As you can see, I've been screwing around with 3D animated physics simulation (with respect to gaming) for some time now.
Today I decided to set myself a new challenge - call it a milestone if you like.. I want to simulate a trebuchet (a medieval siege weapon similar to a catapult).
In order to achieve this, I'll have to build a rudimentary physics engine capable of simulating the following items:
-a 'type one lever and fulcrum' (think seesaw)
-a sling (rope physics)
-articulation of basic 3D boxes
-crude ballistics (I'll probably make it throw spheres for starters..)

Yes, I know there's a bunch of free physics engines out there, but that's not the point of the exercise.. it's an educational project, my hope is to outperform the most popular physics engines, with the resulting engine and its source being 100% open-source (no license of any kind).

Is anyone interested in becoming involved in such a project?
Posted on 2006-05-16 08:40:39 by Homer

As you can see, I've been screwing around with 3D animated physics simulation (with respect to gaming) for some time now.
Today I decided to set myself a new challenge - call it a milestone if you like.. I want to simulate a trebuchet (a medieval siege weapon similar to a catapult).
In order to achieve this, I'll have to build a rudimentary physics engine capable of simulating the following items:
-a 'type one lever and fulcrum' (think seesaw)
-a sling (rope physics)
-articulation of basic 3D boxes
-crude ballistics (I'll probably make it throw spheres for starters..)

Yes, I know there's a bunch of free physics engines out there, but that's not the point of the exercise.. it's an educational project, my hope is to outperform the most popular physics engines, with the resulting engine and its source being 100% open-source (no license of any kind).

Is anyone interested in becoming involved in such a project?

I put together a catapult model for you
Posted on 2006-05-16 15:42:43 by daydreamer
It would be cool if you could outperform PhysX ;P
Posted on 2006-05-17 04:26:20 by f0dder
PhysX cards are selling "like cold cakes".
Regardless, there will be a place for software-based physics for at least the next decade, and perhaps beyond.

Hey, I'd love to be able to claim that my software could keep up with a dedicated math coprocessor, I'll be happy if I can beat up the handful of public C-based engines.

Heard of Unified Field Theory? Einstein talked about it but never got a handle on it.. apparently, Half-Life 2's physics engine is based on a working unified theory, developed as a speedup.. the author was sued by the parent company for merely mentioning it in public !! They claimed that the unified physics formula was their intellectual property, as was any derivative work.. imagine if Einstein had patented e=mc^2 etc..
The author of the work countersued (his employer), claiming that the unified theory had been in existance ever since the Big Bang, and that it was not subject to copyright law..
Posted on 2006-05-18 02:15:52 by Homer
O_o
Posted on 2006-05-18 07:03:43 by f0dder
I've been experimenting with a Molecular physics model.. basically I figured out how to calculate the Inertia Tensor for arbitrary meshes, so I can describe the mass distribution far more accurately.

I'm now looking into using Signed Distance Maps to accelerate the mesh collision detection/response..

My idea is that the 3D artist creates their articulated model using separate meshes, and/or I define submeshes as "subsets of vertices affected by each bone" as described in my skinmesh thread.. having described a series of semirigid bodies, I then calculate the mass properties of each, and also build a SDM for each  unique body.

The SDM is used for microcollision testing, and a sphere test is used for macrocollision testing.
A hierarchy of macrotest spheres based on the body hierarchy is a further option.
Posted on 2006-05-20 00:06:42 by Homer
I was wondering why HL2 was pretty decent when it came to the physics engine (bloated software).
Posted on 2006-05-20 02:46:50 by SpooK

Game Engine Software Engineer at Valve, Jose Garcia discovered the theory. "The game engine ran too slowly. I was assigned the job of speeding it up," he said. "I started out by combining some of the gravity equations with some of the other force equations and found it all started to fit together. After a day, I had fine-tuned the entire physics-animation functions down to four lines of code, which ran a bit faster," he added.


It sounds like they found a way to converge the force equations.. the conventional logic is to calculate forces separately and then to sum them.. I've been staring at the equations for so long now that my eyes are bulging, but unless I reverse-engineer the binary (or study the leaked source like they did at Cornel University), I don't think I'll ever understand :(

Posted on 2006-05-21 01:41:12 by Homer
I thought HL2 used Havok?
Posted on 2006-05-21 02:35:48 by f0dder
I also have some math theories, that I want feedback on, right now I gonna test an approach of bruteforce calculation and saves the result in LUT's
ok I try to make it lowpoly, for it to also work on lowend systems

Posted on 2006-05-21 03:54:41 by daydreamer
f0dder - hl2's engine is known as "Source" to the best of my knowledge Havok was the original (hl1) engine..
Posted on 2006-05-21 04:01:57 by Homer
Original HL didn't really have physics. I know the HL2 engine is Source, which is their own work... but it does seem to incorporate Havok (http://games.ign.com/articles/402/402601p1.html), and probably a few other commercial libraries (not at home right now, so can't check for stuff like the MILES sound system etc).

Btw, the physics are nice, but a bit limited - try playing with GMOD for HL2 and you'll see that you can't add a couple hundreds exploding barrels ;P
Posted on 2006-05-21 04:24:42 by f0dder
Havok Version 1.0 was unveiled at the Game Developers Conference (GDC) in 2000. Version 2.0 was launched at the GDC in 2003.?  A very nice SDK that is used intensively by Microsoft and is found in GameBoy, XBox, etc, etc.

Homer:  You scare me!  :lol:

Paul
Posted on 2006-05-23 08:34:04 by PBrennick

I just see things a little differently to most people.
In terms of the human animal, how often do we believe we've learned everything there is to know, found every solution and every tool, and completely mastered any given subject, only to have some punk amateur see things from a different perspective and make all the experts feel like fools for not seeing it sooner?
I believe in the Paradox of the Expert :)
Posted on 2006-05-23 12:44:17 by Homer

I believe in the Paradox of the Expert :)


I'll drink to that and my OS Development ;)
Posted on 2006-05-23 15:40:13 by SpooK
Havok on GameBoy would be strange - a RAM-limited, FPU-less device that can hardly cope any 3D ^^. And 2D physics aren't as demanding of realism in such games.
Posted on 2006-05-24 07:17:37 by Ultrano