I am having an issue with boxes that land on flat surfaces freaking out when they land, and changing their angle slightly. The following piece of code simulates a box with a flat bottom landing on a flat surface.
import pymunk
space = pymunk.Space()
space.gravity = (0, -1)
box_body = pymunk.Body()
box_shape = pymunk.Poly.create_box(box_body, (1, 1))
box_body.position = (0, 20)
box_body.velocity = (0, 0)
box_shape.mass = 1
floor = pymunk.Segment(space.static_body, (-5, 0), (5, 0), 1)
space.add(floor, box_body, box_shape)
for _ in range(63):
space.step(0.1)
print(box_body.angle)
I would expect the angle to be 0.0, as there is nothing that should be causing it to rotate. However, I actually get a value of -4.8183679268731794e-14. If I change the 63 to 64, I get 7.805375706901558e-15. It's fairly minor in this case, but when other dynamic objects interact with the box, it goes crazy and can often launch other objects at high speed. Why does this happen, and what can I do to prevent it? I've tried increasing the mass significantly, but the results don't change.
Related
Can someone suggest how I can vary the UIPushBehavior magnitude force by distance from source. To mirror the affect of the force of wind from a fan on another object. So the closer the object is to the fan the stronger the force.
if String(describing: identifier) == "fan1" {
self.push = UIPushBehavior(items: [self.object], mode: .instantaneous)
self.push.setAngle(.pi/4, magnitude: 1)
self.animator.addBehavior(self.push)
self.push.addItem(self.object)
}
Use a UIFieldBehaviour to create a linear gravitational field in the direction of the fan. Then you can specify a falloff:
var forceField: UIFieldBehaviour!
// ...
forceField = UIFieldBehavior.linearGravityField(direction: CGVector(dx: 1, dy: -5))
// example values for a "fan" on the bottom left blowing mostly upwards:
forceField.position = CGPoint(x: view.bounds.minX, y: view.bounds.maxY)
forceField.region = UIRegion(radius: 3000)
forceField.minimumRadius = 100
forceField.falloff = 5
forceField.strength = 10
forceField.addItem(view1)
forceField.addItem(view2)
animator.addBehavior(forceField)
Have fun playing around with these values!
Adding collision, another gravity behaviour, and a dynamic item behaviour to two views, we get the following effect:
That feels like a fan on the bottom left to me!
You can also choose a radial gravitational field positioned at where the fan is, if your fan is in a corner and blows "radially", but note that you should use a negative value for strength in that case to say that the field repels rather than attracts.
I've been trying to implement an infinite background animation, which should change between 4 images of equal height and then repeat the sequence. However, it does not seem to work properly.
Note anchorPoint = CGPoint(x: 0.5, y: 0)
func updateBackground(currentTime: TimeInterval){
var delta: CGFloat = 0.0
if lastUpdate != nil {
delta = CGFloat(currentTime - lastUpdate)
}
//First increment position
activeBackground1.position.y += delta*backgroundVelocity
activeBackground2.position.y += delta*backgroundVelocity
//Detect bounds surpass
if activeBackground1.position.y > activeBackground1.size.height + screen.height/2 {
lastSky = (lastSky + 1)%4
sky1 = SKTexture(imageNamed: "sky" + String(lastSky))
activeBackground1.texture = sky1
//Reposition: background1 new position is equal to minus the entire height of
//background2 + its y size.
activeBackground1.position.y = -abs(activeBackground2.size.height-activeBackground2.position.y)
}
if activeBackground2.position.y > activeBackground2.size.height + screen.height/2 {
lastSky = (lastSky + 1)%4
sky1 = SKTexture(imageNamed: "sky" + String(lastSky))
activeBackground2.texture = sky1
activeBackground2.position.y = -abs(activeBackground1.size.height-activeBackground1.position.y)
}
}
The update algorithm works fine, but when it is needed to reposition one of the two background, it seems there is an offset of about 10.0 CGFloat from one background to another. What am I doing wrong?
EDIT: It turned out that the error was located in my image, which presented some blank rows and therefore generated visualisation glitches. So my code works perfectly.
I do the test and most likely you should use something like:
activeBackground2.position.y = activeBackground1.size.height + activeBackground1.position.y
instead of
activeBackground2.position.y = -abs(activeBackground1.size.height-activeBackground1.position.y)
I did this example and it works correctly: https://github.com/Maetschl/SpriteKitExamples/tree/master/InfiniteBackground/InfiniteBackground
Feel free to see and use.
Your problem is floating point math causing rounding errors. I am on a phone right now so I can’t wrote code, but what you want to do is have 1 parent SKNode to handle your entire background.
Add your background slivers to the parent node.
You then place the moving action on the parent only.
As each sliver leaves the screen, you take the sliver, and move it to the end of the other slivers.
This jumping should always be done with integer math, leaving no holes.
Basically:
The floating point moving math is done on the parent node.
The integer based tile jumping is done on each of the slivers.
I'm using code from the pymunk index_video to create a generic function that creates multiple cars which race each other and if they reach the right extreme of the screen, they are removed from Space and re-generated on the left extreme of the screen.
The problem is, that in the example code, each part of the car (chassis, pin joint, motor, wheels) is added to Space separately. I wanted to treat the entire car as a single body whose coordinates I can keep track of by storing the reference of the entire car body in a List and add or delete it to the Space easily.
Also, if the wheels are too close to the chassis, they collide with each other. I presume using a ShapeFilter can help avoid such collisions, but for that I need all parts of the car as a single body.
Please bear with me. I'm completely new to this jargon.
def car(space):
pos = Vec2d(100,200)
wheel_color = 52,219,119
shovel_color = 219,119,52
mass = 100
radius = 25
moment = pymunk.moment_for_circle(mass, 20, radius)
wheel1_b = pymunk.Body(mass, moment)
wheel1_s = pymunk.Circle(wheel1_b, radius)
wheel1_s.friction = 1.5
wheel1_s.color = wheel_color
space.add(wheel1_b, wheel1_s)
mass = 100
radius = 25
moment = pymunk.moment_for_circle(mass, 20, radius)
wheel2_b = pymunk.Body(mass, moment)
wheel2_s = pymunk.Circle(wheel2_b, radius)
wheel2_s.friction = 1.5
wheel2_s.color = wheel_color
space.add(wheel2_b, wheel2_s)
mass = 100
size = (50,30)
moment = pymunk.moment_for_box(mass, size)
chassi_b = pymunk.Body(mass, moment)
chassi_s = pymunk.Poly.create_box(chassi_b, size)
space.add(chassi_b, chassi_s)
vs = [(0,0),(25,45),(0,45)]
shovel_s = pymunk.Poly(chassi_b, vs, transform = pymunk.Transform(tx=85))
shovel_s.friction = 0.5
shovel_s.color = shovel_color
space.add(shovel_s)
wheel1_b.position = pos - (55,0)
wheel2_b.position = pos + (55,0)
chassi_b.position = pos + (0,-25)
space.add(
pymunk.PinJoint(wheel1_b, chassi_b, (0,0), (-25,-15)),
pymunk.PinJoint(wheel1_b, chassi_b, (0,0), (-25, 15)),
pymunk.PinJoint(wheel2_b, chassi_b, (0,0), (25,-15)),
pymunk.PinJoint(wheel2_b, chassi_b, (0,0), (25, 15))
)
speed = 4
space.add(
pymunk.SimpleMotor(wheel1_b, chassi_b, speed),
pymunk.SimpleMotor(wheel2_b, chassi_b, speed)
)
So this question is actually two questions.
A. How to make a "car object" that consists of multiple parts
There is no built in support for this, you have keep track of it yourself.
One way to do it is to create a car class that contains all the parts of the car. Something like this (not complete code, you need to fill in the full car)
class Car():
def __init__(self, pos):
self.wheel_body = pymunk.Body()
self.wheel_shape = pymunk.Circle()
self.chassi_body = pymunk.Body()
self.chassi_shape = pymunk.Poly()
self.motor = pymunk.SimpleMotor(wheel_body, chassi_body, 0)
def add_to_space(self, space)
space.add(self.wheel_body, self.wheel_shape, self.chassi_body, self.chassi_shape, self.motor)
def set_speed(self, speed)
self.motor.rate = speed
def car_position(self)
return self.chassi_body.position
B. How to make parts of the car to not collide with each other
This is quite straight forward, just as you already found the ShapeFilter is the way to go. For each "car", create a ShapeFilter and set a unique non-zero group on it. Then set that ShapeFilter as the filter property on each shape that makes up the car. It doesnt matter if the shapes belong to the same body or not, any shape with a ShapeFilter with a group set will not collide to other shapes with the same group set.
I have a SceneKit game in swift and in it I have a car with a dynamic physics body that is set up like this:
let carScene = SCNScene(named: "art.scnassets/truck.scn")!
let carNode = carScene.rootNode.childNode(withName: "Cube", recursively: true)!
let carPhysicsBody = SCNPhysicsBody(type: .dynamic,shape: SCNPhysicsShape(geometry: SCNBox(width: 5.343, height: 12.125, length: 4.373, chamferRadius: 0)))
carPhysicsBody.mass = 3
carPhysicsBody.friction = 2
carPhysicsBody.contactTestBitMask = 1
carNode.physicsBody = carPhysicsBody
carNode.position = SCNVector3(x: 0, y: 0, z: 5)
carNode.physicsBody?.applyForce(SCNVector3(x: 0, y: 50, z: 0), asImpulse: true)
car = carNode
floorScene.rootNode.addChildNode(car)
The floor's physics looks like this:
As you can see the car gets launched into the air. Then the gravity in the scene makes it fall, and instead of colliding with the floor, it goes right through it.
The gravity looks like this:
What should I change so it will collide with the floor?
I've made a sample macOS playground, available at github.com/heckj/scenekit-physics-playground that shows a physics body interacting. I suspect the core of the problem in your local example is that the floor object doesn't actually have an SCNPhysicsBody associated with it. Until I explicitly set it on SCNFloor (as static), the text blob "fell through" as you originally described.
I recommend adding sceneView.debugOptions = [.showPhysicsShapes] to see the relevant debugging shapes. I've made (and pushed) a few updates to the above repo. Using stock SCNFloor to establish geometry for the physics collection made a tiny target (which is why the slight horizontal impulse made it appear to "pass through"). This last update sets the floor geometry to a long, wide, thin box.
You want to set the floor to be of the type kinematic, not static. With the car being dynamic and the floor being static, the floor isn't interacting with other objects in any fashion, and you explicitly want a collision.
UPDATE: static vs. kinematic doesn't make a difference for the collision interaction, either work effectively the same way, but having a physics body, and verifying it's size or viewing the interaction with .showPhysicsShapes may answer the underlying question of why they're not interacting.
One of posible solutions for u is to set collisionMargin property for your car from 0.0 to probably 0.01, I had the same problem for ball and plane.
I have 10 sprites which I objects of the main sprite I wrote with different images and starting positions etc. But they all behave the same way. They are sub sprites of the main sprite.
I want to be able to hold mouse click on one's rect and move it round the screen which works perfectly fine. But the problem is they all have the same controls click and drag to move them. So if I am clicking on one of the sprite rects and I drag it over another one it picks it up as well. And I don't want that to happen.
Is there a way to only check for collisions with the top most foreground rect or if someone could explain a way of doing this that would achieve similar results. I have had a look at the rect documentation but I can't find a solution.
def update(self,):
self.move(self.rect)
def move(self,rect):
if pygame.mouse.get_pressed() == (1, 0, 0) and the_rect.collidepoint(pygame.mouse.get_pos()):
self.state = 1
elif pygame.mouse.get_pressed() == (0, 0, 0) and the_rect.collidepoint(pygame.mouse.get_pos()):
self.state = 0
if self.state == 0:
the_rect.centerx = the_rect.centerx
the_rect.centery = the_rect.centery
elif self.state == 1:
(the_rect.centerx, the_rect.centery) = pygame.mouse.get_pos()
Rather than using the pygame.mouse.get_pressed() function, use the event queue and check for a pygame.MOUSEBUTTONDOWN event. It will only get fired once when the button is first pressed.