Processing hierarchical eggfiles for Octree optimization

Pipeline
1

Hi,

I’ve been dealing with a conversion of a big egg file into octrees to optimize collision processing, and tried to use the script eggOctree.py

The issue is that if the egg file has a hierarchy of Groups and vertexPool this script doesn’t generate anything

example for such a rejected eggfile

<Group> pipo {
 /* <Dart> { 1 } */
  <Group> groundPlane_transform {}
  <Collide> { Polyset keep descend }
  <Group> Model {
    <Group> Terrain1 {
      <Group> Mesh1 {
        <VertexPool> Mesh1Shape.verts { 
.......
      <Group> Mesh2 {
        <VertexPool> Mesh2Shape.verts {
.......
    <Group> Terrain2 {
.......

I’ve made some progress in pruning the eggfile in a hierarchical way. Here is the tentative code added to eggOctree.py

def findVertexPoolsIn(group,groupList,vertexPoolList):
    for child in iterChildren(group):
        if type(child) == EggGroup:
            findVertexPoolsIn(child,groupList,vertexPoolList)
        if type(child) == EggVertexPool:
            groupList.append(group)
            vertexPoolList.append(child)  
           
def octreefy(infile,outfile):
    """
        octreefy infile and write to outfile
        using the buildOctree functions
    """
    egg = EggData()
    egg.read(Filename(infile))
    eggStripTexture(egg)

    # output eggdata
    ed = EggData()
    ed.setCoordinateSystem(egg.getCoordinateSystem())

    ### THIS IS BASICALLY WHERE THE MODIF OCCURS
    # fill list of group and associated vertexPools
    groupList = []
    vertexPoolList = []
    findVertexPoolsIn(egg,groupList,vertexPoolList)

    # include vertexPools
    for vertexPool in vertexPoolList:
        ed.addChild(vertexPool)

    # now for each embedded group let's octreefy
    ref = 0 # will serve as a group identifier
    for group in groupList:
        ed.addChild(buildOctree(group,`ref`))
        ref = ref + 1

    # finalize: write new egg file
    if listResultingEgg: eggLs(ed)
    ed.writeEgg(Filename(outfile))

An output file is generated, but to be frank I don’t have a clue as to know if it is really optimized or not.

Could one of our Panda’s gurus give a quick look at it and state what is right or what is wrong.

Many thanks

Jean-Claude :slight_smile: