1. 186.162 Seminar (mit Bachelorarbeit) 186.175 Seminar aus Computergraphik SS 2011 Organizers: Matthias Bernhard, Thomas Auzinger Teaching staff: M.Wimmer, O.Mattausch, S.Jeschke, R.Habel, C.Scheiblauer, M. Knecht, M.Ilcik, P.Guerrero, M. Arikan, R. Preiner, T. Sturn Institute of Computer Graphics and Algorithms Vienna University of Technology

2. 1 Seminar Mission statement: Get an idea how scientific work is carried out (in CG) Practice to review literature and get familiar with a particular scientific topic Selecting, reading and understanding Summarizing and explaining (orally and written) Comparing and discussing Practice to give a talk Experience critical scientific discussion

3. 2 Seminar Tasks: Submit initial literature list 3 Lectures: „Wie halte ich einen Vortrag“ (Prof. Purgathofer), „Forschung und wie sie funktioniert“ (Prof. Gröller), „How to write a scientific paper“ (Prof. Wimmer) Write draft of your thesis Review a draft of a colleague Give a presentation Active discussion participation Revision of final document

4. 3 Talk Use Institute’s PowerPoint template for presentations 20 + 5 minutes talk In English Active discussion participation is mandatory and will be graded Reviewer has to begin the discussion

5. 4 Thesis Seminar Thesis (186.175) : STAR 8-12 pages per student Bachelor Thesis (186.162): book chapter 20- 30 pages per student In English Format reports in the style of a scientific paper Use LaTeX Students can work in groups of two

6. Scientific discussion Review of a paper: Draft of thesis one week before presentation Everyone gets a draft to review Typical review form of a conference (Eurographics) Help author to improve the manuscript Reviewer begins discussion after the talk Revision Improve final submission according to reviewers comments 5

7. 6 Important Dates Attend 3 lectures : Fri 11.3. 14:15-15:15, “Forschung und wie sie funktioniert” Fri 15.4. 14:15-15:45, “How to write a scientific paper” Mon 16.5. 14:00-15:30, “Wie halte ich einen Vortrag” Fri 15.4., Deadline: list of literature Thu 2.6., Deadline: draft of your paper Fri 3.6., Receive paper for review Wed 8.6., Deadline: review Thu 9.6. 9:00-…, Presentations Fri 30.9., Deadline: final paper

8. TOPICS Please prepare to take some notes... 7

9. GPGPU – Parallel Programming on Graphics Hardware A survey of parallel programming languages and libraries CUDA, OpenCL, FORTRAN, Compute Shader Thrust, BSGP, BLAS Numerical Applications 8

10. External Sorting of Point Data Laser scanners sample their environment with up to 1 billion points per scan Algorithms work more efficient on sorted data Data does not fit completely in the main memory of a computer, therefore... External Sorting 9 53124765312476 12345671234567 5 6 7 1 2 3 4

11. Multi-threaded Data Structures In recent years the largest enhancements in processing speed are due to parallelization of well known algorithms. Research papers on algorithms that can be used to build and manipulate data structures like the octree or the kd-tree in a parallelized way. 10

12. Image-based 3D Reconstruction Give an overview on current reconstruction methods like Structure from motion, SLAM, Mircrosofts Kinect and PointGreys stereo vision products 11

13. Point clouds from stereo vision / 3d scanner deliver a point cloud of the augmented scene but without any normal vectors Give an overview to the state of the art algorithms for real- time normal estimation on point clouds. 12 RT normal estimation for point clouds

14. Architectural Reconstruction from Sparse Point Clouds manual generation of building models of urban environments are highly costly automatic algorithms to reconstruct geometry from point clouds 13

15. Inverse Rendering recover reflectance properties of surfaces in a real scene from a set of photographs, taking into account both direct and indirect illumination 14 one of several input photographsestimated reflectance properties

16. Post Processing Effects Research and implement real-time post- processing effects like motion blur, depth of field, cell shading, edge detection, bloom... 15

17. HDR-tonemapping Map high-dynamic range colors to a limited range (e.g. for a LCD-display) Tone-mapping operator Computed from image properties (e.g.brightness) Goal: perceptual match between a real scene and a displayed image Thesis: give overview and discuss perceptional issues 16 6 different exposures to generate HDR image Tonemapped image

18. Give an overview on adding camera artifacts to augmented reality scenes to make virtual objects look more realistic. Also available as PR or DA! 17 Camera Artifacts for AR

19. Simulating Depth-of-field in 3D graphics DoF Lens can only focus one point -> limited range where details are perceived in full sharpness In Computer Graphics: Has to be simulated Increase perceived realism Guide attention Thesis: discuss optics, simulation techniques and applications 18

20. Non-Photorealistic Rendering Non-Photorealistic Rendering is inspired by artistic styles such as painting, drawing, technical illustration and animated cartoons Research all kinds of Non- Photorealistic Rendering techniques 19

21. Similarity Detection for Shape Understanding Methods for shape understanding often detect similar parts in a shape to extract its structure Given this structure 'smarter' operations can be applied to the shape 20

22. The Heat Kernel as Shape Descriptor The way heat propagates in a shape tells a lot about that shape's geometry The heat kernel provides a compact representation of the heat propagation in a shape 21

23. Geometry Synthesis of Surface Details by Exemple Given an example model with particular geometric surface details, how can one apply the same kind of details to another surface? 22

24. Semantic Object Deformations A survey of advanced deformations methods 23

25. 24 2D Animation Should be intuitive, direct, user friendly Interesting questions As rigid-as-possible manipulation Intuitive handling of depth 2D animation is fun!

26. Automated Architectural Floor Plans A survey of floor planing algorithms Planar graphs, Optimization, Geometry 25

27. Topics on Image Processing Choose your interest from a large range of topics (resizing, merging, denoising, dehazing, color adjustments, recompositioning, …). Give an (coherent) overview of the state of the art of a single or various techniques. 26

28. 27 Image Resizing What is the best way to resize an image? Very hot topic in recent years Good: Seam carving for context-sensitive resizing Even better: Combine different operators in an optimal way

29. Representing Textures in Gabor Space Gabor Transform for texture representation in the context of texture synthesis for rendering 28

30. Ocean and Wave Dynamics Oceans and waves can be simulated with different approaches Procedural, Numerical, Empirical 29

31. 30 Seminar Questions? http://www.cg.tuwien.ac.at/courses/Seminar/index.html