Simulation projects:
2D Collision Detection STANFORD CS 348C
"Spaghetti & Meatball" 2D collision detection and resolution. Achieved by applying Penalty Forces when edges get too close and impulses between point-edge and point-point pairs to ensure no overlaps occur. Detection is sped up by using a uniform grid subdivision such that only edges that fall in the same subdivision check one another for collision.
Inextensibility Constraints & 2D Collision Detection STANFORD CS 348C
"Spaghetti" 2D collision detection and resolution with inextensibility constraints. Spring forces previously used to join particles effectively removed, with edge length being preserved by constraint impulses.
​
Left: Long unconstrained strand. Maximum edge length error ~60% at each time step after collision.
Right: Long constrained strand. Maximum edge length error ~2% on initial collision and ~0.000001% otherwise.
2D Inverse Kinematics STANFORD CS 248
2D IK on an "Arm" with no joint constraints. End location specified by mouse click.
"Boids" bird flocking simulation STANFORD CS 248
​
Boids are directed via forces: cohesion (bring towards center of flock), separation (penalty force for getting too close repels so no collision), and alignment (boids flying in same direction have same velocity). In the bounded box case a penalty force is applied the farther a boid gets from the box bounds, eventually turning it around. This contains the boids within a defined area.
​
Below: Boids in a bounded box.
Below: Boids in a looped box.
More to be added soon ... (middle of December) STANFORD CS 348C
Voronoi Diagram Procedural Fracture - using Fortune's algorithm & Jump Flood Algorithm:
​
​
​
Project write-up can be found here: