Physics Engine (NEW VERSION)

Link:
http://www.mediafire.com/download.php?kmshe4g05wj


Changes:
Recoded force field interface to allow for different shape types, new black hole force field demonstrated using a sphere shape and a (force/rho/rho) vector field is now in top right corner.
Optimized incline method for increased efficiency when objects are idle.
Fixed issue with projectile's iterate method's ground detection.
Slightly optimized impact method for ground bouncing.
Added a pit with wisps that do not detonate on contact.

Updated Information in BOLD

Features:
-Up to 6 players supported.
-Kobolds and Yetis to hit each other *Press the A button, the regular attack replaces the attack order
-Jumping abilities
-Dash abilities
-Grenade and boulder abilities
-Jumping Johnny Ethereal Bombs
-Hero Stats effect physical attributes and engine behavior
-Collision between objects
-Units take damage if the velocity of a collision or bounce is past a certain threshold
-Tornadoes
-Black Hole

Issues:
-Units moonwalk under certain circumstances.

Plans:
-A "Warlock" style arena map with rounds and abilities and equipment
-The ability for the player to select hero stats to add upon a levelup

looks cool ill check it out soon..

Make it into a minigame?

I have now solved the problem of high speed collision detection. I actually developed a perfect sphere collision detection algorithm, but it's too inefficient to use when a lot objects that are in collision range, due to the calculations involved, so I just check for the object's speed to see if a priori is necessary, or if a posterion will suffice.

I will post an updated version after some short balancing changes, maybe tonight, maybe tomorrow. For those of you who are interested, the algorithm is basically selecting a point on the surface of each sphere that is closest to the other sphere, then solves the distance formula for those points equal to zero (When the distance between the two points closest to the other spheres is equal to zero, that is the exact point in time at which a collision occurs). The both the position of each sphere and angle between the spheres changes with respect to time, which is why its so complicated, but the ending quadratic that I have the algorithm solve using the quadratic formula is as follows:

Spoiler:

Quote :

(v_1x²+v_2x²+v_1y²+v_2y²+v_1z²+v_2z²-2v_1xv_2x-2v_1yv_2y-2v_1zv_2z)t²+

(2(x₁(v_1x-v_2x)+x₂(v_2x-v_1x)+y₁(v_1y-v_2y)+y₂(v_2y-v_1y)+z₁(v_1z-v_2z)+z₂(v_2z-v_1z)))t+

x₁²+x₂²+y₁²+y₂²+z₁²+z₂²-2x₁x₂-2y₁y₂-2z₁z₂-(r₁+r₂)²=0

And yes, I did have to derive that formula through some lengthy algebra, it took me about 2 hours of going in a horribly wrong direction (solving the derivative of the function equal to zero, would have resulted in three times as many terms to deal with before I could realize that everything cancelled) then I had an epiphany and realized that since it was going to be quadratic anyway, I could just solve the original equation equal to zero since I was picking points with respect to time. After I realized that it only took me about a half hour.

Is this the same we played yesterday? If so, it works very well. No slowing issues to date.

Lagger09 wrote:

Is this the same we played yesterday? If so, it works very well. No slowing issues to date.

The version we played yesterday wasnt even using my algorithm, because I forgot to change the group units filter from the global to a local one that I put in with an if-then to check velocities. This is why the bullets were doing unpredictable damage and going through units sometimes. Because they will actually hit things now, I divided their radius by 15 and gave them a random factor for aiming.

Its still very efficient in practical use. The only problem is that JASS is apparently very slow, and with those 20 wisps floating around in the black hole, the new algorithm is used almost 40*20*20 times per second. But my framerate still hovers around 40-45. If I allow ALL units to use the new algorithm, that wisp pit+black hole tears my computer asunder, 5fps and still dropping.

It should be fine, in any real maps there wont be units going fast enough for long enough with enough units nearby to make it hurt.

lol DBC Rapid Peas and myself played a Control the Hill on the new one and that was FUN as hell

2 suggestions for later development:
1. Enable allies so teams can be made
2. Maybe make different arenas with different objectives such as King of the Hill, Castle Siege, etc, etc, etc

2 new ideas:

Make a mage hero so you have a total of 4 heroes, spell could be fireball and blink

capture the flag mode and king of the hill mode

The movement is still too screwy for me. It is difficult to go where you want.

Also, projectiles should be aimed at an angle according to the height of where you target them. Without that, it is very difficult to shoot up or down hills properly.

Dragonheart91 wrote:

The movement is still too screwy for me. It is difficult to go where you want.

Also, projectiles should be aimed at an angle according to the height of where you target them. Without that, it is very difficult to shoot up or down hills properly.

Yeah, the projectile thing was a current plan, and I'm actually having trouble completely disabling pathing for some reason. The units decide to cry when you click over cliffs and so they start moonwalking and your jump wont be in the direction you're moving. I'm not sure how to fix this completely.

RapidFireOmega wrote:

2 new ideas:

Make a mage hero so you have a total of 4 heroes, spell could be fireball and blink

capture the flag mode and king of the hill mode

Yeah, I was thinking some sort of warlockish hero that would bring a stun feature to the table, units can't move while stunned and their decelleration is disabled (Stop and hold position won't slow you down)

Game modes and arenas are also a priority