Faster Invoke for reflected property access and method invocation with AOT compilation

The bane of the iOS programmers life, when working with reflection in Mono, is that you can’t go around making up new generic types to ensure that your reflected properties and methods get called at decent speed. This is because Mono on iOS is fully Ahead Of Time compiled and simply can’t make up new stuff as you go along. That coupled with the dire performance of Invoke when using reflected properties lead me to construct a helper class.

This works by registering a series of method signatures with the compiler, so that they are available to code running on the device. In my tests property access was 4.5x faster and method access with one parameters was 2.4x faster. Not earth shattering but every little helps. If you knew what you wanted ahead of time, then you could probably do a lot better. See here for info.

You have to register signatures inside each class I’m afraid. Nothing I can do about that.

So to register a signature you use:

static MyClass()
{
     //All methods returning string can be accelerated
     DelegateSupport.RegisterFunctionType<MyClass, string>();         
     //All methods returning string and taking an int can be accelerated
     DelegateSupport.RegisterFunctionType<MyClass, int, string>();    
     //All methods returning void and taking a bool can be accelerated
     DelegateSupport.RegisterActionType<MyClass, bool>();             

}

Then when you have a MethodInfo you use the extension method FastInvoke(object target, params object[] parameters) to call it. FastInvoke will default to using normal Invoke if you haven’t accelerated a particular type.

       myObject.GetType().GetProperty("SomeProperty").GetGetMethod().FastInvoke(myObject);
       myObject.GetType().GetMethod("SomeMethod").FastInvoke(myObject, 1, 2);

You can download the source code for FastInvoke from here.

Fixing the width of strings displayed in the Unity GUI

Firstly, I know what you are going to say – you should just use two aligned labels, rather than fiddling around with string formatting – but sometimes that just doesn’t work – like when you want the items in a popup to have columns like here:

No chance for using labels in that component.  So there is a way of effectively padding text to a given width.  You basically work out the width of a ” “, a tab and the string, then use that combination to make a new string which is the correct width.

Just using string.PadRight(20) doesn’t work due to the variable width of characters in the font.

public static class TextHelper
{
	public static string FixTo(this string str, float width, string type="label")
	{
	    var widthOfTab = GUI.skin.GetStyle(type).CalcSize(new GUIContent("\t")).x;
		var widthOfSpace = GUI.skin.GetStyle(type).CalcSize(new GUIContent(" ")).x;
		var widthOfString = GUI.skin.GetStyle(type).CalcSize(new GUIContent(str)).x;
	    return str + new String(' ', (int)((width-widthOfTab)/widthOfSpace)+1) + "\t";
	}
}

You basically use myString.FixTo(150) to fix it based on the width of a “label” in the skin or you can override it to set a different font by using myString.FixTo(200, “box”).

Should work fine in javascript so long as you make a .cs file out of this and put it in your plugins folder.

Extended Unity Coroutines

If you’ve downloaded Unity Serializer you have an updated version of this already

I really like the coroutine pattern in unity but unfortunately it’s not possible to directly extend it. What I would really like is to be able to write cut scene and NPC logic in a single routine, but I need to do things like wait for an animation to complete or an AI task goal to be achieved. Not being able to write my own YieldInstructions I took to the keyboard and wrote an extended coroutine system that uses as Unity’s existing stuff as a base.

To use it you can download and install the package, then when you want to start an extended coroutine you use:

StartCoroutine( RadicalRoutine.Run(yourEnumeratorFunction()) )

YOU MUST insert the RadicalRoutine.Run inside the StartCoroutine call or not much will happen!

Note it only works with the none “text” based version of StartCoroutine at the moment.

Now you can write your own classes that indicate the completion status, they inherit from CoroutineReturn and can set or override two key values.

• finished – set to true / or override and return true when the action has finished
• cancel – set to true / or override and return true if you want the whole coroutine to be aborted

I’ve included one that waits for a part of an animation (end or some point in the middle) to complete and a version that waits for the weight of some animation to reach a value (I use this to keep my blended animations smooth in cut scenes). It’s called WaitForAnimation and there’s an extension class that adds methods to GameObject: gameObject.WaitForAnimation(name, normalizedTime /* default 1f */) and gameObject.WaitForAnimationWeight(name, weight /* default 0f */);

You can also yield any standard Unity YieldInstruction like WWW, WaitForSeconds etc.

You can use the coroutine functions from c# or javascript.

Example

Here’s an example of me using the extensions to run a complete sequence for an NPC. Obviously in this example I’m calling routines in my game that aren’t listed, but they all return CoroutineReturn instances that pause execution until the task is complete. You can see me using the WaitForAnimation and WaitForAnimationWeight methods directly.

What I want to do is move my character over to a table, pick up an envelope, show it to the player and then put it back down again afterwards.

	IEnumerator MoveToPosition_EnterState()
	{
		Vector3 envAngle;

		//Store the current position of the envelope to be picked up
		var rot = msg.transform.rotation;
		var parent = msg.transform.parent;
		var pos = msg.transform.position;
		//Find the point that will carry the envelope
		var ch = gameObject.GetComponentInChildren();

		//Move the character to the envelope's location
		yield return MoveTo((lookPt = msg.transform.Find("EnvelopeLook")).position);
		//Look at the envelope
		yield return RotateTo(envAngle = lookPt.rotation.eulerAngles);
		yield return new WaitForSeconds(0.3f);
		//Play the pickup animation
		Animate("pick_up_from_table");
		//Wait until half way through
		yield return gameObject.WaitForAnimation("pick_up_from_table",0.5f);

		//Put the envelope in the carry point
		msg.transform.parent = ch.transform;
		msg.transform.localPosition = Vector3.zero;
		msg.transform.localRotation = Quaternion.Euler(0,0,0);

		//Wait for the pickup animation to complete
		yield return gameObject.WaitForAnimation("pick_up_from_table");

		yield return new WaitForSeconds(0.2f);
		//Turn to face the camera
		yield return RotateTo(new Vector3(0,10,0));
		//Wait for the rotation animations to complete
		yield return gameObject.WaitForAnimationWeight("left_turn"); //Without a specified second parameter waits for 0
		yield return gameObject.WaitForAnimationWeight("right_turn");
		//Play the excited animation
		yield return Animate("excited");
		yield return new WaitForSeconds(0.5f);
		//Raise the hand containing the envelope
		yield return Animate("hand_raising");
		yield return new WaitForSeconds(1.5f);
		yield return Animate("hand_raising");
		yield return new WaitForSeconds(2.5f);
		//Wave move dramatically
		yield return Animate("waving");
		yield return new WaitForSeconds(0.5f);
		//Turn around to face the table
		yield return RotateTo(envAngle);
		//Play the pickup animation
		Animate("pick_up_from_table");
		//Wait for it to be 40% complete
		yield return gameObject.WaitForAnimation("pick_up_from_table",0.4f);
		//Put the envelope back on the table
		msg.transform.parent = parent;
		msg.transform.position = pos;
		msg.transform.rotation = rot;
		//Wait for the animation to complete
		yield return gameObject.WaitForAnimation("pick_up_from_table");
		//Rotate to face the camera
		yield return RotateTo(new Vector3(0,10,0));
		//Look over shoulder at envelope
		yield return Animate("looking_behind");
		//Sigh
		yield return Animate("relieved_sigh");
		yield return new WaitForSeconds(1);
		//Finish activity
		BlackboardComplete();

	}

Additional

You can also call RadicalRoutine.Create(yourCoRoutineFunction()) which returns an RadicalRoutine instance that can be cancelled. To use that you pass the instances enumerator property to StartCoroutine like this:

r = RadicalRoutine.Create(myRoutine());
StartCoroutine(r.enumerator);

…

r.Cancel(); //Abort on the next iteration

Also the RadicalRoutine has events for cancellation and completion called Cancelled and Finished.

Unity curved path following with easing at constant speed

The Problem

I have a character than needs to follow paths curved through a series of control points as it moves around the world. I also want to ease in and out to speed of movement as the path starts and ends. Sometimes in the middle of a path follow I need to change my mind and have some other path taken or perhaps change from path following to another state. This isn’t possible with iTween at constant speed unless you use PutOnPath, which doesn’t allow easing functions. I would also like to be able to move at a constant speed – either by ignoring the length of individual path elements (and thereby specifying at time fraction for the whole path) or by having a maximum speed my object can travel at.

Solution

Building on the great work done by Andeeee and Renaud, I’ve made a path following class that enables you to specify from 2 to N points and have it build a curved path (if it has more than 2 points!), you can then follow that path using a time from 0…1 and also apply an easing function to make the start and the end points smooth.

It supports Linear, Sine, Quadratic, Cubic, Quintic and Quartic easing functions with control on over whether the path is eased in and out separately.

Line 1 – EaseIn, Line 2 – EaseOut, Line 3 – EaseIn and EaseOut

This allows fine grained control of an eased, spline path in an Update function that is not possible using iTween – this enables you to abort a movement halfway through if something more important is happening .

To use it you download and import the unity package.

Eased spline paths

To use no easing call Spline.Interp(arrayOfPoints, time). Where time is a float between 0 and 1.

The array of points can be an array of Vector3s, GameObjects or Transforms which are implicitly converted to a Spline.Path instance containing an array of Vector3s. DON’T BOTHER to create your own Spline.Path instances!

e.g. transform.position = Spline.Interp(myPathObject.GetComponentsInChildren<Transform>(), time);

This would use the myPathObject and all of its children to define the path. (For example only, don’t go calling GetComponentsInChildren every frame).

To ease the function you can supply one or more of the easing parameters:

Spline.Interp(arrayOfPoints, time, easingType /* e.g. EasingType.Sine */, easeIn /* e.g. true (default) */, easeOut /*e.g. false (default true) */ );

The Spline class also supports Andeeee’s Velocity and GizmoDraw functions to help you with debugging.

PLEASE NOTE: the function uses a different algorithm for short paths so that you can get a number of different effects. If your path has 2 points it does a linear interpolation, 3 points does a Quadratic spline and 4 points does a Cubic spline. The 5 or more point version uses Catmul Rom splines and ensures that the path passes through every point.

You can double up start and end points easily with the Spline.Wrap() call on your path array – you may want to do this for short paths if you want to ensure that all of the points are passed through – or just do it yourself. There should be no effect if you are using InterpConstantSpeed or MoveOnPath when you do this (See below).

Constant Speed

I’ve added an InterpConstantSpeed function too, this tries to make the speed of movement constant between sections. It works in the same way as Interp, with a time value between 0 and 1.

Now just a note on that.  If your path points are moving then you need to watch constant speed interpolations as one section getting much longer or shorter could cause the position to vary wildly.  If your path sections are increasing or decreasing in size then you are better off using Interp and trying to make sure that each path section is roughly the same size (short sections will appear to have lower velocity than large sections, so they need to be “kind of” the same size).  Interp says if your path has 5 sections, each one takes 1/5 of the time to cross.  If you are in section 2 and section 3 gets bigger, you will complete section 2 in the normal time.  Using constant speed, the position at time “t” is along the magnitude of the whole path, if it gets longer or shorter then the position at “t” will change between sections moving the Vector unrealistically forwards or backwards. This is a problem if you are in an early stage of the path and a later path point, far off screen, is moving.  The Vector will stutter for apparently no reason.  If you want constant speed then you need to ensure that your path points maintain exactly the same distance from each other when they move (they need to move in an arc centered on the previous end point).

The second point is that applying “Constant Speed” and “Easing” appears to create a paradox; but InterpConstantSpeed accepts easing functions! Well the “Constant” bit refers to the sizes of path sections being converted to a path magnitude and the easing is applied to the “time” fed into the path calculator: so you can have your cake and eat it 🙂

Path Movement and Character Speed

I’ve also added a speed limited version, this also solves the constant speed issue by having a maximum possible movement speed, you can therefore move those path elements as often as you like – but you can’t specify the time it takes to complete the path. It works kind of like a SmoothDamp – here’s how you go about it:

public class TestFollow : MonoBehaviour {

	public Transform[] path;

	float t = 0;

	// Update is called once per frame
	void Update () {
		transform.position = Spline.MoveOnPath(path, transform.position, ref t, 0.5f);

	}
}

Basically you pass the path, the current position and a reference to a float that will contain the currently targeted path point, the final parameter shown here is the number of world units per second to move.

This routine also has the added benefit that it will move to the start of the path before following it.

Here are all of the parameters to MoveOnPath:

Parameter	Meaning
pts	An array of transforms, Vector3s or GameObjects that describe the path
currentPosition	The current position of the object being moved
pathPosition	The position along the path, this is updated by the call so must be a variable of the class
(Optional)rotation	A quaternion that will be updated to show the rotation that should be used. This can change quickly for slow moving objects, so you might want to smooth it. See below.
maxSpeed	The maximum number of world units to move in one second, default is 1
smoothnessFactor	The smoothing factor is the number of steps on the path that will be used, default is 100
ease	The type from EasingType.Linear, EasingType.Sine, EasingType.Cubic, EasingType.Quadratic, EasingType.Quartic and EasingType.Quintic. Default is Linear
easeIn	true if you want to ease in the function or false if you don’t. Default is true
easeOut	true if you want to ease out the function or false if you don’t. Default is true

Here’s an example of the version that returns a rotation, in this example I use the SmoothQuaternion from a previous post to smooth out the rotations.

#pragma strict

var pathPoints : Transform[];

var t : float;
var sr : SmoothQuaternion;

function Start() {
	sr = transform.rotation;
	sr.Duration = 0.5f;
}

function Update () {
	var q : Quaternion;
	transform.position = Spline.MoveOnPath(pathPoints, transform.position, t, q, 0.5f, 100, EasingType.Sine, true, true);
	sr.Value = q;
	transform.rotation = sr;

}

Javascript

You can use the Spline class from JavaScript, as long as the .cs files are somewhere inside a Assets/Plugins folder (the download does this) and that JavaScript is not inside a Plugins folder.

This example will follow a constant speed path for 20 seconds.

#pragma strict

var pathPoints : Transform[];

var t : float;

function Update () {
	transform.position = Spline.InterpConstantSpeed(pathPoints, t, EasingType.Sine, true, true);
	t += Time.deltaTime/20;
}

The next one moves an item on a path at a maximum of 0.5 world units per second

#pragma strict

var pathPoints : Transform[];

var t : float;

function Update () {
	transform.position = Spline.MoveOnPath(pathPoints, transform.position, t, 0.5f, 100, EasingType.Sine, true, true);

}

Smoothed Vector3, Quaternions and floats in Unity

I’ve put together a set of structs to handle changing the values of floats, Vector3s and Quaternions over time. It’s very simple to use and is far easier to read than having lots of time recognising code or referenced veclocities in the middle of your game logic. You replace a Vector3 (say) with a SmoothVector3 in your code, set its target value, smoothing method and duration, then read its value over time in an Update() call. I use this when I don’t have a gameObject to hand to apply iTween or when I may want to keep changing the position and don’t want to mess with iTween and need the control at every Update call.

You can choose a smoothing mode from slerp, lerp, SmoothDamp and SmoothStep.

To use this in your code, download the .cs file and put it in your Assets/Plugins folder.

Here is an example from JavaScript on how to use the SmoothVector3 to smoothly move an object when a key is pressed.

#pragma strict

var pos : SmoothVector3;

function Start () {
	//Construct a SmoothVector3 by assigning a normal Vector3
	//this initializes the position
	pos = transform.position;

	//The following are optional and override the standard settings

	//Duration in seconds of the transform
	pos.Duration = 1;
	//Smoothing mode from damp, smooth, lerp and slerp
	pos.Mode = SmoothingMode.damp;
}

function Update () {
	if(Input.anyKeyDown)
	{
		//Setting the Value of the SmoothVector3 starts the interpolation
		//This will move the item 1 unit in X,Y from the current position
		//(which may be an interpolated value)
		pos.Value = pos.Value + Vector3(1,1,0);

		//In this example this is equivalent to
		//pos.Value = transform.position + Vector3(1,1,0);

		//An alternative is to use the .Target property
		//which is the current targeted position - in our example
		//this maintains the object on exact boundaries and means
		//that it will accelerate if you click lots of times
		//pos.Value = pos.Target + Vector3(1,1,0);

		//Or of course you could just set the value
		//pos.Value = Vector3(100,100,0);

	}

	//A SmoothVector3 can be used in place of a Vector3 in assignments
	transform.position = pos;

}

With Vector3 and Quaternions you can directly affect the individual elements to set the targeted value.

       mySmoothVec3.x = mySmoothVec3.x + 100;