pmr: point to mesh rendering, a feature-based approach tamal k. dey and james hudson...
Post on 18-Jan-2016
221 Views
Preview:
TRANSCRIPT
PMR: Point to Mesh Rendering,PMR: Point to Mesh Rendering,A Feature-Based ApproachA Feature-Based Approach
Tamal K. Dey and James Hudson
{tamaldey,jhudson}@cis.ohio-state.eduhttp://www.cis.ohio-state.edu/~tamaldey
October 30, 2002
Department of Computer and Information ScienceThe Ohio State University
Department of Computer and Information Science
OverviewOverview
IntroductionAlgorithm: Two Stages
Preprocessing Viewing
ResultsConclusion
Department of Computer and Information Science
Hierarchy ConstructionHierarchy Construction
Points and triangles displayedFeature-dependent, not screen-spaceAdvantageous for large, flat areasGoals:
Quality at all viewing distances Adaptive display Adjustable speed vs. quality setting No input mesh needed
Department of Computer and Information Science
MotivationMotivationTriangles good for quality; points for speedDifference is subtle when far away
Point-based splatting
PMR
Zoom in on the nose...
Department of Computer and Information Science
Splatting vs. PMRSplatting vs. PMR
Point Splatting PMR
When zoomed-in, differences are more noticeable, especially at upper left edge of nose.
Department of Computer and Information Science
Our ApproachOur Approach
Utilizes hierarchy Contains Points and Triangles
Hierarchy: scale independent Depends on model's features
No surface mesh required More flexibility Simpler data structure
Department of Computer and Information Science
PreprocessingPreprocessingDecimate input where "redundant" points exist
Use features to determine thisThreshold guides levels of hierarchyNo new points added; only removal
= 0.2 = 0.3 = 0.4
Department of Computer and Information Science
Feature DetectionFeature DetectionWe use Voronoi diagram to detect features
Can be costly: time + memory Solution: Use octree decomposition of space Maximum of 12000 points per node useful
Department of Computer and Information Science
Feature DetectionFeature DetectionDense point set: long, skinny Voronoi cellsCapture this via height and radius values
Pole vector = estimated normal (AB98) Height estimates distance to medial axis Radius estimates distance between neighbors
Department of Computer and Information Science
Feature DetectionFeature Detection
Decimation is based on ratio
Remove all points with ratio < (threshold) Point with small ratio must have close neighbors Repeat for several values of to give hierarchy
We use values from 0.1 to 1.0
Each leaf node N is processed individually
radiusheight
Department of Computer and Information Science
Point HierarchyPoint HierarchyThe final point hierarchy contains
progressively fewer points
= 0.2 = 0.3 = 0.4
Department of Computer and Information Science
Triangle HierarchyTriangle HierarchyFor point p: We define umbrella of p
Umbrella = set of triangles incident on p and are dual to Voronoi edges intersecting tangent polygon
Department of Computer and Information Science
Triangle HierarchyTriangle Hierarchy
Result: progressively sparser triangle sets
= 0.2 = 0.3 = 0.4
Department of Computer and Information Science
Disk FileDisk FileFor each leaf, store to disk:
Points, estimated normals, hierarchy levelsUmbrella triangles per vertexUmbrella radii per vertexAverage umbrella radius for all points
Map file to memory when viewing
Department of Computer and Information Science
ViewingViewingMust determine pixel size
● Done once per leaf node only● Closest corner point = the one to use● Project two world space points to screen● Gives ratio of world space to screen space● Conservative estimate
Department of Computer and Information Science
Choice of HierarchyChoice of HierarchyChoice of hierarchy level made once per leaf
Metric: Use average umbrella size Try to match umbrella size to pixel size
If too dense: more points to processIf too sparse: detail lost
User can trade speed for quality via scale factor
Just right Too sparseToo dense
Department of Computer and Information Science
Pixel vs. UmbrellaPixel vs. Umbrella
For each point: choose: Pixel vs. Umbrella Compare umbrella radius to (pixel size) (scale factor)
scale factor allows trade-off of quality vs. speed Choose umbrella only if size too big; else choose pixel Conservative estimation performed
Can draw as pixel Must draw as triangles
Department of Computer and Information Science
Scale FactorScale FactorScale factor allows modification of calculation
If scale factor larger, calculations treat pixels as larger
Selects sparser hierarchy level Can modify scale factor to selectively slow
transition between levels, especially at high levels of decimation
Department of Computer and Information Science
Scale Factor TransitionScale Factor Transition
Need to slow transition between sparser levelsDifferences invisible when far away
=0.1 =0.3 =0.8 =1.0
Department of Computer and Information Science
ResultsResults
System used: Pentium 4, 1.7 Ghz, 2 GB RAM Matrox Millenium G450 graphics card Software-only OpenGL rendering
Department of Computer and Information Science
ResultsResults
We varied from 0.1 to 1.0, steps of 0.1 If is large (1.0), features are lost Varying the scale factor
If pixel size is 2 world space units, begin altering Reduce factor linearly until pixel is 4 world space units If pixel is 4 or more units: factor is equal to 1 Net effect: as decimation becomes sparser, slow the
transition between levels.
Department of Computer and Information Science
ResultsResults
0.11 FPS, 4.5M tris, 0 points (Full detail)
0.77 FPS, 670K tris,48K points (PMR)
0.65 FPS, 650K tris, 377K points (PMR)
1.65 FPS, 215K tris, 204K points (PMR)
Varying distances; Blue=triangles, Red=points
Department of Computer and Information Science
ResultsResultsDense level Sparse level
Department of Computer and Information Science
ResultsResults
Comparison of full-detail (=0) vs PMR
Full detail, 0.54 FPS PMR, 3.85 FPS
Department of Computer and Information Science
ResultsResultsComparison of full-detail vs PMR
Full detail, 0.25 FPS PMR, 0.71 FPS
Department of Computer and Information Science
ResultsResultsComparison of full-detail vs PMR
Full detail, 0.11 FPS PMR, 0.61 FPS
Department of Computer and Information Science
ResultsResults
Dense level Sparse level
Factor=10.16 FPS
Factor=30.74 FPS
Factor=50.96 FPS
Department of Computer and Information Science
ResultsResults
DragonHappyBlade
DavidHeadStMatthew
437645542557882954
20006463382855
0.540.430.250.11
0.072
3.853.440.710.780.67
Object Vertices Full PMR
Frames per second. Full denotes the full(=0) mesh; PMR denotes the adaptive
hierarchy scheme with a factor of 5.
Department of Computer and Information Science
Preprocessing TimesPreprocessing Times
DragonHappyBlade
DavidHeadStMatthew
437645542557882954
20006463382855
03:5304:3609:0206:3626:38
97150238512479
Object Vertices Time Size (MB)
Note: Times are in Hours:Minutes.
Department of Computer and Information Science
ConclusionsConclusions
A hybrid rendering scheme Points and triangles employed User-adjustable error tolerance No input surface required
Future work Applications to volume rendering
top related