Input Bindings

An InputBinding represents a connection between an Action and one or more Controls identified by a Control path. An Action can have an arbitrary number of Bindings pointed at it. Multiple Bindings can reference the same Control.

Each Binding has the following properties:

Property	Description
`path`	Control path that identifies the control(s) from which the Action should receive input. Example: `"<Gamepad>/leftStick"`
`overridePath`	Control path that overrides `path`. Unlike `path`, `overridePath` is not persistent, so you can use it to non-destructively override the path on a Binding. If it is set to something other than null, it takes effect and overrides `path`. To get the path which is currently in effect (that is, either `path` or `overridePath`), you can query the `effectivePath` property.
`action`	The name or ID of the Action that the Binding should trigger. Note that this can be null or empty (for instance, for composites). Not case-sensitive. Example: `"fire"`
`groups`	A semicolon-separated list of Binding groups that the Binding belongs to. Can be null or empty. Binding groups can be anything, but are mostly used for Control Schemes. Not case-sensitive. Example: `"Keyboard&Mouse;Gamepad"`
`interactions`	A semicolon-separated list of Interactions to apply to input on this Binding. Note that Unity appends Interactions applied to the Action itself (if any) to this list. Not case-sensitive. Example: `"slowTap;hold(duration=0.75)"`
`processors`	A semicolon-separated list of Processors to apply to input on this Binding. Note that Unity appends Processors applied to the Action itself (if any) to this list. Not case-sensitive. Processors on Bindings apply in addition to Processors on Controls that are providing values. For example, if you put a `stickDeadzone` Processor on a Binding and then bind it to `<Gamepad>/leftStick`, you get deadzones applied twice: once from the deadzone Processor sitting on the `leftStick` Control, and once from the Binding. Example: `"invert;axisDeadzone(min=0.1,max=0.95)"`
`id`	Unique ID of the Binding. You can use it to identify the Binding when storing Binding overrides in user settings, for example.
`name`	Optional name of the Binding. Identifies part names inside Composites. Example: `"Positive"`
`isComposite`	Whether the Binding acts as a Composite.
`isPartOfComposite`	Whether the Binding is part of a Composite.

To query the Bindings to a particular Action, you can use InputAction.bindings. To query a flat list of Bindings for all Actions in an Action Map, you can use InputActionMap.bindings.

Composite Bindings

Sometimes, you might want to have several Controls act in unison to mimic a different type of Control. The most common example of this is using the W, A, S, and D keys on the keyboard to form a 2D vector Control equivalent to mouse deltas or gamepad sticks. Another example is to use two keys to form a 1D axis equivalent to a mouse scroll axis.

This is difficult to implement with normal Bindings. You can bind a ButtonControl to an action expecting a Vector2, but doing so results in an exception at runtime when the Input System tries to read a Vector2 from a Control that can deliver only a float.

Composite Bindings (that is, Bindings that are made up of other Bindings) solve this problem. Composites themselves don't bind directly to Controls; instead, they source values from other Bindings that do, and then synthesize input on the fly from those values.

To see how to create Composites in the editor UI, see documentation on editing Composite Bindings.

To create composites in code, you can use the AddCompositeBinding syntax.

myAction.AddCompositeBinding("Axis")
    .With("Positive", "<Gamepad>/rightTrigger")
    .With("Negative", "<Gamepad>/leftTrigger");

Each Composite consists of one Binding that has InputBinding.isComposiste set to true, followed by one or more Bindings that have InputBinding.isPartOfComposiste set to true. In other words, several consecutive entries in InputActionMap.bindings or InputAction.bindings together form a Composite.

Composites can have parameters, just like Interactions and Processors.

myAction.AddCompositeBinding("Axis(whichSideWins=1)");

There are currently four Composite types that come with the system out of the box: 1D-Axis, 2D-Vector, Button With One Modifier and Button With Two Modifiers.

1D axis

A Composite made of two buttons: one that pulls a 1D axis in its negative direction, and another that pulls it in its positive direction. Implemented in the AxisComposite class. The result is a float.

myAction.AddCompositeBinding("1DAxis") // Or just "Axis"
    .With("Positive", "<Gamepad>/rightTrigger")
    .With("Negative", "<Gamepad>/leftTrigger");

The axis Composite has two part bindings.

Part	Type	Description
`positive`	`Button`	Controls pulling in the positive direction (towards `maxValue`).
`negative`	`Button`	Controls pulling in the negative direction, (towards `minValue`).

You can set the following parameters on an axis Composite:

Parameter	Description
`whichSideWins`	What happens if both `positive` and `negative` are actuated. See table below.
`minValue`	The value returned if the `negative` side is actuated. Default is -1.
`maxValue`	The value returned if the `positive` side is actuated. Default is 1.

If Controls from both the positive and the negative side are actuated, then the resulting value of the axis Composite depends on the whichSideWin parameter setting.

`WhichSideWins`	Description
(0) `Neither`	Neither side has precedence. The Composite returns the midpoint between `minValue` and `maxValue` as a result. At their default settings, this is 0. This is the default value for this setting.
(1) `Positive`	The positive side has precedence and the Composite returns `maxValue`.
(2) `Negative`	The negative side has precedence and the Composite returns `minValue`.

2D vector

A Composite that represents a 4-way button setup like the D-pad on gamepads. Each button represents a cardinal direction. Implemented in the Vector2Composite class. The result is a Vector2.

This Composite is most useful for representing up-down-left-right controls, such as WASD keyboard input.

myAction.AddCompositeBinding("2DVector") // Or "Dpad"
    .With("Up", "<Keyboard>/w")
    .With("Down", "<Keyboard>/s")
    .With("Left", "<Keyboard>/a")
    .With("Right", "<Keyboard>/d");

// To set mode (2=analog, 1=digital, 0=digitalNormalized):
myAction.AddCompositeBinding("2DVector(mode=2)")
    .With("Up", "<Gamepad>/leftStick/up")
    .With("Down", "<Keyboard>/leftStick/down")
    .With("Left", "<Keyboard>/leftStick/left")
    .With("Right", "<Keyboard>/leftStick/right");

The 2D vector Composite has four part Bindings.

Part	Type	Description
`up`	`Button`	Controls representing `(0,1)` (+Y).
`down`	`Button`	Controls representing `(0,-1)` (-Y).
`left`	`Button`	Controls representing `(-1,0)` (-X).
`right`	`Button`	Controls representing `(1,0)` (+X).

In addition, you can set the following parameters on a 2D vector Composite:

Parameter Description

mode Whether to treat the inputs as digital or as analog controls.

If this is set to Mode.DigitalNormalized, inputs are treated as buttons (off if below defaultButtonPressPoint and on if equal to or greater). Each input is 0 or 1 depending on whether the button is pressed or not. The vector resulting from the up/down/left/right parts is normalized. The result is a diamond-shaped 2D input range.

If this is set to Mode.Digital, the behavior is essentially the same as Mode.DigitalNormalized except that the resulting vector is not normalized.

Finally, if this is set to Mode.Analog, inputs are treated as analog (i.e. full floating-point values) and, other than down and left being inverted, values will be passed through as is.

The default is Mode.DigitalNormalized.

Parameter	Description
`mode`	Whether to treat the inputs as digital or as analog controls. If this is set to `Mode.DigitalNormalized`, inputs are treated as buttons (off if below `defaultButtonPressPoint` and on if equal to or greater). Each input is 0 or 1 depending on whether the button is pressed or not. The vector resulting from the up/down/left/right parts is normalized. The result is a diamond-shaped 2D input range. If this is set to `Mode.Digital`, the behavior is essentially the same as `Mode.DigitalNormalized` except that the resulting vector is not normalized. Finally, if this is set to `Mode.Analog`, inputs are treated as analog (i.e. full floating-point values) and, other than `down` and `left` being inverted, values will be passed through as is. The default is `Mode.DigitalNormalized`.

Button with one modifier

A Composite that requires the user to press the button that triggers the Action while holding down another "modifier" button (for example, to represent keyboard shortcuts such as Shift+1). Implemented in the ButtonWithOneModifier class. The buttons can be on any Device, and can be toggle buttons or full-range buttons such as gamepad triggers.

The result is a float.

myAction.AddCompositeBinding("ButtonWithOneModifier")
    .With("Button", "<Keyboard>/1")
    .With("Modifier", "<Keyboard>/leftCtrl")
    .With("Modifier", "<Keyboard>/rightCtrl");

The button with one modifier Composite has two part Bindings.

Part	Type	Description
`modifier`	`Button`	Modifier that has to be held for `button` to come through. If the user holds any of the buttons bound to the `modifier` at the same time as the button that triggers the action, the Composite assumes the value of the `button` Binding. If the user does not press any button bound to the `modifier`, the Composite has a value of 0.
`button`	`Button`	The button whose value the Composite assumes while the user holds both the button that triggers the action, and the `modifier`.

This Composite has no parameters.

Button with two modifiers

A Composite that requires the user to press the button that triggers the Action while holding down two other "modifier" button (for example, to represent keyboard shortcuts such as Ctrl+Shift+1). Implemented in the ButtonWithTwoModifiers class. The buttons can be on any Device, and can be toggle buttons or full-range buttons such as gamepad triggers.

The result is a float.

myAction.AddCompositeBinding("ButtonWithTwoModifiers")
    .With("Button", "<Keyboard>/1")
    .With("Modifier1", "<Keyboard>/leftCtrl")
    .With("Modifier1", "<Keyboard>/rightCtrl")
    .With("Modifier2", "<Keyboard>/leftShift")
    .With("Modifier2", "<Keyboard>/rightShift");

The button with two modifiers Composite has three part Bindings.

Part	Type	Description
`modifier1`	`Button`	The first modifier the user must hold alongside the button that triggers the action, for `button` to come through. If the user does not press any button bound to the `modifier1`, the composite has a value of 0.
`modifier2`	`Button`	The second modifier the user must hold alongside the button that triggers the action, for `button` to come through. If the user does not press any button bound to the `modifier2`, the composite has a value of 0.
`button`	`Button`	The button whose value the Composite assumes while the user presses the button that triggers the action, `modifier1` and `modifier2` at the same time.

This Composite has no parameters.

Writing custom Composites

You can define new types of Composites, and register them with the API. Unity treats these the same as predefined types, which the Input System internally defines and registers in the same way.

To define a new type of Composite, create a class based on InputBindingComposite<TValue>.

// Use InputBindingComposite<TValue> as a base class for a composite that returns
// values of type TValue.
// NOTE: It is possible to define a composite that returns different kinds of values
//       but doing so requires deriving directly from InputBindingComposite.
#if UNITY_EDITOR
[InitializeOnLoad] // Automatically register in editor.
#endif
// Determine how GetBindingDisplayString() formats the composite by applying
// the  DisplayStringFormat attribute.
[DisplayStringFormat("{firstPart}+{secondPart}")]
public class CustomComposite : InputBindingComposite<float>
{
    // Each part binding is represented as a field of type int and annotated with
    // InputControlAttribute. Setting "layout" restricts the controls that
    // are made available for picking in the UI.
    //
    // On creation, the int value is set to an integer identifier for the binding
    // part. This identifier can read values from InputBindingCompositeContext.
    // See ReadValue() below.
    [InputControl(layout = "Button")]
    public int firstPart;

    [InputControl(layout = "Button")]
    public int secondPart;

    // Any public field that is not annotated with InputControlAttribute is considered
    // a parameter of the composite. This can be set graphically in the UI and also
    // in the data (e.g. "custom(floatParameter=2.0)").
    public float floatParameter;
    public bool boolParameter;

    // This method computes the resulting input value of the composite based
    // on the input from its part bindings.
    public override float ReadValue(ref InputBindingCompositeContext context)
    {
        var firstPartValue = context.ReadValue<float>(firstPart);
        var secondPartValue = context.ReadValue<float>(secondPart);

        //... do some processing and return value
    }

    // This method computes the current actuation of the binding as a whole.
    public override float EvaluateMagnitude(ref InputBindingCompositeContext context)
    {
        // Compute normalized [0..1] magnitude value for current actuation level.
    }

    static CustomComposite()
    {
        // Can give custom name or use default (type name with "Composite" clipped off).
        // Same composite can be registered multiple times with different names to introduce
        // aliases.
        //
        // NOTE: Registering from the static constructor using InitializeOnLoad and
        //       RuntimeInitializeOnLoadMethod is only one way. You can register the
        //       composite from wherever it works best for you. Note, however, that
        //       the registration has to take place before the composite is first used
        //       in a binding. Also, for the composite to show in the editor, it has
        //       to be registered from code that runs in edit mode.
        InputSystem.RegisterBindingComposite<CustomComposite>();
    }

    [RuntimeInitializeOnLoadMethod]
    static void Init() {} // Trigger static constructor.
}

The Composite should now appear in the editor UI when you add a Binding, and you can now use it in scripts.

    myAction.AddCompositeBinding("custom(floatParameter=2.0)")
        .With("firstpart", "<Gamepad>/buttonSouth")
        .With("secondpart", "<Gamepad>/buttonNorth");

To define a custom parameter editor for the Composite, you can derive from InputParameterEditor<TObject>.

#if UNITY_EDITOR
public class CustomParameterEditor : InputParameterEditor<CustomComposite>
{
    public override void OnGUI()
    {
        EditorGUILayout.Label("Custom stuff");
        target.floatParameter = EditorGUILayout.FloatField("Some Parameter", target.floatParameter);
    }
}
#endif

Binding resolution

When the Input System accesses the Controls bound to an Action for the first time, the Action resolves its Bindings to match them to existing Controls on existing Devices. In this process, the Action calls InputSystem.FindControls<>() (filtering for devices assigned to the InputActionMap, if there are any) for the Binding path of each of the Action's bindings. This creates a list of resolved Controls that are now bound to the Action.

Note that a single Binding path can match multiple Controls:

A specific Device path such as <DualShockGamepad>/buttonEast matches the "Circle" button on a PlayStation controller. If you have multiple PlayStation controllers connected, it resolves to the "Circle" button on each of these controllers.
An abstract Device path such as <Gamepad>/buttonEast matches the right action button on any connected gamepad. If you have a PlayStation controller and an Xbox controller connected, it resolves the "Circle" button on the PlayStation controller, and to the "B" button on the Xbox controller.
A Binding path can also contain wildcards, such as <Gamepad>/button*. This matches any Control on any gamepad with a name starting with "button", which matches all the four action buttons on any connected gamepad. A different example: */{Submit} matches any Control tagged with the "Submit" usage on any Device.

To query the Controls that an Action resolves to, you can use InputAction.controls. You can also run this query if the Action is disabled.

Choosing which Devices to use

By default, Actions resolve their Bindings against all Devices present in the Input System (that is, InputSystem.devices). For example, if there are two gamepads present in the system, a Binding to <Gamepad>/buttonSouth picks up both gamepads and allows the Action to be used from either.

You can override this behavior by restricting InputActionAssets or individual InputActionMaps to a specific set of Devices. If you do this, Binding resolution only takes the Controls of the given Devices into account.

    var actionMap = new InputActionMap();

    // Restrict the action map to just the first gamepad.
    actionMap.devices = new[] { Gamepad.all[0] };

Note: InputUser and PlayerInput make use of this facility automatically. They set InputActionMap.devices automatically based on the Devices that are paired to the user.

Disambiguation

If multiple Controls are bound to an Action, the Input System monitors input from each bound Control to feed the Action. The Input System must also define which of the bound controls to use for the value of the action. For example, if you have a Binding to <Gamepad>/leftStick, and you have multiple connected gamepads, the Input System must determine which gamepad's stick provides the input value for the Action.

This Control is the "driving" Control; the Control which is driving the Action. Unity decides which Control is currently driving the Action in a process called disambiguation.

During the disambiguation process, the Input System looks at the value of each Control bound to an Action. If the magnitude of the input from any Control is higher then the magnitude of the Control currently driving the Action, then the Control with the higher magnitude becomes the new Control driving the Action. In the above example of <Gamepad>/leftStick binding to multiple gamepads, the Control driving the Action is the left stick which is actuated the furthest of all the gamepads. You can query which Control is currently driving the Action by checking the InputAction.CallbackContext.control property in an Action callback.

If you don't want your Action to perform disambiguation, you can set your Action type to Pass-Through. Pass-Through Actions skip disambiguation, and changes to any bound Control trigger them. The value of a Pass-Through Action is the value of whichever bound Control changed most recently.

Initial state check

Actions with the type set to Value perform an initial state check when they are first enabled, to check the current state of any bound Control, and to set the Action's value to the highest value of any bound Control.

Actions with the type set to Button don't perform any initial state check, so that only buttons pressed after the Action was enabled have any effect on the Action.

Runtime rebinding

Runtime rebinding allows users of your application to set their own Bindings.

NOTE: To download a sample project which demonstrates how to set up a rebinding user interface with Input System APIs, open the Package Manager, select the Input System Package, and choose the sample project "Rebinding UI" to download.

To allow users to choose their own Bindings, use the InputActionRebindingExtensions.RebindingOperation class. Call the PerformInteractiveRebinding() method on an Action to create a rebinding operation. This operation waits for the Input System to register any input from any Device which matches the Action's expected Control type, then uses InputBinding.overridePath to assign the Control path for that Control to the Action's Bindings. If the user actuates multiple Controls, the rebinding operation chooses the Control with the highest magnitude.

IMPORTANT: You must dispose of InputActionRebindingExtensions.RebindingOperation instances via Dispose(), so that they don't leak memory on the unmanaged memory heap.

    void RemapButtonClicked(InputAction actionToRebind)
    {
        var rebindOperation = actionToRebind
            .PerformInteractiveRebinding().Start();
    }

The InputActionRebindingExtensions.RebindingOperation API is highly configurable to match your needs. For example, you can:

Choose expected Control types (WithExpectedControlType()).
Exclude certain Controls (WithControlsExcluding()).
Set a Control to cancel the operation (WithCancelingThrough()).
Choose which Bindings to apply the operation on if the Action has multiple Bindings (WithTargetBinding(), WithBindingGroup(), WithBindingMask()).

Refer to the scripting API reference for InputActionRebindingExtensions.RebindingOperation for a full overview.

Note that PerformInteractiveRebinding() automatically applies a set of default configurations based on the given action and targeted binding.

Showing current Bindings

It can be useful for the user to know what an Action is currently bound to (taking any potentially active rebindings into account) while rebinding UIs, and for on-screen hints while the app is running. You can use InputBinding.effectivePath to get the currently active path for a Binding (which returns overridePath if set, or otherwise returns path).

The easiest way to retrieve a display string for an action is to call InputActionRebindingExtensions.GetBindingDisplayString which is an extension method for InputAction.

    // Get a binding string for the action as a whole. This takes into account which
    // bindings are currently active and the actual controls bound to the action.
    m_RebindButton.GetComponentInChildren<Text>().text = action.GetBindingDisplayString();

    // Get a binding string for a specific binding on an action by index.
    m_RebindButton.GetComponentInChildren<Text>().text = action.GetBindingDisplayString(1);

    // Look up binding indices with GetBindingIndex.
    var bindingIndex = action.GetBindingIndex(InputBinding.MaskByGroup("Gamepad"));
    m_RebindButton.GetComponentInChildren<Text>().text =
        action.GetBindingDisplayString(bindingIndex);

You can also use this method to replace the text string with images.

    // Call GetBindingDisplayString() such that it also returns information about the
    // name of the device layout and path of the control on the device. This information
    // is useful for reliably associating imagery with individual controls.
    var bindingString = action.GetBindingDisplayString(out deviceLayout, out controlPath);

    // If it's a gamepad, look up an icon for the control.
    Sprite icon = null;
    if (!string.IsNullOrEmpty(deviceLayout)
        && !string.IsNullOrEmpty(controlPath)
        && InputSystem.IsFirstLayoutBasedOnSecond(deviceLayout, "Gamepad"))
    {
        switch (controlPath)
        {
            case "buttonSouth": icon = aButtonIcon; break;
            case "dpad/up": icon = dpadUpIcon; break;
            //...
        }
    }

    // If you have an icon, display it instead of the text.
    var text = m_RebindButton.GetComponentInChildren<Text>();
    var image = m_RebindButton.GetComponentInChildren<Image>();
    if (icon != null)
    {
        // Display icon.
        text.gameObject.SetActive(false);
        image.gameObject.SetActive(true);
        image.sprite = icon;
    }
    else
    {
        // Display text.
        text.gameObject.SetActive(true);
        image.gameObject.SetActive(false);
        text.text = bindingString;
    }

Additionally, each binding has a ToDisplayString method, which you can use to turn individual bindings into display strings. There is also a generic formatting method for control paths, InputControlPath.ToHumanReadableString, which you can use with arbitrary control path strings.

Note that the controls a binding resolves to can change at any time, and the display strings for controls might change dynamically. For example, if the user switches the currently active keyboard layout, the display string for each individual key on the Keyboard might change. If bindings are displayed in the UI, this

Control Schemes

A Binding can belong to any number of Binding groups. Unity stores these on the InputBinding class as a semicolon-separated string in the InputBinding.groups property, and you can use them for any arbitrary grouping of bindings. To enable different sets of binding groups for an InputActionMap or InputActionAsset, you can use the InputActionMap.bindingMask/InputActionAsset.bindingMask property. The Input System uses this to implement the concept of grouping Bindings into different InputControlSchemes.

Control Schemes use Binding groups to map Bindings in an InputActionMap or InputActionAsset to different types of Devices. The PlayerInput class uses these to enable a matching Control Scheme for a new user joining the game, based on the Device they are playing on.