3D Visual Computing


Faculty members:

Dr. Sanjeev Kapoor

Prof. K.K. Biswas

Dr. Prem Kalra

Student members:

Alok Kumar

Aditi Sharma

Balpreet Singh

Vivek Kwatra

Projects:

The Graphics team at Indian Institute of Technology, Delhi, has been working towards development of algorithms for viewing of 3-dimensional objects in  a dynamic environment. The research fields include
1. Hidden Surface Removal Algorithms
2. Use of Binary Space Partition trees for representation of 3-D worlds
3. Algorithms to deal with curves and surfaces
4. Shadowing
The projects undertaken in the department are as follows:
Dynamic World Visualisation -
Team members : - Alok Kumar, Balpreet Singh, Vivek Kwatra

1.Dynamic Binary Space Partitioning for Hidden Surface Removal

Description : -

This project aimed at developing a new technique for dynamic binary space partitioning with applications to the dynamic hidden surface removal problem of polygons. The dynamic hidden surface removal problem is to determine and maintain visible surfaces in 3-D set of polygons from a given viewpoint under changes to the scene and it allows for insertions and deletions of polygons  in a 3-D scene.
        In our approach, the polygons are maintained in a space partitioning structure called the separating plane partition tree which is a variation of the traditional binary space partition (BSP) tree.
In this structure, separating planes are the fundamental objects which are used to partition the space. The polygons are fragmented with each of the fragments being stored at the leaves. This allows  for  easy access of polygon fragments for efficient insertion and deletion. Efficient hidden surface removal of the polygons is then achieved by painting the fragments in an order obtained from the space partitioning tree.

The algorithm developed in this project was published in ICVGIP'98 conference -
Dynamic Binary Space Partitioning for Hidden Surface Removal

2.Using Separating Planes between Objects for Efficient Hidden Surface Removal

This project aimed at a new technique for generation of binary space partition trees representing polyhedral objects and
outlines an efficient algorithm for operations such as composition,
deletion and motion of objects. We assume a modified version of BSP trees with all the polygon fragments lying at the leaf nodes described in the above project.
        The efficiency of merging algorithm for two BSP trees depends on the separability of the polyhedras represented by them. We achieve efficiency by use of an ellipsoidal cover of polyhedral object to determine the separating plane between two object. In this project, the primary aim was moving from the domain of polygons to that of objects in dynamic environment.

The algorithm developed in this part of the project will be published in ICVC'99 conference:-

Using Separating Planes between Objects for Efficient Hidden Surface Removal

3.Development of a library for defining objects and motion primitives

     A library of C++ functions, which could be readily used for describing a 3-D view as a set of objects as well as motion of each object, was developed. The library provides for classes of some common shapes (cube etc.) along with functions to perform translation, rotation etc. Complex inputs can be generated by instantiating some of the basic shapes and composing them. Use of C++ functionality allows inheritance of all the motion functions from the instantiated classes. Complex motions can also be described by composition of basic motions.
 

       4. Efficient Algorithm for incorporating Curved Surfaces

Team members - Alok Kumar, Vivek Kwatra

Description  : - The objective of the project is to encorporate curved surfaces in the partition trees. This will help in reducing complexity of problems of hidden surface removal, of objects including curved surfaces. The curved surfaces which were previously approximated by polygons, increases complexity of the algorithm, so efforts have been to encorparate curves in partition trees, without approximating them with polygons.

       5.Dymanic Shadowing

Team members - Aditi Sharma, Balpreet Singh

Description : -The project aims at adding shadows in dynamic scenarios. A new algorithm based on conical partitioning of space is being used to improve on current approaches for shadow generation. The task will also include looking at existing algorithms and compare them.
 

Software Developed :-
Static World Visualisation -
Team members: - Alok Kumar,Balpreet Singh,Vivek Kwatra

Description : - The project aimed at developing an X-windows based software for on-line input and viewing of 3-dimensional objects from a moving view point. The software allowed user to input a view as a set of polygons using a X-windows GUI. User can then go on to view the static input for moving viewpoint.