Prysx, Grade 12 Final Project

[Saad]

For my summative project, I made Prysx, which is a particle and rigid body physics engine put together. This is a demo I put together. The graphics engine uses OpenGL, and for window management SDL was used.

code:

CONTROLS:      If your drawing a polygon      Right click stops the drawing of polygon      Left click to set the starting point and left click to set the final radius at your targeted radius      If point is on a polygon      Middle click breaks apart the polygon      Or else      Left click spawns particles      Right click pulls the particles towards the mouse      Middle clicks blows them away from the mouse

Enjoy

Edit: Use this link for the latest versions

[Aziz]

Is sweet. You know my views about it, I'll post the gist of suggestions for others:

- Colors
- Option to take of side boundaries
- Force of attraction/unattraction proportional to distance

I'm having fun with it (great distraction at work!)

[Euphoracle]

Ooo, pretty and fun. I like it. Too bad my school doesn't let us do anything cool like this to show off a bit for our projects ):

[apomb]

Congrats on getting all the bugs worked out. Its very nice.

[Dan]

Very nice and for the linux fans out there, the exe seems to run fine threw WINE if you don't want to complie it.

[Euphoracle]

Saad is more pro at cross compatibility than most commercial identities, so it seems (:

[CodeMonkey2000]

Saad, what's your secret to blurring?

[zylum]

Pretty nice! I see you only used convex polygons so I'm assuming you used the separating line theorem for collision detection? If not, can you briefly describe your method? I'm writing a physics engine as well but with arbitrary polygons so I'm using a different method (SLT doesn't work with concave polys).

[Saad]

CodeMonkey2000 @ Tue Jun 17, 2008 7:09 pm wrote:

Saad, what's your secret to blurring?

No secret, basically drawing random lines over the current graphics. (Its in Main.cpp should you want to look around).

zylum wrote:

Pretty nice! I see you only used convex polygons so I'm assuming you used the separating line theorem for collision detection? If not, can you briefly describe your method? I'm writing a physics engine as well but with arbitrary polygons so I'm using a different method (SLT doesn't work with concave polys).

I did use convex polygons and separating axis theorem. However should you want to use concave and separating axis theorm, I recommend splitting the polygons into convex polygons. You can achieve this by triangulating the polygons and joining up those triangles that that create convex polygons to create smaller convex polygons. Now you test each of those smaller polygons for collision (Should any collide then you have a collision). I also think you could test it by using intersections of each line its up to you

Edit: Just to say one thing about the collision. It is not perfect right now and the collision doesn't work with polygons that have the same axis (polygons with even number edges, a square has 2 of the same axis). This is why if you noticed the polygon sides was always odd . Perhaps some input?

[Reality Check]

Awesome dude. It's great that you guys did C++ in grade 12 though!

[zylum]

I was wondering which method would be faster as they are both O(n*m). Also they both have very simple loops with few calculations. Though SAT quits when it finds that two polygons aren't intersecting and line intersections quits when it does find an intersection. Typically, it is more likely that two polygons are not intersecting but how much complexity would be added for concave polygons?

[md]

This is pretty cool!

[Saad]

zylum @ Wed Jun 18, 2008 1:10 am wrote:

I was wondering which method would be faster as they are both O(n*m). Also they both have very simple loops with few calculations. Though SAT quits when it finds that two polygons aren't intersecting and line intersections quits when it does find an intersection. Typically, it is more likely that two polygons are not intersecting but how much complexity would be added for concave polygons?

For concave polygons you will be testing for which of the convex polygons, but I would go with SAT since it wouldn't check some edges that have no need to be checked.
Also SAT would let you project your edges with velocities which makes it much easier find if they'll collide in a frame or not aswell as give you collision normals and translation vectors so that they don't collide.

And thanks all for the great comments

[corriep]

The link doesn't work for me

[saltpro15]

bad necro paul!