Papervision3D - 3.0
Switch branches/tags
Nothing to show
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
src
.gitignore
README

README

Papervision3D - 3.0

This code is for discussing the core architecture.

Your comments are welcome!

Discussion:
==============================================
geometry:
VertexGeometry holds a mesh's vertices and maintains some Vector.<Number> arrays
which we can pass easily to Utils3D.projectVectors() 

camera:
The camera *should* be part of the scenegraph (ie: scene.addChild(camera);)

projection:
* assuming a righthanded system.
* uses the default OpenGL projection matrix (gluPerspective)
  => Flash's native PerspectiveProjection is not used on purpose.
     => This is up for grabs. Think its not well suited for clipping etc. (I'd rather work in normalized clip space)
* while traversing the scenegraph matrices are post-multiplied (ie: earth.rotationY++ will orbit sun)
* we need a mechanism to distinguish between "local" and "global" rotations.
  => maybe roll / pitch / yaw to make earth spin its own axis (local)
  => maybe rotationX / Y / Z to make earth spin its parent (sun, global)
  => or do we? Maybe I'm missing something here.
* projection is divided in some steps:
  1. transform all objects including camera to world space
  2. update camera: calculate its inverse world transform
     => cull objects by bbox / bshpere outside camera frustum
  3. transform all objects to view / camera space
     a] check whether objects are behind, in front or straddling the near-plane, clip if needed
        => think this can only be done in view / camera space *before* the perspective transform
     b] cull triangles outside view
  4. perspective divide (mult the view matrix with the projection matrix)
  5. draw
  
  Clipping can be done in multiple spaces: we can clip triangles in world, view(?) or in screen space.
  For clipping in screen space we *must* be sure all vertices are in front of the near plane.
  => ie: all triangles straddling the near plane should have been clipped already
  Clipping in world space has advantage that we can move from object to screen space in one go.
  Clipping in view space: not sure, its ideal for clipping to the near plane