G - O - V : Teaching

[an error occurred while processing this directive]

[an error occurred while processing this directive]

Summer Semester 2004

Seminar

Graphics meets Vision

real-world photograph (left) and synthetic image (right) (from Samuel Boivin's project web page)

Computer Graphics rendering is all about creating realistic-looking images from a description, or model, of a scene. Computer Vision, on the other hand, is the art of inferring such abstract scene descriptions from real world-recorded images. In this seminar we are going to investigate how image synthesis can aid in image analysis.

With your expertise in image synthesis techniques from CG-I, you will re-create a real-world photograph by computer graphics means: You are asked to take a picture of some real-world object, consider the effects that are responsible for the object's appearance, build a model of the depicted scene, and render it using ray tracing or graphics hardware. By varying the model's parameter values, you will be able to match the synthetic image to the photograph as closely as possible, this way analyzing the illumination, surface reflectance, 3D structure, or motion in the image.

• Thursdays 14-16, beginning at 2:00pm sharp
• Building: MPII, Geb. 46
• Room: 019 - Visualization Theater

• Attend the first seminar lectures to learn about analysis-by-synthesis, photometric calibration and about scientific soft skills.
• Stroll through nature and take one (or several) pictures of an interesting scene you want to model. Present your picture in class, explain the visible effects and the techniques you intend to use to model the scene.
• Build a computer model of the scene you've recorded either for your ray tracer or for graphics hardware rendering.
• Find model parameter values so that the rendered image appears as similar to your photograph as possible. You might want to employ some optimization algorithm to find the parameter values automatically (analysis-by-synthesis).
• Prepare a compelling 15-minute Powerpoint presentation showing your results and describing what you have done.
• Along with your oral presentation, prepare a web page explaining your project.
• Attend the other students' presentations to learn about their work, give them feed-back on their presentations, and learn for your own talk's preparations.

Topic	Students	Project Page	Presentations
Tea Cup	Benjamin Peters		PDF
Kenny	Jens Weber, Andreas Kleer		PDF
Wrinkled Sheet of Paper	Tobias Schild, Markus Wilhelm
Ludo	Patrick Kapahnke, Dennis Schwarz		PDF
Bricks	Christian Weber, Denis Weber		PDF
Coffee Box	Martin Koster		PDF

Inverse Rendering/Analysis by Synthesis
• Samuel Boivin's homepage
  • Siggraph 2002 paper
  • Presentation Slides, Siggraph 2001
• Polarization-based Inverse Rendering
• Model-based Camera Calibration Using Analysis by Synthesis
• Model-based Analysis of Multi-Video Data

Realistic Ray Tracing

• Peter Shirley, Realistic Ray Tracing, AK Peters, 2000
• Henrik Wann Jensen, Realistic Image Synthesis Using Photon Mapping, AK Peters, 2001
• Philip Dutre, Kavita Bala, Philippe Bekaert, Advanced Global Illumination, AK Peters, 2002

Realistic Rendering on Graphics Hardware

• Andrew Woo, et al., OpenGL Programming Guide, 3. Ausgabe, Addison-Wesley, 1999
• Randima Fernando and Mark Kilgard, The Cg Tutorial, Addison Wesley, 2003
• Tomas Moeller and Eric Haines, Real-Time Rendering, AK Peters, 2000


Remarks

• The seminar starts April 22th.
• The seminar will be held in German or English, depending on the students' preferences.

---

Home / Research Units / G-O-V: Home Page / Teaching

---

Max-Planck-Institut für Informatik
About the Institute | Research Units | News & Activities | Location | People | Services | Search the Site | Intranet

---

Copyright © 1998-2003 by Max-Planck-Institut für Informatik. All rights reserved. Impressum and legal notices.
Page maintained by Andrei Linţu <lintu@mpi-sb.mpg.de>
www site design and concept by Uwe Brahm