Multidimensional Adaptive Sampling and Reconstruction for Ray - PowerPoint PPT Presentation

Multidimensional Adaptive Sampling and Reconstruction for Ray Tracing Toshiya Hachisuka * Wojciech Jarosz * Richard Peter Weistroffer † Kevin Dale ‡ Greg Humphreys † Matthias Zwicker * Henrik Wann Jensen * * University of California, San Diego † University of Virginia ‡ Harvard University

4 Sampling and Realistic Rendering • Distributed ray tracing [Cook et al. 84] Pixel Lens Objects Image Plane

8 Adaptive Sampling Image based adaptive sampling • Recursive subdivision [Whitted 80] • Stochastic adaptive sampling [Mitchell 87, 91] • Perceptually based criterion [Bolin and Meyer 98] • Divergence based criterion [Rigau et al. 03] • MISER [Press and Farrar 90] [Leeson 03]

9 Adaptive Sampling Image based adaptive sampling • Recursive subdivision [Whitted 80] • Stochastic adaptive sampling [Mitchell 87, 91] • Perceptually based criterion [Bolin and Meyer 98] • Divergence based criterion [Rigau et al. 03] • MISER [Press and Farrar 90] [Leeson 03]

10 Motion Blur

11 Motion Blur

12 Motion Blur

13 Motion Blur Time

13 Motion Blur Time Pixel

17 Adaptive Sampling [Mitchell 91] Time Pixel

36 Adaptive Sampling [Mitchell 91] Time Pixel Sampled Exact

37 Motion Blurred Sphere 1 Samples / Pixel 4 16 64 Reference

38 Motion Blurred Sphere 1 Samples / Pixel 4 16 64 Reference

39 Adaptive Sampling [Mitchell 91] Time Pixel Sampled Exact

40 Key Idea of Our Work • Conventional adaptive sampling • Only adaptive on pixel axis • Multidimensional adaptive sampling • Adaptive on all axes

41 Multidimensional Adaptive Sampling Time Pixel

62 Multidimensional Adaptive Sampling Time Pixel Sampled Exact

63 Motion Blurred Sphere - Comparison [Mitchell 91] Our method 1 4 16 64 Reference

64 Motion Blurred Sphere - Comparison [Mitchell 91] Our method 1 4 16 64 Reference

65 Motion Blurred Sphere Our Method 20 Low Discrepancy 15 Mitchell 10 5 0 1 4 16 64 256

66 Rendering is Integration • Each pixel value, L(x, y), is defined as multidimensional integration � � L ( x, y ) = f ( x, y, u 1 , . . . , u n ) d u 1 . . . d u n · · ·

67 Rendering is Integration • Motion blur is 1D integration over time (t) � L ( x, y ) = f ( x, y, t ) d t

68 Rendering is Integration • DOF is 2D integration over lens (u, v) �� L ( x, y ) = f ( x, y, u, v ) d u d v

69 Analytical Solution Time t Pixel L ( x ) x � L ( x ) = f ( x, t )d t

70 Analytical Solution Time t Pixel L ( x ) x � L ( x ) ≈ w i f ( x, t i ) i

71 Conventional Adaptive Sampling Time t Pixel L ( x ) x � L ( x ) ≈ w i f ( x, t i ) i

72 Multidimensional Adaptive Sampling Time t Pixel L ( x ) x � L ( x ) ≈ w i f ( x, t i ) i

73 Key Features • Exploits coherency beyond the image plane • Adaptive sampling based on deterministic samples • Theory applies to various effects

Results

75 Motion Blur (3D) - Reference 512 samples 27,488 sec

76 Motion Blur (3D) - Our Method 8 samples 672.2 sec

77 Motion Blur (3D) - [Mitchell 91] 12.67 samples 676.4 sec

78 Motion Blur (3D) - Comparison Our method [Mitchell 91] 8 samples / pixel 12.67 samples / pixel 672.2 sec 676.4 sec RMS: 0.0034 RMS: 0.0099

79 Depth of Field (4D) - Reference ‏ 512 samples 11,960 sec

80 Depth of Field (4D) - Our Method 16 samples 993 sec

81 Depth of Field (4D) - [Mitchell 91] ‏ 38.25 samples 980 sec

82 Depth of Field (4D) - Comparison Our method [Mitchell 91] 16 samples / pixel 38.25 samples / pixel 993 sec 980 sec RMS: 0.132 RMS: 0.192

Sampling Density 83

84 Area Light Source (4D) Our method Mitchell Our Method Mitchell Scene Setup 1 8 Area Light Source Samples / Pixel Occluders 64 64

85 Area Light Source (4D) Our method Mitchell Metropolis Our Method Mitchell Low Discrepancy Metropolis 1 8 Samples / Pixel 64 64

86 Area Light Source (4D) Our method Mitchell Metropolis Samples / Pixel 64 Area Light Source Occluders

87 Motion Blur + DOF (5D)

88 Future Work • More higher dimensional adaptive sampling • New adaptive sampling criterion • Application to more complex light transport

89 Conclusion • Produces images with less noise by fewer samples • Adaptive sampling in multidimensional space • Reconstruction from multidimensional samples • Applicable to various rendering effects

90 Acknowledgement • NSF CPA 0701992 • UCSD FW Grid Project • NSF Research Infrastructure Grant EIA-0303622 • All UCSD graphics lab members

91 Thanks

Multidimensional Adaptive Sampling and Reconstruction for Ray - PowerPoint PPT Presentation

Multidimensional Adaptive Sampling and Reconstruction for Ray Tracing Toshiya Hachisuka * Wojciech Jarosz * Richard Peter Weistroffer Kevin Dale Greg Humphreys Matthias Zwicker * Henrik Wann Jensen * * University of California, San

Recent Advances in Adaptive Sampling and Reconstruction for Monte

SURE-based Optimization for Adaptive Sampling and Reconstruction Supplementary Materials Tzu-Mao

Adaptive Incident Radiance Field Sampling and Reconstruction Using Deep Reinforcement Learning

1 Sampling and Aliasing Image Processing pipeline Artifacts due to undersampling or poor

A new multidimensional-type reconstruction and limiting procedure for unstructured (cell-centered)

Sampling and Reconstruction Digital Image Processing How does it help? Filtering reduces

Multidimensional Transcoding for Adaptive Video Streaming Jens Brandt, Lars Wolf Institut f

Overview of Sampling Topics (Shannon) sampling theorem Impulse-train sampling

Multi-Armed Bandits: Non-adaptive and Adaptive Sampling Instructor: Sham Kakade 1 The

AdaGeo: Adaptive Geometric Learning for Optimization and Sampling Gabriele Abbati 1 , Alessandra

Adaptive Sampling for Sensor Networks Seminarvortrag von Philippe Bourquin Betreuung: Silvia

Estimating and Sampling Graphs with Multidimensional Random Walks Group 2: Mingyan Zhao,

Application of optimal sampling lattices on CT image reconstruction and segmentation or three

Adaptive Quiz Generation Using Thompson Sampling Fuhua (Oscar) Lin, PhD Athabasca University,

3D RECONSTRUCTION Reconstruction method Reconstruction from images Reconstruction from video

Realistic Image Synthesis - Spatio-temporal Sampling and Reconstruction. Exploiting Temporal

Adaptive Sampling for Imaging Malena Sabate Landman, Sergey Dolgov, Silvia Gazzola, and Tom Davis

Importance Sampling Methodology for Multidimensional Heavy-tailed Random Walks Jose Blanchet

Estimation of cosmological parameters using adaptive importance sampling Gersende FORT LTCI,

Reconstruction of post-adaptive images of Io observed with LBTI and KECK M. Bertero. P. Boccacci

Fast Convergence to Wardrop Equilibria by Adaptive Sampling Methods . . . . . Gouleakis

ADAPTIVE TWO-STAGE INTEGRATORS FOR SAMPLING ALGORITHMS BASED ON HAMILTONIAN DYNAMICS E.

SociableSense: Exploring the Trade- offs of Adaptive Sampling and Computation Offloading for

Convergence and Efficiency of Adaptive Importance Sampling techniques with partial biasing