computer-animation-simulation

Lab 3: SPH Fluid Simulation

Methods

Fully Compressible Method

• Simplest version
• Water compress too much
• Speed of sound for pressure
• Spiky Kernel shows better results than Poly6
• The most simple model -> does not shows the best results
• Less parameters to modify -> more try and error to get good results

Weakly Compressible Method

• Eurographics 2019 approach
• State Equation for pressure
• Added boundary particles
• Tried to use Cubic Spline kernel, but failed

Iterative Weakly Compressible Method

• Iterate until convergence or timeout
• Results are very similar to Weakly Compressible Method, but with a lower stiffness

Kernel Equations

Poly6

The most easy to understand, not recommended to use for pressure nor viscosity though.

Spiky

Easy to understand, nice results when applied

Viscosity Gradient kernel

Recommended for viscosity instead of Poly6 nor Spiky

CubicSpline

The kernel does not give me plausible results, and probably I misscalculated the gradient derivative, so I end up not using it

Pressure Equation

Speed of sound

• Used for the Fully Compressible Method
• Not an intuitive parameter to modify

State Equation

• Set stiffness as parameter
• Used for the Weakly Compressible Methods
• More intuitive to understand and modify

Boundary Particles

• Used in the Weakly Compressible Methods.
• Helps with the collission in extreme situations and avoid crashes.
• With the uncorrect parameters and situation, it can act as a glue for the particles, sticking them in the cube walls.

Lab 2: Cloth Simulation

Collisions

Particle-collider DONE

Sphere Collision

Box Collision

Prevent self-intersections

Use Particles & Particle Hash DONE

Adjusted the Particle Spatial Hash from Lab1

Consider Rest Distance and avoid Jittering DONE

Set a minimum distance between the Rest Distance and collision distance

Use Substepping, not CCD DONE

Set substepping=10 for the integrator step and collision detection

Stable Cloth-Cloth Friction DONE

Integrator Used

Better results with Verlet, but other integrators work as well.

Interactions

Grab and move particles DONE

Taking advantage of the spatial hash, perform a ray-particle collision detection.

Fix/Release particles DONE

Blackhole DONE

Boat Scene :o

Verlet Integrator is not working well for the fountain, so had to try others like RK2 to avoid using custom integrators for cloths and other for fountain particles.

Issues

Self-intersection not looking properly

Even adding the recommended steps to achieve self-intersection, it still looks flat, and by increasing the thickness, the cloth starts to become unstable.

Cloth Simulation exploded easily

Starting with Symplectic Euler, increasing the kElastic to +400.0 helps with stability, then implementing relaxation steps, and finally moving to Verlet integrator which is Physically Based Dyanmics (more stable!).

Sphere not colliding when moved

Had to adjust a bit the formula for Verlet by updating the prevPos with the velocity added because of the collision.

---

Lab 1: Particle Systems

I have chosen the Midpoint Integrator since it has more appealing results when activating the blackhole gravitational attraction force with the lowest computation (RK2 and RK4 also give appealing results).

Particle systems with 1000s of particles DONE

Tested with 2000 particles smoothly in release mode

Collisions: test & response DONE

• Plane
• AABB
• Sphere

Gravity + another force DONE

• Gravity
• Linear Drag Force
• Quadratic Drag Force
• Gravitational attraction (Blackhole)

Mouse interaction DONE

By selecting the object in the UI, we can move it by dragging the mouse and pressing the right-click

Bonus: Spatial Hash Table DONE

Bonus: Snowball Scene DONE

---

Lab 0: Warm Up

Integrators DONE

The integrator solvers can be chosen from the comboBox next to each SolverN at the UI and pressing restart.

• Symplectic Euler
• Midpoint
• Verlet
• RK2
• RK4

Forces DONE

Wind force can be set in the UI as accelerations in each X,Y,Z axis. As for the air resistance, you can choose between no drag, linear drag, and quadratic drag.

• Wind as constant acceleration force
• Linear Drag
• Quadratic Drag

SceneProjectiles DONE

• Added an additional particle for more testing
• Simple collision to plane with condition y<=0
• Added bouncing and friction coefficients editable in UI