Controlled Topology Simplification (Voxel Based Object Simplification) - PowerPoint PPT Presentation

About This Presentation
Title:

Controlled Topology Simplification (Voxel Based Object Simplification)

Description:

Controlled Topology Simplification (Voxel Based Object Simplification) T. He, L. Hong, A. Varshney, S. Wang Udeepta Bordoloi Why Simplification? Levels of Detail High ... – PowerPoint PPT presentation

Number of Views:53
Avg rating:3.0/5.0
Slides: 18
Provided by: Preferr1378
Category:

less

Transcript and Presenter's Notes

Title: Controlled Topology Simplification (Voxel Based Object Simplification)


1
Controlled Topology Simplification(Voxel Based
Object Simplification)
  • T. He, L. Hong, A. Varshney, S. Wang
  • Udeepta Bordoloi

2
Why Simplification?
  • Levels of Detail
  • High frame rate
  • Removal of high frequencies (anti-aliasing)
  • Moire patterns

3
Conceptually
  • Simplification can be in two areas
  • Geometry
  • Reduce the number of geometry primitives
  • Topology
  • Reduce the number of holes, tunnels etc.

4
The factor of Topology
  • Simplifying
  • Will change the topology!!!
  • Preserving
  • Will limit the amount of simplification possible
  • Will keep certain high frequencies
  • Can we take the middle path?

5
Pipeline
6
Sampling
  • Input models
  • Polygonal
  • Range scanned
  • Volume
  • Mathematical function
  • Sample the 3-d space
  • Get a 3-d signal (volume as we know in sci-viz)

7
Controlled(?) Filtering
  • Use low-pass filtering to create
    frequency-limited versions of the sampled volume
  • To create a series of LOD volumes, use a series
    of low-pass filters.
  • End result volume pyramid
  • Can easily get interpolated volumes for
    in-between levels

8
Compare this volume hierarchy with -
  • Locality based clustering of geometry
  • Mip-maps (which use averaging, also a kind of
    low-pass filtering)

9
Pipeline
10
Reconstruction
  • Volume reconstruction
  • Surface reconstruction
  • Iso-surface
  • Too many faces
  • Can output high frequency patterns
  • Need to preserve topology of the iso-surface
    while simplifying
  • Method of choice improved Splitting Box Algo.

11
Splitting Box
  • Vertices above/equal iso-value black
  • Vertices below iso-value white
  • Edge is MC if
  • Color changes at most once along the sequence of
    its vertices
  • Face is MC if all 4 edges are MC
  • Box is MC if all 6 faces are MC

12
Splitting Box
  • Start with the whole grid as a single box
  • Recursive division of the box along its longest
    edge until
  • 2x2x2 box
  • MC box B with satisfactory approximation
  • What is a satisfactory approximation?

13
Satisfactory approximation
  • Sub-boxes of B, B1 and B2, are MC
  • Approximation triangles intersect the common face
    of B1 and B2 in between grid points where there
    is a color change
  • Triangle chain of B geometrically coincident with
    that of B1 and B2 (?)
  • Thus, approximate surface cannot be more than one
    sampling distance in error

14
Adaptive Marching Cubes
  • Edge is AMC if
  • Color changes at most once along the sequence of
    its vertices
  • Face is AMC if all 4 edges are AMC
  • Box is AMC if all possible grid lines in the
    encompassed 3d space are AMC

15
How is AMC different from Splitting Box?
  • Splitting Box can have unnecessary splits
  • E.g., non-MC box can get split along an MC edge,
    thus sub-boxes would still be non-MC
  • Non-AMC splits are done by finding the axes which
    have non-AMC edges
  • Different approximation checking
  • Within a user-specified distance from the true
    (MC) surface

16
Multi-layered surface rendering
  • Render multiple surfaces at different opacities
  • Visibility order determination a problem

17
Results
  • Slides
Write a Comment
User Comments (0)
About PowerShow.com