Heads up: Even though this problem arose while I was working with Unity, it has nothing specific to Unity, and is more about programming logic, so please don't shy away.
I'm using Unity and rotating an object by script. The thing is, if I rotate it to, say, 180 degrees, the object does not rotate exactly to that much and tends to stop at between 179 and 181 degrees. So, to check if rotation is complete I check if the rotation angle is targetAngle +/- 1, which works.
I check using
if (transform.eulerAngles.z > lowerLimit && transform.eulerAngles.z < upperLimit)
where
lowerLimit = targetAngle-1;
upperLimit = targetAngle + 1;
Now, the problem arises when the targetAngle is 0. In this case, my script checks if rotation angle is between -1 and 1. But, -1 should really be 359, so it needs to check if the angle lies between 359 and 1.
How can I implement this?
In other words, I guess I'm asking how to implement a wrap-around number system.
EDIT
Found one work-around. If targetAngle is 0, I treat is specially. It works, but isn't the most elegant.
if (targetAngle == 0.0)
{
if ((transform.eulerAngles.z > 359.0 && transform.eulerAngles.z <= 360.0) || (transform.eulerAngles.z >= 0.0 && transform.eulerAngles.z <= 1))
{
rotate = false;
}
}
else
{
if (transform.eulerAngles.z > targetAngle - 1 && transform.eulerAngles.z < targetAngle + 1)
{
rotate = false;
}
}
You could do ...
lowerLimit = (targetAngle % 360) + 359; //360 - 1;
upperLimit = (targetAngle % 360) + 361; //360 + 1;
if (((transform.eulerAngles.z + 360) % 360) > lowerLimit
&& ((transform.eulerAngles.z + 360) % 360) < upperLimit)
This moves the check away from the zero and you wouldn't have to deal with positive/negative checking.
EDIT
The % operator on the targetAngle restricts the rotating to +/-359 degrees, so a target angle of 721 would come down to 1, and a target angle of -359 would come down to 1. This should do nicely for all cases I think.
EDIT 2
To fix the last case you mentioned in your comment, I guess you'd need to apply the same wrapping logic to your transform.eulerAngles.z values. Probably best to put this wrapping in an extra function now, so try this:
int wrapNumber(int input) // replace int with whatever your type is
{
// this will always return an angle between 0 and 360:
// the inner % 360 restricts everything to +/- 360
// +360 moves negative values to the positive range, and positive ones to > 360
// the final % 360 caps everything to 0...360
return ((input % 360) + 360) % 360;
}
lowerLimit = wrapNumber(targetAngle) + 359; //360 - 1;
upperLimit = wrapNumber(targetAngle) + 361; //360 + 1;
if (wrapNumber(transform.eulerAngles.z) + 360 > lowerLimit
&& wrapNumber(transform.eulerAngles.z) + 360 < upperLimit)
Depending on how often you need to use this, checking for some cases might remove unneeded overhead. For example, the final % 360 within wrapNumber is only needed if the input was positive. If you're calling this ten times per minute it probably won't matter. If you're calling it a hundred times per second, you may want to check how it performs in this situation.
This may be an old thread but after looking at many different snippets all trying to deal with Wrapping I found that Unity has a nice builtin function that simply takes care of business, At least in my case that the end result i required was a lerp so i only had to change it to LerpAngle and it returned a solid result.
Mathf.LerpAngle is your friend.. solved all my issues with popping etc..
http://docs.unity3d.com/ScriptReference/Mathf.LerpAngle.html
Related
Suppose I have a ball moving in a table and I want to evaluate the reflections it will do in the walls.
Suppose it will have a given movement vector:
I know the end position is the red dot, following this logic. For simplicity, let's suppose the table is 4x2 and centered in (0,0). My code to evaluate the position is:
ball.position += ball.movement;
if(ball.position.x > 2) ball.position.x = 2 - (ball.position.x - 2);
else if(ball.position.x < -2) ball.position.x = -2 + (-2 - ball.position.x);
if(ball.position.y > 1) ball.position.y = 1 - (ball.position.y - 1);
else if(ball.position.y < -1) ball.position.y = -1 + (-1 - ball.position.y);
I'd like to know if there is a shorter / faster / smarter way to approach this evaluation, instead of 4 if-else statements.
The for if-else statements can be reduced to two by using Mathf.Abs so a check e.g.
if(x < -2)
else if(x > 2)
simply becomes
if(Mathf.Abs(x) > 2)
then in both cases you can use Vector2.Reflect. And in roder to instantly also clamp the position use Mathf.Clamp
ball.position += ball.movement;
if(Mathf.Abs(ball.position.x) > 2) ball.movement = Vector3.Reflect(ball.movement, Vector2.up);
if(Mathf.Abs(ball.position.y) > 1) ball.movement = Vector3.Reflect(ball.movement, Vector2.right);
// and for instantly also clamp the position
ball.position = new Vector2(Mathf.Clamp(ball.position.x, -2f, 2f), Mathf.Clamp(ball.position.y, -1f, 1f);
Note that you have to change movement component after reflection. So
ball.movement.vx = abs(ball.position.x) > 2 ? -ball.movement.vx: ball.movement.vx;
similar for y
Also note that more reliable model includes preliminary calculation of reflection moment instead of post-factum changing
I have a simple chart whereby the user can determine a start & end direction in radians. The control then draws the chart using an override of OnRender. I am drawing the arcs with StreamGeometryContext.ArcTo. This method has an IsLargeArc property which determines how the arc is drawn (true for > 180 degrees (PI), false for < 180 degrees). I am determining this value from a condition which works fine:
//Rule does not exceed 180 degrees in direction (radian), IsLargeArc= False else true
if (Start < Math.PI && (End - Start) < Math.PI || //currently unknow condition in here to deal with < PI when start angle is > then end angle?)
{
//IsLargeArc = false;
}
else
{
//IsLargeArc= true;
}
The issue comes when the start < end. e.g. From 270 degrees to 120 degrees. I need a condition to satisfy an angle over 180 degrees (PI) in this situation. Maths is not my strong point. I think I need to add PI*2 to the end and then somehow compare the two values but not sure on how to achieve this?
Well, you could add a full circle to the end, (or start; according to the direction) angle e.g.:
if (start < end)
start += 2 * Math.PI; //full circle in radians.
This way you'll add a full circle to the end angle, which doesn't change the position for your drawing and results in a valid and correct angle if you subtract them (start - end).
Although I must say, I would expect a start > endcondition.
If start > end or visa versa, than this tells you something about the direction.
You can use Math.Abs method to get the absolute value of the difference.
Your code might look like the following:
if ((Start < Math.PI && Math.Abs(End - Start) < Math.PI) ||
(Start > Math.PI && End - Start < 0 ))
{
//IsLargeArc = false;
}
else
{
//IsLargeArc= true;
}
private bool IsLargeArc(Styled2DRange range)
{
double angleDiff = GetPositiveAngleDifference(End - Start);
if (angleDiff > Math.PI)
{
return true;
}
return false;
}
private double GetPositiveAngleDifference(double angleDiff)
{
return (angleDiff + (2*Math.PI))%(2*Math.PI);
}
My game involves ships moving and rotating around a target (i.e., an enemy ship). Rotation depends on whether the user wants to rotate by port/starboard, or just by the closest side to the enemy.
The problem:
The angles are being wrapped with MathHelper.WrapAngle(). Keeping the angles between PI and -PI works great, until the rotating ship gets to the point where -3.141 becomes 3.141 (and vice versa). For example, the ship is rotating to port correctly, then when it hits this line it flips over starboard, then back to port again, then starboard again and so on!
I would be very grateful if the community could point out:
What I can do to make the ship rotate logic work correctly when going over the PI/-PI wrap 'barrier'
Point out any inefficiencies in my code (I'm sure there are many, and I'm sure there are many other ways to do this more efficiently)
Link to any relevant articles or tutorials that can help me overcome this issue (this is my first game)
Additional Information:
Ship.ShipMoveState.NoMoveRotate is essentially a flag that tells the ship to rotate (starting at 100th of max speed up until we hit max speed) either port or starboard, whichever is closest. The ship rotates to these sides as that is where the weapons are located. ShipMoveState.AwaitFurtherOrders is tells the ship to rotate depending on the difference in angle between closest side (port/starboard) and angle to enemy.
ShipCompartment primeCompartment = TargetShip.CenterCompartment;
if (FireState == ShipFireState.FireAtTarget)
primeCompartment = TargetCompartment;
// If ship is to the left of target, below will work
Vector2 distanceToDestination = primeCompartment.Position - CenterCompartment.Position;
float angleToEnemy = (float)Math.Atan2(distanceToDestination.Y, distanceToDestination.X);
angleToEnemy = MathHelper.WrapAngle(angleToEnemy);
CenterCompartment.Rotation = MathHelper.WrapAngle(CenterCompartment.Rotation);
float portBatteryAngle = MathHelper.WrapAngle(CenterCompartment.Rotation - Helpers.RightAngle);
float starboardBatteryAngle = MathHelper.WrapAngle(CenterCompartment.Rotation + Helpers.RightAngle);
float allowance = 0.005f;
bool portIsClosest = false;
switch (primaryFacing)
{
case PreferredFacing.None:
// If port battery not facing enemy
if (angleToEnemy > (MathHelper.WrapAngle(portBatteryAngle + allowance))
|| angleToEnemy < (MathHelper.WrapAngle(portBatteryAngle - allowance)))
{
// And starboard battery not facing either
if (angleToEnemy > (MathHelper.WrapAngle(starboardBatteryAngle + allowance))
|| angleToEnemy < (MathHelper.WrapAngle(starboardBatteryAngle - allowance)))
MoveState = Ship.ShipMoveState.NoMoveRotate;
else
MoveState = ShipMoveState.AwaitFurtherOrders;
}
else
{
portIsClosest = true;
MoveState = ShipMoveState.AwaitFurtherOrders;
}
if (MoveState == ShipMoveState.AwaitFurtherOrders)
{
float diff = 0f;
if (portIsClosest)
diff = angleToEnemy - portBatteryAngle;
else
diff = angleToEnemy - starboardBatteryAngle;
RotateShip(diff);
}
else if (MoveState == Ship.ShipMoveState.NoMoveRotate)
{
// Turn to port (if target is between 6 and 9 o'clock)
if (angleToEnemy < portBatteryAngle)
RotateShip(-MaxRotation / 100);
// Turn to starboard (if target is between 3 and 6 o'clock)
else if (angleToEnemy > starboardBatteryAngle)
RotateShip(MaxRotation / 100);
else
{
if (angleToEnemy > portBatteryAngle && angleToEnemy < starboardBatteryAngle)
{
// Turn to starboard (if target is between 9 and 12 o'clock)
if (angleToEnemy < CenterCompartment.Rotation)
RotateShip(MaxRotation / 100);
// Turn to port (if target is between 12 and 3 o'clock)
else
RotateShip(-MaxRotation / 100);
}
}
}
break;
}
Please let me know if you require any further information. Thank you very much for your assistance.
Instead of, for instance,
a < wrap(b-c)
use
0 < wrap(b-c-a)
or
0 > wrap(a-b+c)
This makes it a little less readable, but is the correct way to compare (supposedly small) angle differences.
I'm working on a small game, I have objects which I want to elevate up and down. Object moves to max value of Y -> Object moves to min value of Y -> Repeat. I have a rough idea of how to do this, I would put this in a timer/my update method.
if(Y >= max)
{
direction = "down";
}
if(y =< min)
{
direction = "up";
}
if(direction == "up") Y -= speed;
if(direction == "down") Y += speed;
(Could also use a bool ofcourse but, for the sake of implicity)
But it feels like I'm just re-inventing the wheel, is there by any chance a built in method/math function to do this automatically? eg. SomeFunction(min, max, increment).
I'm using the XNA framework, so functions built into that are ofcourse welcome as well.
Forget having a separate direction flag.
Just use a negative speed for "up" to simplify the code:
if ((Y >= max) || (Y <= min)) // Hit an edge?
speed = -speed; // Reverse direction.
Y += speed;
I have a bouncing ball application and I have to extend it to prevent overlapping of the balls.
When ball overlaps another, they should move away as in real life.
I have to extend the given MoveBall method:
private void MoveBall()
{
prevX = x;
prevY = y;
x += xVelocity;
y += yVelocity;
// Is there too closed ball?
foreach (Ball ball in parentForm.balls)
{
distance = Math.Sqrt(Math.Pow((double)(ball.prevX - prevX), 2) +
Math.Pow((double)(ball.prevY- prevY), 2));
overlap = ((radius + ball.radius) - distance);// +ball.radius;
if (ball.id != this.id &&
ball.id != lastID &&
overlap > 0)
{
lastID = this.id;
if (xVelocity > 0) // roading right
{
xVelocity = -xVelocity;
x -= xVelocity - ball.xVelocity;
}
else if (xVelocity <= 0) // roading left
{
xVelocity = -xVelocity;
x += xVelocity + ball.xVelocity;
}
if (yVelocity > 0)
{ // going up
yVelocity = -yVelocity;
y -= yVelocity - ball.yVelocity;
}
else if (yVelocity <= 0) // down
{
yVelocity = -yVelocity;
y += yVelocity + ball.yVelocity;
}
}
}
// ***********************************************
// ***************** END MY CODE *****************
if (x > parentForm.Width - 10 - (radius) || x < 0)
{
if (x < 0) x = 0;
if (x > parentForm.Width - 10) x = parentForm.Width - 10 - radius;
xVelocity = -xVelocity;
}
if (y > parentForm.Height - 40 - (radius) || y < 0)
{
if (y < 0) y = 0;
if (y > parentForm.Height - 40) y = parentForm.Height - 40 - (radius);
yVelocity = -yVelocity;
}
}
x,y, xVelocity, yVelocity, radius, prevX, prevY declared as int.
overlap, distance as double.
When 2 overlap, they are getting stuck. Why?
Unfortunately, I can't upload all source code because there are lot of modules.
I'm using Visual C# Express 2010.
As no Question is asked explicitly, I will assume the question "Why are the balls sticking together?"
You have only shown one loop in source code, that's not enough ;-) To check all possible collisions, you need to check n*(n-1)/2 possible collisions. That is normally done with two loops. You have to put in careful measures to avoid handling the same collision twice.
The reason that your balls get stuck is that you handle the same collision multiple times. For example two balls colliding exactly horizontal: The left one has velocity 5 and x-position of 100. The other one shall have a position of 110 and velocity of -6. When the collision happens:
x is set to 105.
Collision detected: x is set to 104 and velocity to -5.
The other Ball handles the same collision:
He moves according to his velocity to position 104.
Collision handling: His velocity becomes 6 and position becomes 105.
The balls were at 100 and 110 resp. and have been moved to 104 and 105. While the velocities are now pointing away from each other, the collision handling in the following step will invert them again. So the positions are close together and the velocities are changing sign every frame. The balls seem "stuck to each other".
I hope the answer helps you to understand your problem. For a better implementation of an elastic collision (that handles each collision exactly once) look here: Ball to Ball Collision - Detection and Handling
Having stumbled upon similar issues when I made my first attempts at collision detection algorithms, I'll try to describe what I think is the problem here.
Maybe the balls move fast enough so that, before collision is even detected by your code, they are already partially "inside" each other. When collision detection comes and notices that, it does what it's supposed to do: change the planned trajectories of the objects according to the details of the collision that just happened. The problem is that, because these objects got sort-of "merged" before collision detection caught them, they can't get unstuck because collision detection is fired again, trapping them with each other.
If this is the source of the problem, then maybe the above code would work with a small enough velocity vector. Of course, that's not a real solution, but if it does work for very small velocities, it probably confirms my hypothesis and you have some idea regarding how to proceed.