Deep Impact is 20 years old! A look back at one of ILM's classic water sims shot



>Successive Over-Relaxation and forward-integrated timestep techniques described in the paper were fundamentally unstable.
Aye, the semi-Lagrangian method is unconditionally stable but the paper about the method was published by Jos Stam about a year later. If i’m not mistaken, in Newton’s physics Verlet integration is also more preferred than Runge Kutta methods.

>a voxel grid (I didn’t know it was called that at the time, so I called them a stacked cube array)
In my opinion it’s just a finite difference method, a way of discretisation of the PDEs, no mater how many dimensions. “Voxel grid” is in my view a more fancy word rather than it’s related to numerical analysis or computer science data structures.

>I only had C at the time, not even C++
It doesn’t matter at all. Engineer doesn’t deal with programming languages, engineer deals with mathematical models, computational models like Turing machines. There’s an interesting article “Programming in the early days of the computer age”: lads in 50s of the previous century were solving differential equations for “inner workings of the nuclear furnace that keep stars shining” on EDSAC, a computer where logic was implemented by vacuum tubes.


Great movie for its time.

Unfortunately CG water effects do not age particularly well.

Not even today’s better water shots will look good when the movie is viewed ten years from now.