| Paul Ostrowski's Home Page |
|
| Paul Ostrowski's Home Page |
I created this page with two goals in mind. The first goal was to provide a link to my resume, both in HTML format, and as a word document. The second goal was to provide a place to showcase some of my cooler outputs from school projects or whatever. I hope you enjoy them.
Last year I worked with Scott Dillman on an awesome COM project that let you create COM object using base classes that I designed, and itegrate them in a visual environment Scott developed. It is really cool and very powerful. Check it out at: http://www.eas.asu.edu/~cse591f/project/project.htm
I've been working (very slowly) on a new Role-playing game. I don't have a cool name for it yet, but I think of it as the next generation of role-playing games. It is very technical, perhaps too technical for the general game-player, but it is a very acurate representation of real hand-to-hand combat. Check it out - Next Generation Role-Playing game
If you have any comments, suggestions, or just want to say 'Hi', click
here to 
Send me email
I worked with Scott Dillman & Shaung Lui to create a cool 3-D robot arm animation program, using either OpenGL or native API calls (depending upon what you have installed on your machine). To check it out, download Robot.zip and extract to somewhere on your hard-drive. Then execute the Robot.exe file. When it starts, you will want to open the file newsamp.bot from the File/Open menu. This program was written on WindowsNT, and I think also runs on Windows 9x.
In the past, I have studied a lot about wavelets and fractals. For those of you interested in wavelets, what they are and how they are used, I have found an excellent source for wavelet information: Mathsoft.com. I have been developing my own algorithm that I use for terrain generation. The basic algorithm is self-recursive, thus fractal in nature, but it incorporates all (OK, maybe its just most - I'm still working on the math) properties of wavelets. In this way, I can generate a partial landscape with exactly as much detail as would be visible. As the user moves or zooms in on a local section, I expand the terrain as needed, but I ensure that the higher resolution image is best approximated by the previous lower resolution image! I have not implemented it in 3-D yet, but it is easily adaptable to any dimention, and can work for vector fields, or any other complex data structures. If fractals and wavelets interest you, click on it for more information, including a windows executable demo.
During a CSE340 course at ASU, I developed an algorithm for text retrieval, or any type of linear search. The algorithm executes with about 1/8th to 1/10th the number of computation and comparisons of the standard Knuth-Morris-Pratt (KMP) search algorithm. Unfortunately, I just found out that I have re-invented the wheel. A very similar algorithm already exists, its called the Boyer-Morris algorithm :(
I have also spent a good amount of time playing with compression routines, both lossy and lossless. In particular, the LZW, Huffman, Arithmatic and Discrete Cosine Transform (used only for graphic images). At GM Desert Proving Grounds, I attained an average 12.5:1 lossless compression for the real-time data aquisition pipeline used in the test vehicles.
I have also finished my first project in VRML 2.0. Its a simple war game in which you aim a cannon to fire a shell at a truck that moves along a road before you. It's not perfect, but its interesting. You will need to install a VRML plug in for your browser if you don't already have one. I developed the script using Microsofts VRML plug-in for internet explorer. If you use a different browser, I can't gaurentee the quality, since I've heard VRML script work differntly on different browsers.
Below I have included some of the work I have completed from my computer graphics classes:
This is the output from my volume visualization lab from the advanced computer graphics class. It was created by casting rays through a density map data set taken from a CAT scan of a dried human skull. The data set was 64 X 64 X 68, and the image created is 300 X 300. On my pentium 60 MHz, it took about 30 minutes to complete, running under windows 3.1
This is the output from my vector-field-flow visualization lab. It represents the airflow over a car. The program allows the user to create a curvelinear grid, impose a 2D vector field onto the grid vertices, and then isolate all the critical points in the map. A critical point is point at which there is NO airflow. A saddle point is a critical point that has two airflows coming into it, and two exiting it. These points also have two real eigenvalues, one positive, one negative, of the jacobian matrix. By tracing the vectors, in both positive and negative time, along the eigenvectors of the jacobian matrix, we can isolate local flow fields. This helps us analyze the topology of the flow-field. The grey triangles show areas that contain a critical point. The red lines are the vector traces of the eigenvectors from each and every saddle point.
If this stuff interests you, I have 4 other screen dumps from this project:
I lost the source code to this project when my hard drive crashed a few years ago, but I do have the windows executable. Like I said, I don't have the source code anymore, so if there are bugs in it, oh well, sorry.
