Projects
Elif Tosun
New York University



Shape Optimization using Reflection Lines
E.Tosun, Y.Gingold, J.Reisman, D.Zorin

Abstract
Many common objects have highly reflective metallic or painted finishes. Their appearance is primarily defined by the distortion the curved shape of the surface introduces in the reflections of surrounding objects.
Reflection lines are commonly used for surface interrogation, as they capture many essential aspects of reflection distortion directly, and clearly show surface imperfections that may be hard to see with
conventional lighting.
In this paper, we propose the use of functionals based on reflection lines for mesh optimization and editing.  We describe a simple and efficient discretization of  such functionals based on screen-space surface
parameterization, and we demonstrate how such discrete functionals can be used for several types of surface editing operations.









In Proceedings, Symposium on Geometry Processing 2007:193-202. Barcelona, Spain.

Paper

Supplementary video
Addendum tech report
Presentation at SGP07

Thesis Proposal

Recently I proposed the topic of my dissertation.  Briefly, my proposal is to find functionals to minimize on a surface such that the resulting surface would satisfy some visual quality constraints. The functionals are derived from methods in the field of surface interrogation. Surface interrogation refers to a step in surface construction where a surface is tested by means of methods based on geometry before claimed visually fair and can be manufactured, for example in car hood manufacture. A few examples of such tools are reflection lines, curvature lines, isophotes, etc.  Our goal is to eliminate this step completely such that when the surface is generated it will automatically satisfy such constraints.  Our work is mostly discrete, where the functionals are applied to surfaces defined as meshes. In order to test  our approaches we use the meshopt package still in development at NYU.

Related Links:
Proposal 
Defense  
Bibliography


Manifold Based Surface Construction
E.Tosun, D.Zorin

I worked on the extension of the manifold-based surfaces of Ying and Zorin [YZ04] to surfaces that have piecewise smooth boundaries (including convex and concave corners). Additionally, we modified the method to construct surfaces with prescribed smoothness using tensor product B-splines as local descriptions of the surface per chart. This construction  also supports surfaces with piecewise smooth boundaries. Furthermore, we showed  that these manifold-based constructions yield surfaces that are at least 2-flexible everywhere.  
  


I have presented a portion of this work as part of my depth qualifying exam which is equivalent to getting a degree in MSc.


Related Links:
Paper(in preparation)
DQE Presentation

[YZ04] L. Ying,D. Zorin "A Simple Manifold Based Construction fo Surfaces of Arbirary Smoothness". SIGGRAPH 04.


Simulation Project
As a final project in my physically-based simulation class, I implemented an algorithm of my college advisor Ileana Streinu.  Although the emphasis of the class was on dynamic simulation, my project was an implementation of a kinetic simulation. The algorithm is for convexifying polygons based on a concept called pseudo triangulations. The details of the algorithm can be found at Ileana Streinu's website. My implementation involved glui as well as opengl and glut for user controls and petsc for non-linear solvers. More details and code can be provided upon request.

Related Links:
Project Presentation


DetectIt
E.Tosun, B.Wellington

I worked on a rather unrelated project as the final for my Operating Systems class. With a colleague (Ben Wellington) we developed a peer to peer, scalable plagiarism detection software. Built on top of the tapestry framework, it uses a fingerprinting technique in order to detect "plagiarized" documents. Effectiveness of this system increases with number of users, since there will be more documents in the system to check against.  Our implementation was in Java. 

Related Links:
Project Presentation
Paper


© Elif Tosun 2007