Mesh Decimation. Mark Pauly

Transcription

1 Mesh Decimation Mark Pauly

2 Applications Oversampled 3D scan data ~150k triangles ~80k triangles Mark Pauly - ETH Zurich 280

3 Applications Overtessellation: E.g. iso-surface extraction Mark Pauly - ETH Zurich 281

4 Applications Multi-resolution hierarchies for efficient geometry processing level-of-detail (LOD) rendering Mark Pauly - ETH Zurich 282

5 Applications Adaptation to hardware capabilities Mark Pauly - ETH Zurich 283

6 Size-Quality Tradeoff error size Mark Pauly - ETH Zurich 284

7 Outline Applications Problem Statement Mesh Decimation Methods Vertex Clustering Iterative Decimation Extensions Remeshing Variational Shape Approximation Mark Pauly - ETH Zurich 285

8 Problem Statement Given: M = (V, F) Find: M = (V, F ) such that 1. V = n < V and M M is minimal, or 2. M M < ɛ and V is minimal M M Mark Pauly - ETH Zurich 286

9 Problem Statement Given: M = (V, F) Find: M = (V, F ) such that 1. V = n < V and M M is minimal, or 2. M M < ɛ and V is minimal hard! look for sub-optimal solution Mark Pauly - ETH Zurich 287

10 Problem Statement Given: M = (V, F) Find: M = (V, F ) such that 1. V = n < V and M M is minimal, or 2. M M < ɛ and V is minimal Respect additional fairness criteria normal deviation, triangle shape, scalar attributes, etc. Mark Pauly - ETH Zurich 288

11 Outline Applications Problem Statement Mesh Decimation Methods Vertex Clustering Iterative Decimation Extensions Mark Pauly - ETH Zurich 289

12 Vertex Clustering Cluster Generation Computing a representative Mesh generation Topology changes Mark Pauly - ETH Zurich 290

13 Vertex Clustering Cluster Generation Uniform 3D grid Map vertices to cluster cells Computing a representative Mesh generation Topology changes Mark Pauly - ETH Zurich 291

14 Vertex Clustering Cluster Generation Hierarchical approach Top-down or bottom-up Computing a representative Mesh generation Topology changes Mark Pauly - ETH Zurich 292

15 Vertex Clustering Cluster Generation Computing a representative Average/median vertex position Error quadrics Mesh generation Topology changes Mark Pauly - ETH Zurich 293

16 Computing a Representative Average vertex position Low-pass filter Mark Pauly - ETH Zurich 294

17 Computing a Representative Median vertex position Sub-sampling Mark Pauly - ETH Zurich 295

18 Computing a Representative Error quadrics Mark Pauly - ETH Zurich 296

19 Error Quadrics Squared distance to plane p = (x, y, z, 1) T, q = (a, b, c, d) T dist(q, p) 2 = (q T p) 2 = p T (qq T )p =: p T Q q p Q q = a 2 ab ac ad ab b 2 bc bd ac bc b 2 cd ad bd cd d 2 Mark Pauly - ETH Zurich 297

20 Error Quadrics Sum distances to vertex planes dist(q i, p) 2 = i p T Q qi p = p T ( i Q qi ) p =: p T Q p p i Point that minimizes the error q 11 q 12 q 13 q 14 q 21 q 22 q 23 q 24 q 31 q 32 q 33 q p = Mark Pauly - ETH Zurich 298

21 Comparison average median error quadric Mark Pauly - ETH Zurich 299

22 Vertex Clustering Cluster Generation Computing a representative Mesh generation Clusters p {p 0,...,p n }, q {q 0,...,q m } Connect (p,q) if there was an edge (p i,q j ) Topology changes Mark Pauly - ETH Zurich 300

23 Vertex Clustering Cluster Generation Computing a representative Mesh generation Topology changes If different sheets pass through one cell Not manifold Mark Pauly - ETH Zurich 301

24 Outline Applications Problem Statement Mesh Decimation Methods Vertex Clustering Iterative Decimation Extensions Mark Pauly - ETH Zurich 302

25 Incremental Decimation General Setup Decimation operators Error metrics Fairness criteria Topology changes Mark Pauly - ETH Zurich 303

26 General Setup Repeat: pick mesh region apply decimation operator Until no further reduction possible Mark Pauly - ETH Zurich 304

27 Greedy Optimization For each region evaluate quality after decimation enqeue(quality, region) Repeat: pick best mesh region apply decimation operator update queue Until no further reduction possible Mark Pauly - ETH Zurich 305

28 Global Error Control For each region evaluate quality after decimation enqeue(quality, region) Repeat: pick best mesh region if error < ε apply decimation operator update queue Until no further reduction possible Mark Pauly - ETH Zurich 306

29 Incremental Decimation General Setup Decimation operators Error metrics Fairness criteria Topology changes Mark Pauly - ETH Zurich 307

30 Decimation Operators What is a "region"? What are the DOF for re-triangulation? Classification Topology-changing vs. topology-preserving Subsampling vs. filtering Inverse operation progressive meshes Mark Pauly - ETH Zurich 308

31 Vertex Removal Select a vertex to be eliminated Mark Pauly - ETH Zurich 309

32 Vertex Removal Select all triangles sharing this vertex Mark Pauly - ETH Zurich 310

33 Vertex Removal Remove the selected triangles, creating the hole Mark Pauly - ETH Zurich 311

34 Vertex Removal Fill the hole with triangles Mark Pauly - ETH Zurich 312

35 Decimation Operators Vertex Removal Vertex Insertion Remove vertex Re-triangulate hole Combinatorial DOFs Sub-sampling Mark Pauly - ETH Zurich 313

36 Decimation Operators Edge Collapse Vertex Split Merge two adjacent triangles Define new vertex position Continuous DOF Filtering Mark Pauly - ETH Zurich 314

37 Decimation Operators Half-Edge Collapse Restricted Vertex Split Collapse edge into one end point Special vertex removal Special edge collapse No DOFs One operator per half-edge Sub-sampling! Mark Pauly - ETH Zurich 315

38 Edge Collapse Mark Pauly - ETH Zurich 316

39 Edge Collapse Mark Pauly - ETH Zurich 317

40 Edge Collapse Mark Pauly - ETH Zurich 318

41 Edge Collapse Mark Pauly - ETH Zurich 319

42 Edge Collapse Mark Pauly - ETH Zurich 320

43 Edge Collapse Mark Pauly - ETH Zurich 321

44 Edge Collapse Mark Pauly - ETH Zurich 322

45 Edge Collapse Mark Pauly - ETH Zurich 323

46 Edge Collapse Mark Pauly - ETH Zurich 324

47 Edge Collapse Mark Pauly - ETH Zurich 325

48 Priority Queue Updating Mark Pauly - ETH Zurich 326

49 Incremental Decimation General Setup Decimation operators Error metrics Fairness criteria Topology changes Mark Pauly - ETH Zurich 327

50 Local Error Metrics Local distance to mesh [Schroeder et al. 92] Compute average plane No comparison to original geometry Mark Pauly - ETH Zurich 328

51 Global Error Metrics Simplification envelopes [Cohen et al. 96] Compute (non-intersecting) offset surfaces Simplification guarantees to stay within bounds Mark Pauly - ETH Zurich 329

52 Global Error Metrics (Two-sided) Hausdorff distance: Maximum distance between two shapes In general d(a,b) d(b,a) Computationally involved A d(a,b) B d(b,a) Mark Pauly - ETH Zurich 330

53 Global Error Metrics Scan data: One-sided Hausdorff distance sufficient From original vertices to current surface Mark Pauly - ETH Zurich 331

54 Global Error Metrics Error quadrics [Garland, Heckbert 97] Squared distance to planes at vertex No bound on true error Q 1 Q 2 Q 3 = Q 1 +Q 2 v 3 p 1 p 2 p i TQ i p i = 0, i={1,2} solve v 3 TQ 3 v 3 = min < ε? ok Mark Pauly - ETH Zurich 332

55 Complexity N = number of vertices Priority queue for half-edges 6 N * log ( 6 N ) Error control Local O(1) global O(N) Local O(N) global O(N 2 ) Mark Pauly - ETH Zurich 333

56 Incremental Decimation General Setup Decimation operators Error metrics Fairness criteria Topology changes Mark Pauly - ETH Zurich 334

57 Fairness Criteria Rate quality of decimation operation Approximation error Triangle shape Dihedral angles Valence balance Color differences... Mark Pauly - ETH Zurich 335

58 Fairness Criteria Rate quality after decimation Approximation error Triangle shape Dihedral angles Valence balance Color differences r 1 e 1 < r 2 e 2 r 1 e 1... e 2 r 2 Mark Pauly - ETH Zurich 336

59 Fairness Criteria Rate quality after decimation Approximation error Triangle shape Dihedral angles Valence balance Color differences... Mark Pauly - ETH Zurich 337

60 Fairness Criteria Rate quality after decimation Approximation error Triangle shape Dihedral angles Valence balance Color differences... Mark Pauly - ETH Zurich 338

61 Fairness Criteria Rate quality after decimation Approximation error Triangle shape Dihedral angles Valence balance Color differences... Mark Pauly - ETH Zurich 339

62 Fairness Criteria Rate quality after decimation Approximation error Triangle shape Dihedral angles Valence balance Color differences... Mark Pauly - ETH Zurich 340

63 Fairness Criteria Rate quality after decimation Approximation error Triangle shape Dihedral angles Valence balance Color differences... Mark Pauly - ETH Zurich 341

64 Fairness Criteria Rate quality after decimation Approximation error Triangle shape Dihedral angles Valance balance Color differences... Mark Pauly - ETH Zurich 342

65 Incremental Decimation General Setup Decimation operators Error metrics Fairness criteria Topology changes Mark Pauly - ETH Zurich 343

66 Topology Changes? Merge vertices across non-edges Changes mesh topology Need spatial neighborhood information Generates non-manifold meshes Vertex Contraction Vertex Separation Mark Pauly - ETH Zurich 344

67 Topology Changes? Merge vertices across non-edges Changes mesh topology Need spatial neighborhood information Generates non-manifold meshes manifold non-manifold Mark Pauly - ETH Zurich 345

68 Comparison Vertex clustering fast, but difficult to control simplified mesh topology changes, non-manifold meshes global error bound, but often not close to optimum Iterative decimation with quadric error metrics good trade-off between mesh quality and speed explicit control over mesh topology restricting normal deviation improves mesh quality Mark Pauly - ETH Zurich 346

69 Outline Applications Problem Statement Mesh Decimation Methods Vertex Clustering Iterative Decimation Extensions Mark Pauly - ETH Zurich 347

70 Out-of-core Decimation Handle very large data sets that do not fit into main memory Key: Avoid random access to mesh data structure during simplification Examples Garland, Shaffer: A Multiphase Approach to Efficient Surface Simplification, IEEE Visualization 2002 Wu, Kobbelt: A Stream Algorithm for the Decimation of Massive Meshes, Graphics Interface 2003 Mark Pauly - ETH Zurich 348

71 Multiphase Simplification 1. Phase: Out-of-core clustering - compute accumulated error quadrics and vertex representative for each cell of uniform voxel grid 2. Phase: In-core iterative simplification - compute fundamental quadrics - iteratively contract edge of smallest cost Mark Pauly - ETH Zurich 349

72 Multiphase Simplification 1. Phase: Out-of-core clustering - compute accumulated error quadrics and vertex representative for each cell of uniform voxel grid 2. Phase: In-core iterative simplification - compute fundamental quadrics - use accumulated quadrics from clustering phase - iteratively contract edge of smallest cost achieves a coupling between the two phases Mark Pauly - ETH Zurich 350

73 Multiphase Simplification Original (8 million triangles) Uniform clustering (1157 triangles) Multiphase (1000 triangles) Garland, Shaffer: A Multiphase Approach to Efficient Surface Simplification, IEEE Visualization 2002 Mark Pauly - ETH Zurich 351

74 Multiphase Simplification Garland, Shaffer: A Multiphase Approach to Efficient Surface Simplification, IEEE Visualization 2002 Mark Pauly - ETH Zurich 352

75 Out-of-core Decimation Streaming approach based on edge collapse operations using QEM Pre-sorted input stream allows fixed-sized active working set independent of input and output model complexity Wu, Kobbelt: A Stream Algorithm for the Decimation of Massive Meshes, Graphics Interface 2003 Mark Pauly - ETH Zurich 353

76 Out-of-core Decimation Randomized multiple choice optimization avoids global heap data structure Special treatment for boundaries required Wu, Kobbelt: A Stream Algorithm for the Decimation of Massive Meshes, Graphics Interface 2003 Mark Pauly - ETH Zurich 354