
> started.

I will help as I can.

> Is there anything available that might help me better understand
> cvisual at
> this level?

There is no technical documentation, sorry :(

> What
> I am seeing in the code, to the extent I follow it, is not what I would
> expect.  If I look
> at the code for a glut sphere or cone I see it being built
> geometrically from
> triangle fans and strips.

There are basically 3 kinds of code in Visual primitives

(1) code to implement the Python interface (attributes + methods).  We use
the CXX library, which wraps C++ classes around Python types and vice-versa,
to make this easier.

(2) code to render the primitive (using triangle strips or other GL
primitives).  This will look a little like the implementations you are used
to, but Visual cannot use OpenGL matrix math (it is too inaccurate when the
scale of the scene is very large or small, which is important for physics).
Therefore the code does transformation and lighting itself using the tmatrix
class.

(3) code to implement mouse picking. This is raytracing code.  The line
between the camera position and the mouse cursor forms a ray which is
intersected with all objects in the scene.  The closest intersection is
taken to be mouse.pick/mouse.pickpos.

> Cvisual seems to look more like ray-tracing code - which I have
> never been able to follow - though wish I could.

Actually, a simple raytracer is a relatively gentle introduction to
graphics, if you find one written for pedagogical value rather than for
efficiency.  Maybe someone needs to write one in Python.

Dave