Soft body physics.

[99TheCreator]

I'm not even going to put this in suggestions, as it will never happen. But if you have been poking around the internet recently you may have found a game called BeamNG drive, in that game all cars have soft body physics, which means if I was driving head on into a pole the car would somewhat "wrap" around the pole, like in real life, someone else can probably explain it better. But I think this would be awesome in KSP because say if you are approaching the mun at a higher speed than the landing legs can support, the legs would bend and buckle under the weight and speed of the rocket. This could also be used if you collide with another object, instead of parts exploding they would bend, snap, and leak fuel if it were a fuel tank while colliding with the object, maybe the rocket would still work, although not as well. One of my favorite reasons is if your engine gets damaged, while landing or something, instead of it just exploding, the nozzle (depending on what engine it is) could bend, or it would be much less fuel efficient, produce less thrust (possibly by leaking fuel through the nozzle, which could ignite and explode, maybe further damaging your rocket) and could possibly give you a chance at going back home. Last reason, also my favorite, is reentry. While reentering the atmosphere, if you go in to steep or at a bad angle parts of the capsule could start to melt. Say you are reentering and you go in nose first, if you dont turn around fast enough the parachute could melt if it is ontop of the capsule, and after that the whole front of the capsule starts melting, until a hole is made and, sadly, your precious kerbalnaughts die. Or if you go in lopsided and one side is more exposed than the other, a hole could be made there, killing the kerbal on that side, but there could be a slim chance you correct yourself and the others survive, or the capsule flips out and they all die. I think this would be awesome if it ever were to happen. BeamNG Drive is made with unity, and if im not mistaken ksp is being ported to unity (?), so with a miracle this could happen. Share your thoughts? Do you think it's a good idea? Or does it suck?

[Whackjob]

KSP isn't being ported into Unity. It *is* Unity.

Also, carriage returns are awesome.

[Zeblote]

BeamNG.drive is made with torque3d, a (bad) open-source engine

[99TheCreator]

BeamNG.drive is made with torque3d, a (bad) open-source engine

I could have sworn I saw the Unity logo when I had it, hmm. Still would be awesome though if this could happen.

[arq]

Some of the things you're suggesting (IE rocket deformation affecting thrust characteristics) are not deformable-body physics but rather fluid dynamics, which are gloriously hideous and would require very sophisticated (read: expensive) computations to do in a setting like KSP.

Some of the other things you're mentioning, like re-entry damage, are already planned features for KSP. Though you're suggesting a very sophisticated ablation model which would ultimately be nearly indistinguishable from a simple 'hitpoint' based model.

I feel that rigid-body physics offer a good tradeoff between realism, computational load (though I really hope Unity can get a GPU-accelerated physics engine), and user transparency. My biggest physics-related wish-item is the new aerodynamics model they're eventually planning to implement, as the current one is *not* a very good characterization of realistic aerodynamics. FAR is better, but is still limited by some of the core KSP mechanics and messes with the gameplay balance a bit (ie significantly reduced dV to orbit) and I'd be much more willing to use a stock model.

[Whackjob]

Ahh, fluid dynamics. Euphemistically called "How many beers can I theoretically chug at once."

[Wallace]

I don't like the idea of this, I much prefer the keep the damage simple, yes it would be cute if the legs on my lander bent if I hit too hard, although as far as gameplay is concerned, there's no real difference to them just dropping off, either way, the lander is ruined, and that is what is important. likewise, if the parachutes melts all cool looking or just explodes and is gone, it's the same thing as far as the soon to be pavement-pancake Kerbals are concerned.

There are games I've played that have had very sophisticated damage models, my favourite being an offroading game I used to play where your car basically got slowly wrecked and distorted out of shape as you went along, the damage was a key part of gameplay as careful driving meant a better condition car later in the race, and learning to keep a car with a twisted chassis going straight was a skill. Also, it looked totally awesome when your car got all battered and messed up,

KSP is different in that constant damage is not a major part of gameplay, most of the time your ship takes no damage or catastrophic damage. You might occasionally get an Apollo 13 moment where you have partial damage and have to get your kerbals home, but mostly this would just be a heap of CPU time devoted just to making crashes look cooler, which, I'm not against cool looking crashes but it certainly is not worth adding more complexity and wrecking gameplay for anyone without a monster PC, smooth running and good frame rate is much more important.

Also, I have a feeling that fancy soft physics + the modular built rockets of KSP would be a mighty glitch fountain for sure.

[Beachernaut]

Meh... I have BeamNG, and enjoy it for what it is..... A car crashing simulator. IMO it's a bad idea to try to integrate soft body physics into KSP. So take a game that is already pretty resource intensive, then add in much more processor intensive physics. You'd need a pretty healthy (above average) gaming machine to launch a simple rocket, much less the stations and transport/colony ships that we see now.

[99TheCreator]

Meh... I have BeamNG, and enjoy it for what it is..... A car crashing simulator. IMO it's a bad idea to try to integrate soft body physics into KSP. So take a game that is already pretty resource intensive, then add in much more processor intensive physics. You'd need a pretty healthy (above average) gaming machine to launch a simple rocket, much less the stations and transport/colony ships that we see now.

Thats why I did not put it in suggestions, I just thought it would be a cool thing that could happen if we all had monster PC's.

[DaveofDefeat]

I would put soft bodied physics in the magical imaginary ksp that only exists in my head, where the physics are multithreaded and run on a real physics engine. the game is 64 bit and not in origin and all of my parts are not made out of nitroglycerin and explode on impact.

[Milkshakefiend]

Thats why I did not put it in suggestions, I just thought it would be a cool thing that could happen if we all had monster PC's.

You'd need seriously monster PCs for on-the-fly fluid dynamics simulation in KSP. I mean like Godzilla PCs. I'm not really up to speed on game engine physics simulations - I saw a quite impressive particle-based simulation of water done by PhysX a few months ago (

) - but I do have to do fluid/structure interaction simulations using finite element software, and that seriously kills PCs for anything other than trivial problems. It's not just CPU demand either - the matrices that have to be manipulated each time frame can get absolutely gargantuan, which eats RAM and as we know we can't address enough of that already as it is

Also, perhaps someone in the know can answer this, but I don't believe all these lovely fluid simulations are truly coupled simulations - and by that I mean that solids affect a fluid flow, but the fluid does not cause any sort of 'loading' of the solid itself? If so, the computational demand of these simulations seen thus far is smaller than what would be needed for KSP. It wouldn't be enough for the sea or the air to react and move accurately to disturbance (in the case of air that would be fairly pointless as we can't see it most of the time) but the fluid would need to present a force upon the solid for the solid to deform/heat up/vibrate like a bee/etc., otherwise the whole shebang would be a waste of processing.

Some of the time physics problems can be simulated without all the pesky computation by making some gross assumptions and/or using analytic solutions. At some point I want to try to integrate exhaust mach diamonds (if Unity's particle system will allow this) and you can be sure I will not be simulating wave behaviour!