David Luebke University of Virginia …and other cool graphics stuff.

David LuebkeUniversity of Virginia

…and other cool graphics stuff

Overview

I’ll talk about a bunch of current projects:– Scanning Monticello– Immersive display– Level of detail– GPGPU– Interactive ray tracing– Graphics architectures

Scanning Monticello

Fairly new technology: scanning the world

Scanning Monticello

Want a flagship project to showcase this

Idea: scan Thomas Jefferson’s Monticello– Historic preservation– Virtual tours– Archeological and architectural research,

documentation, and dissemination– Great driving problem for scanning &

rendering research

Scanning Monticello—Demo

Scanning Monticello:Where To Next?

Building structured light scanner– Idea:

Shine bars of light at object Discern its shape from their shape

– Jeff Peirson is building a portable structured-light rig for scanning objects, friezework, and so on at Monticello

Scanning Monticello:Where To Next?

Capture high dynamic range color to go with scanned geometry– Show HDR, tone mapping examples– Idea:

Scan w/ laser, then take lots of pictures Lots of pictures, different exposures, different

locations (maybe even video?) Figure out camera location very accurately For all “correct” pixels in each image:

Project onto geometry Store for that location and direction

– Might work with Rhythm & Hues on this

Immersive Display:New Orleans Museum of Art

Jefferson’s America & Napoleon’s France – Major exhibition commemorating

bicentennial of the Louisiana Purchase Included a “Virtual Monticello” exhibit

– Virtual window– Barrier stereogram

Show NOMA videos

Immersive Display:Here at UVA

We have the pieces of a cool virtual-reality immersive display system – Screens (3 silver, 2 rear-projection)– Projectors (6 dedicated, 3 others)– Head tracker – Spatial audio– Even part of the NOMA “window”!

Need to bring all the pieces together– Got a good start already

Immersive Display:Research

Design and conduct experiments with perception lab– Study the perceptual/cognitive effects of

added layers of realism– Almost guaranteed to produce papers

Also fits into other research efforts– Perceptual level of detail/animation– Perceptual interactive ray tracing– Chromium

Level of Detail

A recurring theme in computer graphics: trade fidelity for performance– Reduce level of detail of distant, small,

or unimportant objects

249,924 polys 62,480 polys 7,809 polys 975 polys

Level of Detail

A recurring theme in computer graphics: trade fidelity for performance– Reduce level of detail of distant, small,

or unimportant objects

Level of Detail:Motivation

Big models! – David: 56 million polygons

– St. Matthew: 372 million polygons

Cou

rtes

y D

igita

l Mic

hela

ngel

o P

roje

ct

Level of Detail Research:VDSlib

Comparatively recent technique: continuous level of detail– Adapt LOD continuously, not discrete

chunks Better fidelity/triangle Hard to make fast on today’s hardware

– VDSlib: efficient view-dependent LOD Cool data structures & algorithms Could make a big difference in 3D game

design Still work to be done to make fast enough

Level of Detail Research:GLOD

GLOD: – High-level library for LOD in OpenGL

Goal: make high-quality LOD easy to integrate in games and other graphics applications

Includes and interfaces with VDSlib Cool software (interface design) issues Free membership in the Church of Glöd

– Show GLOD video

GPGPU

General-purpose computation on graphics hardware (GPUs)

Modern GPUs are fantastically capable– Extremely programmable– Full floating-point precision– Fast, so fast (20 GFLOPS sustained on GFX

5900) We want to harness that power for

scientific computing– But it’s harder than it sounds

GPGPU:CA Simulations on GPU

Promising domain: cellular automata (CA) simulations– Ex: Conway’s “Game of life”

Application: modeling cardiac tissue fibrillation – Collab with Don Jordan, MAE, and med school

personnel Possible application: simulate

microvascular remodeling– Collab with Tom Skalak, BME

GPGPU: Other domains

Fluid mechanics simulation – Flow across a flapping wing– Modeling viscoelastic fluids– Joint work with Hossein Haj-Hariri, MAE– Cool math (PDEs)

Heat transfer– Simulating heat propagating across chip– Joint work with Kevin Skadron, Mircea Stan

Others…

The Big Question

How should we evaluate and regulate the visual fidelity of our simplifications?

Measuring Fidelity

Fidelity of a simplification to the original model is often measured geometrically:

METRO by Visual Computing Group, CNR-Pisa

Measuring Visual Fidelity

However…– The most important measure of fidelity is

usually not geometric but perceptual: does the simplification look like the original?

Therefore:– We are developing a principled framework for

LOD, based on perceptual measures of visual fidelity

Perceptual LOD

Idea: measure local simplification measures against a perceptual model to predict whether the user can could see the effect of simplification

Model: contrast sensitivity function

Perception 101: Contrast Sensitivity Function

Contrast grating tests produce a contrast sensitivity function– Threshold contrast

vs. spatial frequency– CSF predicts the

minimum detectablestatic stimuli

Campbell-Robson Chart by Izumi Ohzawa

Your Personal CSF

Perceptual Graphics:Where To Next?

Incorporate eccentricity, velocity (attention?)

Protect copyrighted media: imperceptible “watermarking” via mesh distortion

Interactive ray tracing…

Interactive Ray Tracing

Interactive ray tracing – Once a joke, now a reality– Interesting opportunity: prioritized pixel

sampling for perceptually driven rendering Sample edges, center of gaze, etc more Requires reconstruction of sparse samples! Cool perceptual, signal processing issues

System: OpenRT– Would like to install on our 24-node cluster– Cool systems hacking issues

Graphics Architecture

We are exploring research topics in computer graphics architecture– Thermal-aware graphics

Joint work with Kevin Skadron

– Building a simulator to explore these and other issues

My Projects:

Proposed projects:– Scanning Monticello– Immersive display– Level of detail– GPGPU– Perceptual graphics– Interactive ray tracing– Graphics architecture

Summary

Proposed projects:– Scanning Monticello

Structured light HDR capture

– Immersive display– Level of detail– GPGPU– Perceptual graphics– Interactive ray

tracing– Graphics

architecture

Summary

Proposed projects:– Scanning Monticello– Immersive display

System building Perception expts

– Level of detail– GPGPU– Perceptual graphics– Interactive ray

tracing– Graphics

architecture

Summary

Proposed projects:– Scanning Monticello– Immersive display– Level of detail

VDSlib GLOD

– GPGPU– Perceptual graphics– Interactive ray

tracing– Graphics

architecture

Summary

Proposed projects:– Scanning Monticello– Immersive display– Level of detail– GPGPU

CA sims Fluid/heat sims

– Perceptual graphics– Interactive ray

tracing– Graphics

architecture

Summary

Proposed projects:– Scanning Monticello– Immersive display– Level of detail– GPGPU– Perceptual graphics

LOD/ray tracing Watermarking Integrate w/ display

– Interactive ray tracing– Graphics architecture

Summary

Proposed projects:– Scanning Monticello– Immersive display– Level of detail– GPGPU– Perceptual graphics– Interactive ray

tracing Perceptual

heuristics Sampling/recon

– Graphics architecture

Summary

Proposed projects:– Scanning Monticello– Immersive display– Level of detail– GPGPU– Perceptual graphics– Interactive ray

tracing– Graphics

architecture

Summary

Proposed projects:– Scanning Monticello– Immersive display– Level of detail– GPGPU– Perceptual graphics– Interactive ray

tracing– Graphics

architecture

Desired student attributes:– Willing to commit 10-15

hours per week– Self-starter– Good discipline & time

management skills– Here over summer– Good programmer– Intro graphics (or equiv)– Hardware capable– Strong math skills– Intro perception

knowledge