Prevent Collision Forces in Unity Physics in Collision Callback - unity3d

How do I prevent a collision from applying forces in Unity? I am using 2D physics and want an arrow to stick into a crate. I can easily remove the rigid body and collider in the collision callback, but it seems that a frame of collision force is still applied to the arrow, causing slight jumps in position and rotation. Settings isKinematic on the rigid bodies in the collision callback also appears to not prevent this one frame of force being applied.
I am hoping to tell Unity to not apply physics for the collision.
Using kinematic for the life time of the arrow is not an option because the arrow needs to fly realistically until it hits something.
Here is the code for the crate object that handles the collision:
protected virtual void HandleCollision(ArrowScript arrow, Collision2D coll)
{
StickArrow(arrow, coll);
if (DestroyAfterSeconds >= 0.0f)
{
Destroy(arrow.gameObject, DestroyAfterSeconds);
}
}
private void OnCollisionEnter2D(Collision2D coll)
{
ArrowScript script = coll.gameObject.GetComponent<ArrowScript>();
if (script != null)
{
HandleCollision(script, coll);
}
}
private bool StickArrow(ArrowScript arrow, Collision2D coll)
{
Vector2 surfaceNormal = coll.contacts[0].normal;
float surfaceAngle = Mathf.Atan2(surfaceNormal.y, surfaceNormal.x);
float arrowAngle = Mathf.PI + (arrow.transform.eulerAngles.z * Mathf.Deg2Rad);
float angleDifference = Mathf.Abs(BowAndArrowUtilities.DifferenceBetweenAngles(surfaceAngle, arrowAngle));
float penetration = arrow.PercentPenetration * PenetrationPercentageModifier * (1.0f - angleDifference);
if (penetration <= MinimumPenetrationPercentage)
{
arrow.PercentPenetration = 0.0f;
return false;
}
// Make the arrow a child of the thing it's stuck to
arrow.transform.parent = transform;
arrow.gameObject.transform.Translate(new Vector3(-penetration * arrow.Length, 0.0f, 0.0f));
SpriteRenderer thisSpriteRenderer = GetComponent<SpriteRenderer>();
if (thisSpriteRenderer != null)
{
arrow.GetComponent<SpriteRenderer>().sortingLayerID = thisSpriteRenderer.sortingLayerID;
arrow.GetComponent<SpriteRenderer>().sortingOrder = Mathf.Max(0, thisSpriteRenderer.sortingOrder - 1);
}
BowAndArrowUtilities.PlayRandomSound(arrow.CollisionAudioClips, penetration * 5.0f);
// destroy physics objects from the arrow (rigid bodies, colliders, etc.). This unfortunately doesn't prevent this frame from apply force (rotation, position) to the arrow.
arrow.DestroyPhysicsObjects();
return true;
}
Unity version is 5.3.4.

I ended up making the arrow head a trigger. Inside of OnTriggerEnter2D, I then perform a circle cast in the direction the arrow is pointing with a width of the arrow head sprite. Triggers do not get affected by Unity physics calculations.
private void OnTriggerEnter2D(Collider2D coll)
{
ArrowScript script = coll.gameObject.GetComponent<ArrowScript>();
if (script != null)
{
Vector2 dir = -script.ArrowHead.transform.right;
// ray cast with the arrow size y value (thickness of arrow)
RaycastHit2D[] hits = Physics2D.CircleCastAll(script.ArrowHead.transform.position, script.Size.y, dir);
foreach (RaycastHit2D hit in hits)
{
// collider2d is a member variable assigned in Start that is the Collider2D for this object
if (hit.collider == collider2d)
{
HandleCollision(script, hit.normal);
break;
}
}
}
}

Your problem is that OnCollisionEnter and OnTriggerEnter are called after all the collisions are resolved.
The simplest way to solve this without affecting anything would be to change the weight of the box, arrow, or both.
Set the weight of the crate to a high value, and the weight of the arrow to a low value.
Another way is to use trigger colliders, as you have done. However trigger colliders have problematic side-effects. For example, it doesn't call OnCollisionEnter or OnTriggerEnter on the crate. You will have to do all the logic inside the arrow script, which is not much of a problem.
There are a lot of other ugly hacks however. You could set the velocity of the box to 0 after impact, but it would freeze the crate if you hit it as it was moving. You could use the collision information to cancel the force applied to the crate to solve the collision. You could save the last velocity of the crate every frame, and reapply it to the rigid body during the OnCollision call.
I wouldn't suggest any of these, but they are possible.

Related

Raycast2D hits only one side of Collider

I want to make sure that various objects moving at high speed cannot pass through walls or other objects. My thought process was to check via Raycast if a collision has occurred between two moments of movement.
So the script should remember the previous position and check via Raycast for collisions between previous and current position.
If a collision has occurred, the object should be positioned at the meeting point and moved slightly in the direction of the previous position.
My problem is that works outside the map not inside. If I go from inside to outside, I can go through the walls. From outside to inside not.
Obviously I have misunderstood something regarding the application with raycasts.
using System.Collections;
using System.Collections.Generic;
using UnityEngine;
public class ObsticalControll : MonoBehaviour
{
private Vector3 positionBefore;
public LayerMask collisionLayer = 9;
private Vector3 lastHit = new Vector3(0, 0, -20);
// Start is called before the first frame update
void Start()
{
positionBefore = transform.position;
}
private void OnDrawGizmos()
{
Gizmos.DrawCube(lastHit, new Vector3(.2f,.2f,.2f));
}
// Update is called once per frame
void Update()
{
Vector3 curruentMovement = transform.position;
Vector2 dVector = (Vector2)transform.position - (Vector2)positionBefore;
float distance = Vector2.Distance((Vector2)positionBefore, (Vector2)curruentMovement);
RaycastHit2D[] hits = Physics2D.RaycastAll((Vector2)positionBefore, dVector, distance, collisionLayer);
if(hits.Length > 0)
{
Debug.Log(hits.Length);
for (int i = hits.Length -1 ; i >= 0 ; i--)
{
RaycastHit2D hit = hits[i];
if (hit.collider != null)
{
Debug.Log("hit");
lastHit.x = hit.point.x;
lastHit.y = hit.point.y;
Vector3 resetPos = new Vector3(hit.point.x, hit.point.y, transform.position.z) + positionBefore.normalized * 0.1f;
transform.position = new Vector3(resetPos.x, resetPos.y, transform.position.z);
}
}
}
positionBefore = transform.position;
}
}
Theres a better way to deal with this that unity has built in.
Assuming the object thats moving at a high speed has a RigidBody(2d in your case) you can set its Collision Detection to Continuous instead of Discrete.
This will help collisions with high speed collision, assuming that its moving at high speed and the wall is not moving.
If for some reason you cannot apply this to your scenario, Ill try to help with the raycast solution.
However, I am still wondering about the collision behavior of my raycast script. That would be surely interesting, if you want to calculate shots or similar via raycast
Alright, so your initial idea was to check if a collision had occurred, By checking its current position and its previous position. And checking if anything is between them, that means a collision has occurred. And you would teleport it back to where it was suppose to have hit.
A better way todo this would be to check where the GameObject would be in the next frame, by raycasting ahead of it, by the distance that it will travel. If it does hit something that means that within the next frame, it would have collided with it. So you could stop it at the collision hit point, and get what it has hit. (This means you wouldn't have to teleport it back, So there wouldn't be a frame where it goes through then goes back)
Almost the same idea but slightly less complicated.
Problem would be that if another object were to appear between them within the next frame aswell, it could not account for that. Which is where the rigidbody.movePosition shines, And with OnCollisionEnter you can detect when and what it collided with correctly. Aswell as without the need to teleport it back

GameObject Follow cursor yet also follows enemies?

I'm making a simple character that follows the player's cursor. What I also want is for when the game object "enemy" appears the character then goes to that location to alert the player. Once the enemy is gone the character continues to follow the cursor like normal. Is there a reason why my script won't work. How else can I paraphrase it?
public class FollowCursor : MonoBehaviour
{
void Update ()
{
//transform.position = Camera.main.ScreenToWorldPoint( new Vector3(Input.mousePosition.x,Input.mousePosition.y,8.75f));
if (gameObject.FindWithTag == "Enemy")
{
GameObject.FindWithTag("Enemy").transform.position
}
if (gameObject.FindWithTag != "Enemy")
{
transform.position = Camera.main.ScreenToWorldPoint(new Vector3(Input.mousePosition.x,Input.mousePosition.y,8.75f));
}
}
}
You are not using FindWithTag correctly, as it is a method that takes a string as parameter you need to use it like this:
GameObject.FindwithTag("Something") as stated in the Unity scripting API
Now to apply this to your code you would need to do the following to set your players position based on wether or not an enemy is found (assuming this script is on your actual player object):
if(GameObject.FindWithTag("Enemy"))
{
//If an enemy is found with the tag "Enemy", set the position of the object this script is attatched to to be the same as that of the found GameObject.
transform.position = GameObject.FindWithTag("Enemy").transform.position;
}
else
{
//when no enemy with the tag "Enemy" is found, set this GameObject its position to the the same as that of the cursor
transform.position = Camera.main.ScreenToWorldPoint( new Vector3(Input.mousePosition.x,Input.mousePosition.y,8.75f));
}
However this code will just snap your player instantly to the position of the found Enemy. If this is not the desired behaviour you could use a function like Vector3.MoveTowards instead to make the player move to it gradually.
This code also has room for optimisation as searching for a GameObject every update frame is not the ideal solution. But for now it should work.
I'm going to code coding all the function for you, I'm not pretty sure about the beavihour of your code, I understand a gameobject will be attached to the mouse position, so not really following....
Vector3 targetPosition;
public float step = 0.01f;
void Update()
{
//if there is any enemy "near"/close
//targetPosition = enemy.position;
//else
//targetPosition = MouseInput;
transform.position = Vector3.MoveTowards(transform.position, targetPosition , step);
}
For the f you can use a SphereCast and from the enemies returned get the closest one.

OnCollisionEnter() not functioning with rigid-body and continuous detection while bouncing..?

So, as the title implies, my OnCollisionEnter is not being called. I'm not sure why. The objects are bouncing off surfaces they contact.
Here's the relevant code:
static Rigidbody m_ProjectileRigidbody;
internal void FireProjectile(GameObject projectile, float speed)
{
projectile.transform.position =
State.PlayerTransform.position + State.PlayerTransform.forward;
projectile.transform.rotation = State.PlayerTransform.rotation;
m_ProjectileRigidbody = projectile.GetComponent<Rigidbody>();
m_ProjectileRigidbody.AddForce
(State.PlayerTransform.forward * speed, ForceMode.Impulse);
if (State.PlayerState.Consumes)
{
State.PlayerState.ConsumeCellEnergy(EnergyConsumption);
State.PlayerState.GenerateCellHeat(HeatProduction);
}
}
void OnCollisionEnter(Collision collision)
{
Debug.Log("Collided With: " + collision.gameObject.name);
}
If you are working with 2D colliders and rigidbodies, use OnCollisionEnter2D instead of OnCollisionEnter.
And make sure in Edit -> Project Settings -> Physics the collision matrix is properly set.
And also, double check that:
Both objects have collider, rigidbody properly set up.
Both objects are active.
You do not accidentally disable collider, rigidbody or set
isKinematic, isTrigger from your script.

Camera Move and Rotate Follow Player very jerky

I make a character move on the surface of a circle. I let the camera move and rotate follow character. But the camera move and rotate very jerky. If I increase the value of the third parameter, the shock increases. and to reduce the value of the third parameter, the camera does not rotate to keep up the character. Help me fix it
My Code Camera Follow Player
public class CameraFollow : MonoBehaviour
{
public Transform player;
GameController gc;
public float speed = 2;
Vector3 pos = new Vector3 (0, 0, -10);
// Use this for initialization
void Start ()
{
gc = FindObjectOfType (typeof(GameController)) as GameController;
}
void FixedUpdate ()
{
if (gc.gameState == GameController.GameState.playing || gc.gameState == GameController.GameState.changeWave) {
transform.position = player.position + pos;
transform.rotation = Quaternion.Slerp (transform.rotation,
player.transform.rotation,
speed * Time.deltaTime);
}
}
}
Setting the position of a transform inside of FixedUpdate is a red flag for sure, especially when you're reporting that it's "jerky". Fixed update happens at an irregular interval compared to the frames displayed. This is because Physics needs to update using a fixed time step. The reason why this is the case is out of scope for this question.
Long story short, try changing FixedUpdate to Update and that should fix things looking "jerky".
Let me know if this doesn't work and I'll look for other possible causes.
If you are using a Rigidbody2D to move the character, make sure to set its Interpolate property to 'Interpolate'. This should fix it.

Drag and Drop between 2 gameObjects

I have 2 Spheres in my scene. I want to be able to drag and drop my mouse from one sphere to the other, and have an indicator (a straight line for example) while dragging. After releasing the mouse button, I want to store in the first sphere the other sphere (as GameObject).
I need this in UnityScript, but I can accept C# ideas.
So far I have thought about onMouseDown event on the first Sphere, and then onMouseEnter to the other sphere, I'll store the data in some global variable (which I donno how to do yet) and in case of onMouseExit I'll just put the global variable as null.
Then onMouseUp on the first Sphere I'll store the global variable in the pressed object (the sphere).
Any tips and tricks on how to do it?
Assumptions/Setup
You're working in a 2D worldspace; the logic for dragging the object for this part of the solution would need changing.
Setting parent after click drag, referred to setting the object you didn't click to be the child of the parent you did click on.
The camera should be an orthographic camera; using a perspective camera will cause the dragging to not align with where you think it should be.
In order to make the dragging work, I created a quad that was used as the 'background' to the 2D scene, and put that on a new layer called 'background'. Then setup the layermask field to only use the background layer.
Create and setup a material for the line renderer, (I'm using a Particles/Additive shader for the above example), and for the parameters of the line renderer I'm using Start Width: 0.75, End Width: 0, and make sure Use World Space is ticked.
Explanation
Firstly setup the otherSphere field to be pointing to the other sphere in the scene. OnMouseDown and OnMouseUp toggle the mouseDown bool, that is used to determine if the line renderer should be updated. These are also used to turn on/off the line renderer, and set/remove the parent transform of the two spheres.
To get the position to be dragged to, I'm getting a ray based off of the mouse position on screen, using the method ScreenPointToRay. If you wanted to create add smoothing to this drag functionality, you should enable the if (...) else statement at the bottom, and set the lerpTime to a value (0.25 worked well for me).
Note; when you release the mouse, the parent is immediately set, as such both objects end up getting dragged, but the child will return to it's former position anyway.
[RequireComponent(typeof(LineRenderer))]
public class ConnectedSphere : MonoBehaviour
{
[SerializeField]
private Transform m_otherSphere;
[SerializeField]
private float m_lerpTime;
[SerializeField]
private LayerMask m_layerMask;
private LineRenderer m_lineRenderer;
private bool m_mouseDown;
private Vector3 m_position;
private RaycastHit m_hit;
public void Start()
{
m_lineRenderer = GetComponent<LineRenderer>();
m_lineRenderer.enabled = false;
m_position = transform.position;
}
public void OnMouseDown()
{
//Un-parent objects
transform.parent = null;
m_otherSphere.parent = null;
m_mouseDown = true;
m_lineRenderer.enabled = true;
}
public void OnMouseUp()
{
//Parent other object
m_otherSphere.parent = transform;
m_mouseDown = false;
m_lineRenderer.enabled = false;
}
public void Update()
{
//Update line renderer and target position whilst mouse down
if (m_mouseDown)
{
//Set line renderer
m_lineRenderer.SetPosition(0, transform.position);
m_lineRenderer.SetPosition(1, m_otherSphere.position);
//Get mouse world position
if (Physics.Raycast(Camera.main.ScreenPointToRay(Input.mousePosition), out m_hit, m_layerMask))
{
m_position.x = m_hit.point.x;
m_position.y = m_hit.point.y;
}
}
//Set position (2D world space)
//if (m_lerpTime == 0f)
transform.position = m_position;
//else
//transform.position = Vector3.Lerp(transform.position, m_position, Time.deltaTime / m_lerpTime);
}
}
Hope this helped someone.