A friend of mine and I were discussing different idea for allow materials in a world to be destroyed in a very piecemeal fashion and he proposed the idea of representing solids as very viscious fluids. My intuition is that this would be either: A) very difficult, or B) very resource intensive, but I am unsure.

Is such a model feasible for an RPG/adventure rpg/fps?

edit:clarification: The idea is for collisions and objects breaking to be handled by this system. To basically allow any object handled this way to be destructable instead of having scripted destroyable objects.

  • \$\begingroup\$ why does the genre of the game matter? \$\endgroup\$
    – Adam Harte
    Aug 26, 2010 at 20:22
  • 4
    \$\begingroup\$ I love the idea of having the world slowly melt. \$\endgroup\$
    – eli
    Aug 26, 2010 at 20:36
  • \$\begingroup\$ the genre not so much, more of an example of the scale of a world in this case. I could see some sort of puzzle game being able to get away with alot less processing for this then a game with an entire world. \$\endgroup\$
    – lathomas64
    Aug 26, 2010 at 22:28

3 Answers 3


FastLSM is sorta like what you're after.

Alec has also made 2D game engine based on deforming lattice (instead of voxels as in FastLSM). It's called Physical. Look at the demo asteroids style game to see what can be done and the performance.

Alec also built a full game using physical called Sopwith IV.

I never played Sopwith IV, but I spent plenty of time playing with the asteroids demo.

One of the key features that make it fun is, everything is destructible. Including you own ship. It's hilarious when you snap off on one rocket, but can still control it.

  • 2
    \$\begingroup\$ that's more then sorta, I think FastLSM is exactly what i need. It even handles fracturing and soft/rigid body in combination! \$\endgroup\$
    – lathomas64
    Aug 27, 2010 at 4:46
  • \$\begingroup\$ I said sorta as the bodies will bend and fracture, but they aren't fluid. The volume of an arm can't be pushed into the rest of the body causing it to swell. The bodies are soft bodies instead of rigid, but they are still bodies not fluids. \$\endgroup\$
    – deft_code
    Aug 27, 2010 at 5:02
  • 1
    \$\begingroup\$ They aren't fluids but it is a pretty good solution to the problem the liquid idea was trying to address. \$\endgroup\$
    – lathomas64
    Aug 27, 2010 at 21:24

Glasses are basically that: matter arranged amorphously, that is without an order (a hallmark of a liquid) but with the molecules unable to move very far from their initial positions (very very high viscosity). If it works in real life, there's no reason it can't work in a simulation.

However you have to understand that when I say very very high viscosity, I mean very very very high viscosity. To contrast, water has a viscosity of 0.01 poise. The glass transition occurs (by definition) at 10^13 poise. That is a viscosity that is 15 orders of magnitude (a quadrillion times) higher than what we expect for a regular liquid. The expected motion of said liquid during the duration of the game will be pretty much exactly zero.

Well, that's just perfect! If the liquid does not move, we do not have to simulate it! We can just fill our liquid grid volumes and be done with it. Congratulations, you have just re-invented voxel rendering!

It's an excellent choice for destructible geometry if you can spare the horsepower.

  • \$\begingroup\$ Most (all?) physics engines will not be able cope with a numbers like 0.01 and 10^13 simultaneously and remain stable. I'd bet it can't be done uniformly. The simulation would have to divided into two separate systems, one that models fluids, and another that fakes fluidity in "highly viscous solids". \$\endgroup\$
    – deft_code
    Aug 26, 2010 at 21:25
  • \$\begingroup\$ I actually had a para on that that I snipped out, since I draw the conclusion that you would be silly to actually simulate motion of something that will for all intents and purposes will never move on human timescales. If you scale up time enough, well, even rock flows. Panta rei my brother, panta rei. \$\endgroup\$
    – drxzcl
    Aug 26, 2010 at 21:31
  • \$\begingroup\$ Thanks. I've not done a whole lot in the graphics realm of programming if thats obvious. destructible and/or deformable geometry is precisly the main intent here. Well destructible/constructable anyways. \$\endgroup\$
    – lathomas64
    May 24, 2011 at 21:12

This sort of makes me think about Soft Body Dynamics, but they would never return to their original form?


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .