Dan Orban

Cancer Cell Migration

This project worked towards understanding which properties of a cell and its environment are involved in cell movement. Cancer cell migration is a problem that is both high dimensional (i.e. has many variables like stiffness, motors, clutches, traction forces, actin flow patterns, etc...) and has complex spatial temporal relationships (i.e. cell body, actin, and arms).

Using the c++ version of the Cell Migration Simulator, developed in collaboration with the interactive visualization lab and David Odde's biomedical engineering lab, we investigated how to explore parameter relationships. The parameter space is large and nonlinear with complex spatial features. See my Doctoral Dissertation for results.

Shock Physics

In collaboration with the Data Science at Scale team at Los Alamos National Lab, I worked for two summers developing interactive visualization techniques for understanding the high-dimensional space in shock physics applications.

Shock physics is a research area in material science that uses high powered gas guns or lasers to shock a material of interest. The resulting elastic and fluid reactions provide detailed information about the strength of materials, giving insight into physics and the possibility of creating new material phases (e.g. stronger materials).

Cardiac Lead Design

In order to understand the complex input and output space for cardiac lead design, we implemented an instance comparison tool called Bento Box. This tool allows users to interactively explore and compare spatially relevant features. Bento Box is a virtual reality visualization and 3D user interface technique for comparative analysis of data ensembles, such as this set of 10 time-varying simulations of blood flow around a cardiac lead in the right atrium of the heart. Each column shows a simulation with different parameters, here varying the length and stiffness of the lead. Each row shows a different view of the data. The top row is a zoomed-out overview. Users add additional rows of complementary, zoomed-in views during analysis for memory intensive fluid-structure interaction simulations in virtual reality.

Particle Flow

Visualizing complex fluid flow in virtual reality (VR) is a necessary part of state-of-the-art climate study, physical system design, and biomechanical engineering. However, the most effective way to support real-time advection and rendering of hundreds of thousands of multi-dimensional particles on multiple high-framerate stereo displays remains an open challenge. Particle Flow is particle advection and rendering system for addressing this challenge in the context of current graphics hardware.

Sandbox Graphics Engine

In order to interact with visualize large simulation data sets with state of the art in graphics technology I created the Sandbox Graphics Engine. It focuses on high performance graphics rendering across multiple shared contexts and devices. The Sandbox is an entity component framework written in C++ using Vulkan and OpenGL.

The ultimate goal of this project is to enable rendering on multiple immersive display types (Desktop / Web / Cave / Head mounted display) while efficiently minimizing data transfer and flexible interaction.

Supercomputer Interaction

One important rule in visualizing large scientific data sets is "Do not move the data." The data is often too large to efficiently move for real-time interactive graphics. At the same time most remote data access is handled over the web. This project integrates the high-performance of c++ back end (e.g. off-screen rendering using Vulkan) where the data is stored with a javascript front end. The application uses web sockets to communicate over the network to the super computer also enabling simulation steering. The example uses P3.js as a simple graphics library for interaction.

Cloth Simulation

The purpose of my cloth simulation was to compare the features of multiple simulation integration schemes. In this solution I implemented and investigated explicit Euler, simi-implicit Euler, explicit Runga-Kutta 4th Order (RK4), and implicit backwards Euler. implicit backwards Euler was definitely the most stable. Surprisingly, simi-implicit Euler, in terms of stability, performed almost as well as RK4.

As a feature, I also explored interaction techniques in the HTC Vive using MinVR, an open source C++ VR library that I helped develop. Using a virtual sphere as a cursor, it was easy to manipulate the cloth through sphere collisions.

Finite Element Simulation

For a physics based animation class, I imlemented a CPU Finite Element Method (FEM) simulator in C++. It explores using explicit and implicit integration methods. A GPU version has been started.

Ray Tracer

For a graduate level graphics class, I implemented a ray tracer that handles lighting, reflection, refraction, and depth of field. I am very proud of this picture of a deer overlooking Lake Tahoe.

Haptics

For exploring user interaction, I have used the Open Haptics library to interact with a Phantom device. This allows users to feel forces. For example, users can pull the edge of a cell arm to manipulate a simulation.

To get experience creating our own haptics device, a friend, David Swanson, and I created a virtual button (pictured on the left). Using the HTC Vive, and our active haptics device we were able to virutally touch and push buttons. Moving a motorized plate while our hand moved (location based), provided the feeling of force feedback.