Opinions beginners or Advanced:

I don’t mean to harp on a problem, but this is an old argument from work. It came to mind because of the discussions in http://www.opengl.org/discussion_boards/ubb/Forum2/HTML/008949.html as to what is a good solution and what is not.

When I am asked to add a particle explosion, I do my best to make it “look” good. Using 3 or 4 imposter based textures per effect (smoke, flame, shock wave, etc), plus point list, etc. (soon point sprites). there is a minor physics engine, and I do mean minor, gravity based on distance (since it has to look “reasonable” at any altitude) linear drag, though pressure estimate by altitude, etc.

I have never been out for any awards, this has never been used as a “explosion simulation”, only as a good look-and-feel system for the customer to know… “ahhh something happened!” Interestingly enough, the customers love it, but the supporting personel hate it. All the technical people want to add full fluid dynamics simulations to it, etc. But how much pay off do you really get from something that looks good, vs. exact science? Last I remember from looking at dynamics equations even a P4 will choke down on frames to bite down on that math…

Anyhow… My bosses are pleased and that is all that really matters, I will probably tweak additions from Siggraph and other places as I find them, and may expand the physics engine little by little; so no one out there is going to hurt my feelings. I am just curious if there is that big of a payoff in exact science vs gaming physics as it is often called?

Go and find a book about Extreme Programming and give to the idiots you work with.

“Do the simplist thing that could possibly work”

Don’t add extra functionality because you aren’t gonna need it"

The guy paying the bills decides when you are finished.

It sounds to me like the solution you’ve implemented is a success: the customer loves it, your bosses are pleased and it works older machines.

I wouldn’t be suprised if your colleagues, in support, are a bit jealous; they seem to immediately point out that “fluid dynamics” is the way to go. Yes, it would be if you had more time and the client base was running dual processor machines.

I would think that implementing fluid dynamics would be much more complex, and therefore “harder to work on”. A “bit” of contradiction, wouldn’t you say?


Video-programming is oftem a matter of “doing like if” rather than “doing like this!”


“If it looks right, it is right”

Sounds to me like your solution is perfect. A little bit of simple physics to fool the viewer into thinking it’s proper physics.

Chances are if you implemented full fluid dynamics it would look crap. And the resource requirements for it would be insane.

Another good rule of thumb is KISS. “Keep It Simple, Stupid”