As a friend informed me, some mining companies have discovered a way around the horrors of discrete analysis and have gotten surprising results form it. If game engines can be used for predicting the effects of real world explosions, can they be used for other simulations?
Here’s what I’m thinking. Frame rate doesn’t matter at all in this case, so it’s okay to use a high number of physics spheres represented by simple geometries (can you parent sprites to them?). Whenever a collision is detected some checks will be made: how much energy is in either molecule, what chemical does it represent, how much energy should be transferred, do we have enough energy to combust, does it release more energy, etc.? In the latter case molecules will be replaced. Molecules temperatures will be represented by color. Orthographic cameras will be set up before the simulation to record the simulation.
You’re first thought is probably that game engines are designed to look good, not be accurate. As a rebuttal, you probably haven’t seen just how “approximate” engineering analysis methods are From my experience, it seems a system actually trying to get at the core of what we’re dealing with can exceed anything we’re working with now.
Can I get some feed back on this? What problems do you see from the programming side? Do you think it will work? It seems trivial compared to a game to me, but I may be way off base. One problem that worries me is high pressure situations. Do you think that might cause glitches to occur? I think if the frame rate was forced high enough it would go without a hitch. Energy conservation might be a killer though. Thanks!
So a lot when depend on what/where/why you are trying to simulate but there is no reason why not.
The only question is whether some other tool might do it better. (I.e. jme3 is a generic tool, there might be something already developed specifically for the field you are trying to simulate).
Well, there are certainly programs designed for just this sort of thing, but it’s sort of like imperial units sticking around because of old people in industry are used to it. Computer performance changes fast and it makes me think that we’re at a point where direct particle simulation would be better than the traditional methods.
I did a little with physics in jme a while back, but haven’t really done much with jmonkey in total. If my summer isn’t too busy I might try to set up something simple and compare experimental data with it. On a promising note, I found some videos on something like this from a company called Prometech Software from a few years back.
To me the approach also seems a bit cumbersome. As @zarch hinted, lots of effort is made in games just to make them look real because its utterly impossible to simulate reality. Most physics simulations are based on Newton and thus are only meant for specific mechanic simulations really. So you are kind of trying to simulate something to make something that allows you to simulate the former…
If you try to make a perfect simulation, you’re gonna have a bad time. From what I’ve read, even our best simulations don’t work at all for 20% of cases. These programs use fluid equations based off the kinetic theory of gases, which assumes particles behave in a classical, newtonian sense. By doing a particle simulation under the same assumptions (what game engines already do) you cut out the equations and hours of setup time.
From my experience, it seems a system actually trying to get at the core of what we’re dealing with can exceed anything we’re working with now.