Object Space EWA Surface Splatting: An Equipment Quickened Way to deal with Top notch Point Rendering.


64 views
Uploaded on:
Description
CMU Object Space EWA Surface Splatting: An Equipment Quickened Way to deal with Astounding Point Rendering Liu Ren Hanspeter Pfister Matthias Zwicker Inspiration Point-based representation needs amazing composition separating Fantastic point rendering needs equipment bolster
Transcripts
Slide 1

CMU Object Space EWA Surface Splatting: A Hardware Accelerated Approach to High Quality Point Rendering Liu Ren Hanspeter Pfister Matthias Zwicker

Slide 2

Motivation Point-based illustrations needs brilliant composition sifting High quality point rendering needs equipment bolster GPU execution outpaces CPU [Wolfman Geforce 4 demo] Pure Hardware Accelerated + High Quality? Yes!

Slide 3

Related Work Object Space EWA Splatting EG 02

Slide 4

Surface Element (Surfel) ordinary y surfel digression plane z x No availability No composition maps, no typical maps, and so forth 2D remaking channel Point-based Surface Representation

Slide 5

twist screen space article space surfel reproduction channel y x associating distorted recreation channel y z x Surfel Rendering: Splatting

Slide 6

twist screen space item space distorted recreation channel screen space resampling channel low-pass channel Surfel Rendering: Screen Space EWA Filtering reproduction channel Elliptical Weighted Average (EWA) sifting EWA Splat = low-pass channel distorted recreation channel Screen space EWA splatting not bolstered by illustrations equipment.

Slide 7

twist screen space article space distorted low-pass channel recreation channel digression space resampling channel low-pass channel Surfel Rendering: Object Space EWA Filtering Tangent space resampling channel = twisted low-pass channel remaking channel View ward channel

Slide 8

typical surfel digression plane Textured polygons Tangent space resampling channels Texture mapping EWA splats in edge cradle Additive alpha mixing Warped surface composition reproduction Hardware Accelerated Point-based Rendering

Slide 9

Correct perceivability in equipment ? No gaps, concealed surface splats evacuation Lack of A-cradle support ? EWA resampling channel View subordinate Texture or polygon not altered Challenges

Slide 10

1. Perceivability splatting Disable casing cradle upgrades Render dark quad for each surfel Generate profundity picture with a little balance 2. Resampling channel splatting Disable Z-cradle overhauls Render textured polygons with added substance alpha mixing Two Pass Algorithm Overview

Slide 11

surface impediment ancient rarities profundity picture profundity picture QSplat z camera space camera space Object Space EWA First Pass: Visibility Splatting Schemes

Slide 12

curved gaussian (digression space resampling channel) composition (unit gaussian) (0,1) (1,1) match (0,0) (1,0) surface mapping surfel polygon with obscure geometry digression space quad with known geometry vertex calculation textured quad Second Pass: Handle View Dependent EWA Resampling channel

Slide 13

Per-pixel standardization: Read back information from edge cushion Post-preparing plan Bad for equipment speeding up Per-surfel standardization: Pre-figure the surfel standardization weight Pre-handling plan Good for equipment quickening without standardization with standardization no antiquities shifting brilliance Normalization Issues

Slide 14

Splatting No Filtering V.S. Item Space EWA Filtering Object Space EWA/Points V.S. Anisotropic Texture Filtering + Accuview/Triangle Mesh Demo: Checkerboard on Geforce 4 Ti 4600 :

Slide 15

Salamander 103K Surfels Demo: Surfel Models on ATI Radeon 9700

Slide 16

Demo: Surfel Models on ATI Radeon 9700 Chameleon 102K Surfels

Slide 17

Demo: Surfel Models on ATI Radeon 9700 Wasp 273K Surfels

Slide 18

Demo: Surfel Models on ATI Radeon 9700 Fiesta 352K Surfels

Slide 19

Performance with Phong Shading

Slide 20

Conclusion New question space definition of EWA surface splatting Completely equipment quickened methodology with unimportant CPU contribution Benefits from GPU execution changes

Slide 21

Future Work Semitransparent point models View subordinate BRDF shading Animated point models Optimization with forthcoming equipment highlights

Slide 22

CMU Acknowledgments Jessica Hodgins, Paul Heckbert Micheal Doggett, Evan Hart, Jeff Royle Henry Moreton Jennifer Pfister, Wei Li, Wei Chen Http://www.cs.cmu.edu/~liuren/research.htm

Recommended
View more...