1.  Marc Levoy Graphics research and courses at Stanford http://graphics.stanford.edu

2.  Marc Levoy Graphics faculty Chris Bregler animation, motion capture Ron Fedkiw simulation, natural phenomena Pat Hanrahan rendering, architectures Marc Levoy sensing, modeling, rendering Leo Guibas modeling, geometry

3.  Marc Levoy Related areas Bernd Girod (EE) imaging, video, networking Terry Winograd human-computer interaction Bill Dally computer architecture Mark Horowitz VLSI, hardware

4.  Marc Levoy Research projects Digital Michelangelo project Solving the Forma Urbis Romae Visualizing cuneiform tablets Modeling subsurface scattering Multi-image digital photography Measuring and modeling reflectance Acquisition and display of light fields Image-based modeling and rendering Real-time volume rendering Interactive workspaces Parallel graphics architectures Stanford immersive television project Texture analysis-synthesis methods Motion analysis / synthesis Automatic illustration systems Physics-based modeling and simulation Visualization of computer systems Real-time programmable shading …and many more

5.  Marc Levoy Digital Michelangelo project (Levoy) very large geometric models scientific tool for art historians virtual museums, multimedia, replicas lasting archive of important cultural artifacts

7.  Marc Levoy David’s left eye

8.  Marc Levoy Research challenges vision problems –aligning and merging scans –automatic hole filling –inverse color rendering –automated view planning digital archiving problems –making the data last forever –robust 3D digital watermarking –indexing and searching 3D data –real-time viewing on low-cost PCs

9.  Marc Levoy The Forma Urbis Romae (Levoy) 60’ x 45’ x 4” marble map of ancient Rome, carved 200 A.D. shows the city at a scale where you can see every room now in 1,163 fragments, an open problem for 500 years

10.  Marc Levoy Solving the puzzle algorithms must be fast minimize false positives robust to effects of weathering linear features 2D contours 3D surfaces

11.  Marc Levoy Real-Time Programmable Shading (Hanrahan) high-level languages for programmable graphics hardware RenderMan in real-time guide the future of graphics hardware parallelizing scientific computations on the same hardware

12.  Marc Levoy Modeling subsurface scattering (Hanrahan, Levoy) translucency is caused by multiple scattering approximated by volumetric diffusion validation using physical measurements

13.  Marc Levoy Stanford Immersive Television Project (Bregler, Dally, Girod, Hanrahan, Horowitz, Levoy) light field cameras real-time range scanning sensing vision compression transmission decompression graphics Intel DTV tuner card

14.  Marc Levoy Light field cameras (Horowitz, Levoy, Hanrahan) video light field camera spherical light field camera

15.  Marc Levoy Real-time range scanning space time camera projector light stripe object one pixel over time one frame

16.  Marc Levoy holes can be found and filled on-the-fly object or scanner can be handheld / shoulderheld video framerange data merged model (159 frames)

17.  Marc Levoy Motion analysis / synthesis (Bregler) ? Acquisition Analysis Animation Kinematics Dynamics Language

18.  Marc Levoy Physics-based modeling and simulation (Fedkiw) new computational algorithms for numerical simulation of physical phenomena Water - simulated using the Navier Stokes equations and the level set method for implicit surface evolution. A solid “invisible” sphere initiates the splashing.

19.  Marc Levoy Physics-based modeling and simulation (Fedkiw) new computational algorithms for numerical simulation of physical phenomena Smoke - simulated as a scalar in a flow field generated using the Navier Stokes equations. Photon mapping is used for the visualization.

20.  Marc Levoy Virtual Human(oid) Project (Fedkiw) derive and improve physics- based models of tissues, organs, organ systems, clothing

21.  Marc Levoy Kinetic Data Structures (Guibas) A kinetic data structure (KDS) maintains an attribute of interest in a collection of moving or deforming objects. Examples include many kinds of proximity, visibility, or connectivity information. This yields efficient algorithms for collision detection, visibility maintenance, and aggregation or communication among mobile nodes.

22.  Marc Levoy Interactive workspaces (iRoom) (Winograd, Fox, Hanrahan) multiple display surfaces multiple interaction devices flexible display architecture facilitates group work The ultra-high resolution Interactive Mural integrates desktop access, sketching, 3D models, and images under pen-based control

23.  Marc Levoy Courses CS 148 – Introductory Computer Graphics CS 248 – Introduction to Computer Graphics CS 348A – Mathematical Foundations (modeling) CS 348B – Image Synthesis Techniques (rendering) CS 348C – Animation Techniques CS 338 – Level Set Methods CS 368 – Geometric algorithms (computational geometry) CS 448 – Topics in Computer Graphics CS 468 – Topics in Geometric Algorithms

24.  Marc Levoy Examples of topics CS 448 - Topics in Computer Graphics –experiments in digital television –interactive workplaces –modeling appearance This year: –graphics architectures (Autumn, Hanrahan) –digital photography (Spring, Levoy) CS 468 - Topics in Geometric Algorithms –matching techniques and similarity measures

25.  Marc Levoy PhD students Maneesh Agrawala Sean Anderson Robert Bosch Ian Buck Cindy Chen Milton Chen Scott Cohen Joao Comba James Davis Matthew Eldridge Reid Gershbein Francois Guimbretiere Olaf Hall-Holt David Hoffman Greg Humphreys Homan Igehy Brad Johanson Menelaos Karavelas Dave Koller Song Sam Liang Tamara Munzner Bradley Nelson John Owens Lucas Pereira Matt Pharr Kekoa Proudfoot Katheline Pullen Timothy Purcell Ravi Ramamoorthi Szymon Rusinkiewicz Gordon Stoll Chris Stolte Diane Tang Yelena Vileshina Li-Yi Wei http://graphics.stanford.edu