I have created a collision class to detect collisions using pixels. In my class I've also developed some functions for determining the collsion angle. Based on this I created some examples:
http://megaswf.com/serve/25437/
http://megaswf.com/serve/25436/
(Space to change gravity, right/left to give some speed to the ball.)
As you will probably notice, there are strange things that happen:
When the ball speed is very low
When the direction of the ball is
almost tangent to the obstacle.
collision http://img514.imageshack.us/img514/4059/colisao.png
The above image shows how I calculate the collision angle.
I call the red dots the keypoints. I search for a maximum number of keypoints (normally 4). If I find more then 2 keypoints, I choose the 2 farthest ones (as shown in one of the blue objects). Thats how I then find the normal angle to where in the surface the object collided. I don't know if this is an obsolete way of doing things.
based on that angle, I rotate the speed vector to do the bouncing.
The piece of code to do the maths is here:
static public function newSpeedVector(speedX: Number, speedY: Number, normalAngle: Number): Object{
var vector_angle: Number = Math.atan2(speedY, speedX) * (180/Math.PI);
var rotating_angle: Number = (2*(normalAngle - vector_angle) + 180) % 360;
var cos_ang: Number = Math.cos(rotating_angle/DEGREES_OF_1RAD);
var sin_ang: Number = Math.sin(rotating_angle/DEGREES_OF_1RAD);
var final_speedX: Number = speedX * cos_ang - speedY * sin_ang;
var final_speedY: Number = speedX * sin_ang + speedY * cos_ang;
return {x_speed: final_speedX, y_speed: final_speedY};
}
This is how the new speed vector is calculated...
My question is, has anyone faced this kind of problem or has some idea on how to avoid this from happening?
Without seeing your code, this is the best I can provide.
Collision physics should have some velocity threshold that is considered "stopped". That is, once the velocity gets small enough you should explicitly mark an object as stopped and exempt it from your collision code. This will help stabilize slow moving objects. Trial and error is required for a good threshold.
When a collision happens, it is important to at least attempt to correct any inter-penetration. This is most likely the reason why tangent collisions are behaving strangely. Attempt to move the colliding object away from what it hit in a reasonable manner.
Your method of determining the normal should work fine. Game physics is all about cheating (cheating in a way that still looks good). I take it rotation and friction isn't a part of what you're going for?
Try this. You have the contact normal. When you detect a collision, calculate the penetration depth and move your object along the normal so that is isn't penetrating anymore. You can use the two points you have already calculated to get one point of penetration, you'll need to calculate the other though. For circles it's easy (center point + radius in direction of normal). There are more complex ways of going about this but see if the simple method works well for you.
I have read and recommend this book: Game Physics Engine Development
I did something a little like that and got a similar problem.
What I did is that when the pixels from the bouncing object (a ball in your case) overlap with the pixels from the obstacle, you need to move the ball out of the obstacle so that they are not overlapping.
Say the ball is rolling on the obstacle, before your render the ball again you need to move it back by the amount of 'overlap'. If you know at what angle the ball hit the obstable just move it out along that angle.
You also need to add some damping otherwise the ball will never stop. Take a (very) small value, if the ball velocity is bellow that value, set the velocity to 0.
You can try one more thing, which I think is very effective.
You can make functions such as getNextX() and getNextY()(Which of course give their position coordinated after the next update) in your game objects and Check collision on objects based on their next position instead of their current position.
This way, the objects will never overlap, you'll know when they are about collide and apply your after collision physics gracefully!
Related
I'm creating this 2D, pixel art game. When the camera follows the player (it uses easing), on the final approach, the position gets several subpixel adjustments.
If I have smoothing ON (on my graphic assets), the graphics look good (sharp. it's pixel art) but the subpixel motion is jerky/jumpy.
If I have smoothing OFF, the subpixel motion is smooth, but the pixel art graphics look blurry.
I'm using Flash player v21. I've tried this with Starling and with Flash's display list.
You have a pixelated object that is moving in increments of less than the pixel size, but you don't want to restrict your mathematical easing to integers, or even worse, factors of 8 or what have you. The solution I am using in my project for this exact issue is posted below (I just got it working last week!)
Concept
create a driver that is controlled by the easing using floating point numbers.
Allow this driver to then control where the actual display object is rendered. We can use a constraint to only allow the display object to render on your chosen resolution.
Code Example
// you'll put these lines or equivalent in the correct spots for your particular needs.
// SCALE_UP will be your resolution control. If your pixels are 4 pixels wide, use 4.
const SCALE_UP: int = 4;
var d:CharacterDriver = new CharacterDriver();
var c:Character = new Character();
c._driver = d; // I've found it useful to be able to reference the driver
d._drives = c; // or the thing the driver drives via the linked object.
// you don't have to do this.
then when you are ready to do your easing of the driver:
function yourEase(c:Character, d:CharacterDriver):void{
c.x = Math.ceil(d.x - Math.ceil(d.x)%SCALE_UP);//this converts a floating point number into a factor of SCALE_UP
c.y = Math.ceil(d.y - Math.ceil(d.y)%SCALE_UP);
Now this will make your character move around 4 pixels at a time, but still be able to experience easing!
The bit with the modulo (%) operator is the key. For instance, 102-102%4 = 100. 103-103%4 = 100. 104-104%4 = 104.
In case anyone is confused by that, look at what 102%4 does: 4 goes into 102 25 times with a remainder of 2. so 102%4 = 2. Then 102 - 2 = 100.
In your case, since the "camera" is following the player (i.e. the background is moving, right?) then you really need to apply drivers to everything in the background instead, but it is basically the same idea.
Hope this helps.
since you specifically mentioned the "final approach" i think your problem comes from the fact that the easing equations puts your graphics at fractional coordinates, especially while getting closer to the target, but you should also notice it during the rest of the animation.
depending on the easing "engine" that you're using you should be able to set a "round values" flag, so all the coordinates set will be integer values and not fractional
if that's not possible, find a way in your display objects to round the x and y values every time they change
I´m programing a space ship side scroll in as3. The bottom of the stage are mountains and here comes the problem, when I try to detect the ship collision against the mountains..
Because the poor collision detection and the need of avoid large loops my idea is create an object that works as a collider itself detecting a collision and avoiding parse all the stage or more selective metod.
I place "by hand" in the flash stage several instances of circles with a class for manage them where I place the If(this.collider.hits(ship)....
I spent looong time but I can find the way to make it work some of the mistakes i get are like this
Error 1061: Call to a possibly undefined method hitTestObject through a reference with static type Class.
some Idea? Thanks in advance
when you hit test with points it is important that the point being tested is relative to the object being tested against, eg
if(mountain.hitTestPoint(this.x + circle1.x, this.y + circle1.y))
will return true if the circles are inside the object calling the function because their position relative to the mountain is now relative to it rather then relative to the ships xy position within the clip... hope that makes sense.
btw I have done this myself in the past but I would have to remind you that you can only hit test with the points so there is no need to have circles, use blank sprites instead and set the visible flag in the properties panel to false, no drawing will make it slightly faster... not that you will notice, also sprites/graphics use less memory then movie clips.
also I would recommend hard coding some points in the clips rather then actually adding the clips in the sprite/clip itself, this will make it easier to work with them and scale later on (believe me this will annoy the hair from your head to do something later and slow the game to scale on the fly)
try something like this... you can determine the points values by adding a clip to the movie clip and getting its position from the properties if you must.
private var hitPoints:Vector.<Point> = new Vector.<Point>
hitPoints.push(new Point(10, 40));
hitPoints.push(new Point(30, 40));
//...do this for all your points
//loop through all your points and check if the hit relative to the ships position.
for(var i:int = 0; i < hitPoints.length; i++)
{
if (scene.hitTestPoint(ship.x + hitPoints[i].x, ship.y + hitPoints[i].y))
{
//do your hit stuff here
break;//don't forget to break
}
}
in this code you will need to make sure the scene object is a reference to your scenery at the bottom of the screen.
I hope this helps but if this is not enough help then you should post some of your code here so we can have a look and see where it can be improved.
I'll try to explain my problem in a clear and short way.
I'm writing a grid-game. In this game when the player clicks somewhere, the player moves, along the grid with a path calculated by the computer (because there's obstacles that the player avoid : walls...) to the final point.
It's an isometric grid but it's like it was a basic 2D grid.
So I have my path, which is a Point Array (key points screen coordinates) :
path : [x=10,y=100],[x=40,y=172], .. etc.
To display the movement of the player, I tried to use tweens (TweenLite/TweenMax).
There's no option on TweenLite to wait for a tween to finish before starting the next. In any case the solutions seem complicated (shitty delays/onFinish:function).
The solution I found is acceptable : TweenLite provides a LinePath2D function which works exactly like I wished. The only problem is that it works with only one function (path is the complete array):
var pathanimate:LinePath2D = new LinePath2D(path);
pathanimate.addFollower(Player);
TweenMax.to(pathanimate, 1, { progress:1, ease:Linear.easeNone });
So I can't "touch" anything during the movement. INCLUDING the aspect of the sprite of the player that must changes with the direction during each step of the path (it would be more simple if it LinePath worked with a loop).
I don't know if this is clear (i'm french huh) but I see only but two options :
Keep the LinePath and have, on each frame, some kind of counter/timer that "measures" the direction of the player. Could be heavy in ressources, but keeps the LinePath2D that works very well alone
Find another solution
I'd be glad to hear your ideas and code !
Thanks in advance
Actually I can provide several solutions, but they will include many lines of code. Briefly speaking:
System of waypoints. You have algorithm to set collection of waypoints to your character, and render him (updating from the main gaming loop), and character reaches one waypoint after another by shifting them from the collection.
Working with tweens, append them (so there will be queue).
In both options, you can set new path for the character, all you need is simple logic to approximate current position of the character to the closest grid point, and calculate new path from there, It's very easy with basic grid systems, like:
closestGridCellX = Math.round(this.x / gridCellSize);
closestGridCellY = Math.round(this.y / gridCellSize);
As i am working on side view bike stunt game in which am trying to add slow motion effect when a bike performs stunts.
Is there any solution for showing such kind of slow motion effect in Box2D. can anybody help me in this regard
Thanks and regards,
Chandrasekhar
As mentioned already, changing the length of the time step can give a slow motion effect. It also has the side-effect of altering the way gravity affects bodies, and can complicate other things like recording a replay or syncing state across multiplayer games for example.
Another option is to use a fixed time step length for every time step, and keep track of the previous position and angle for all bodies. Then you can interpolate between the last frame and the current frame to draw them at an in-between frames position. This means that you are always drawing things a tiny bit behind their current position in the physics engine, but it should not be noticeable with typical frame-rates of 30-60fps.
An easy way to do this effect with or without Box2D is to step up a time modifier.
so lets say you move the player in a run function like so:
player.x += vel_x;
player.y += vel_y;
you could setup a time modifier variable that is initialized to 1
var time_mod:Number = 1;
then update all your movements as followed
player.x += vel_x * time_mod;
player.y += vel_y * time_mod;
then when you want "the slow motion effect" change your time_mod variable. For half of real time change your time_mod to 0.5. If you want to super speed change it to 2 or 3, super slow? change to 0.3
you get the idea?
I know that there are people out there creating classes for this (ie http://coreyoneil.com/portfolio/index.php?project=5). But I want to learn how to do it myself so I can create everything I need the way I need.
I've read about BitMap and BitMapData. I should be able to .draw the MovieClips onto a BitMap so I could then cycle the pixels looking for the collisions. However, It's weird and confusing dealing with the offsets.. And it seams like the MyBitMap.rect has always x = 0 and y = 0... and I can't seam to find the original position of the things...
I'm thinking of doing a hitTestObject first, then if this was positive, I would investigate the intersection betwen the movieclips rectangles for the pixel collisions.
But then there is also another problem (the rotation of movieclips)...
...I need some enlightment here on how to do it.
Please, any help would be appreciated..
If you're using BitmapData objects with transparency you can use BitmapData.hitTest(firstPoint:Point, firstAlphaThreshold:uint, secondObject:Object, secondBitmapDataPoint:Point = null, secondAlphaThreshold:uint = 1):Boolean.
You'll have to change from global coords to the local BitmapData coords which will require a bit of math if it is rotated. That's easily achieved (look up affine transform for more info on wiki):
var coordTransform:Matrix = new Matrix();
coordTransform.rotate(rotationRadians);
coordTransform.translate(x, y);
coordTransform.transformPoint(/* your point */);
A classic reference for pixel perfect collision detection in flash is this Grant Skinner's article. It's AS2, but the logic is the same for AS3 (there are ports available if you google a bit).
If I recall correctly, this particular implementation worked as long as both tested objects had the same parent, but that can be fixed.
About BitmapData x and y values, I understand it could be confusing; however, the way it works makes sense to me. A BitmapData is just what the name implies: pixel data. It's not a display object, and cannot be in the display list; so having x or y different than 0 doesn't really make sense, if you think about it. The easiest way to deal with this is probably storing the (x,y) offset of the source object (the display object you have drawn from) and translate it to the global coordinate space so you can compare any objects, no matter what's their position in the display list (using something like var globalPoint:Point = source.parent.localToGlobal(new Point(source.x,source.y)).
I've previously used Troy Gilbert's pixel perfect collision detection class (adapted from Andre Michelle, Grant Skinner and Boulevart) which works really well (handles rotation, different parents, etc.):
http://troygilbert.com/2007/06/pixel-perfect-collision-detection-in-actionscript3/
http://troygilbert.com/2009/08/pixel-perfect-collision-detection-revisited/
and from there he has also linked to this project (which I've not used, but looks really impressive):
http://www.coreyoneil.com/portfolio/index.php?project=5
I managed to do it after all, and I already wrote my class for collision detections,/collisions angle and other extras.
The most confusing process is maybe to align the bitmaps correctly for comparing. When whe draw() a movieclip into a a BitmapData, if we addChild() the corresponding Bitmap we can see that part of it is not visible. it appears to be drawn from the center to right and down only, leaving the top and left parts away from beeing drawn. The solution is giving a transform matrix in the second argument of the draw method that aligns the bitmap and makes it all be drawn.
this is an example of a function in my class to create a bitmap for comparing:
static public function createAlignedBitmap(mc: MovieClip, mc_rect: Rectangle): BitmapData{
var mc_offset: Matrix;
var mc_bmd: BitmapData;
mc_offset = mc.transform.matrix;
mc_offset.tx = mc.x - mc_rect.x;
mc_offset.ty = mc.y - mc_rect.y;
mc_bmd = new BitmapData(mc_rect.width, mc_rect.height, true, 0);
mc_bmd.draw(mc, mc_offset);
return mc_bmd;
}
in order to use it, if you are on the timeline, you do:
className.createAlignedBitmap(myMovieClip, myMovieClip.getBounds(this))
Notice the use of getBounds which return the rectangle in which the movie clip is embedded. This allows the calculation of the offset matrix.
This method is quite similar to the on shown here http://www.mikechambers.com/blog/2009/06/24/using-bitmapdata-hittest-for-collision-detection/
By the ways, if this is an interesting matter for you, check my other question which I'll post in a few moments.