Top Banner
Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen
10

Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Dec 17, 2015

Download

Documents

Albert Byrd
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projectionof Convex Polyhedra

Stefan RoettgerThomas Ertl

University of Erlangen

Page 2: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

2

Introduction

• Unstructured volume rendering

• Cell-projection = PT algorithm of Shirley and Tuchman (’90)

• Main bottlenecks: sorting and tetrahedral decomposition (Wittenbrink ‘99)

• Current performance 600,000 tet/s (Guthe ‘02)

• Above 1,000,000 tet/s performance is completely memory bandwidth limited

Page 3: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

3

• Use emissive optical model (Max ‘95)

• Does not require sorting

• Ray integral = length of ray segment times average emission (assuming linear interpolation)

Emissive Optical Model

Page 4: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

4

• The graphics hardware can take over the projection of arbitrary convex polyhedra

• Based on bounded layered fog (Mech JGT ‘01)

Projection of Convex Polyhedra

Page 5: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

5

• 1st pass– enable A writing and front face culling

– draw primitive with alpha=(d-min d)/d

• 2nd pass– enable subtractive blending

and back face culling

– draw primitive again

• 3rd pass– disable A writing and culling

– enable additive blending with alpha multiplied

– draw primitive with rgb=emission/2

PCP Algorithm

Page 6: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

6

• Emissive model: 212,000 hex/s on NVIDIA GeForce3 (Guthe ‘02: 120,000 hex/s)

• For maximum intensity projection one only needs to render each face once

• Performance is about 600,000 hex/s• Performance drop is mainly due to additional

passes

Performance

Page 7: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

7

Example Images

Bluntfin8 fps

Campfire50 fps

Neghip22 fps

Page 8: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

8

• Triangulated terrain

• Place prism on every base triangle

• Assume constant emission in each prism

Ground Fog

Page 9: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

9

Emission vs. MIP

25 fps 50 fps

Page 10: Cell-Projection of Convex Polyhedra Stefan Roettger Thomas Ertl University of Erlangen.

Cell-Projection of Convex Polyhedra Volume Graphics ‘03 Stefan Roettger, University of Erlangen Tokyo

10

Thanks for your attention!

Fin