---
title: "Input and Cameras"
description: "Map physical controls to actions and configure 2D and 3D cameras."
canonical: https://docs.nodetool.ai/game-engine/input-cameras
markdown: https://docs.nodetool.ai/game-engine/input-cameras.md
product: NodeTool
source: https://github.com/nodetool-ai/nodetool/blob/main/docs/game-engine/input-cameras.md
---

# Input and Cameras

Declare what the game understands as actions and, in 3D, axes. Physical keyboard, mouse, gamepad, and touch bindings turn into the same input frames used by simulation and replay.

![Signal Forge HUD showing WASD movement, Space jump, E interaction, R recovery, and drag look](/assets/game-engine/signal-forge.png)

*Signal Forge's displayed controls include its authored E action. Document bindings and on-screen instructions should agree. [Capture source](/game-engine/screenshots).*

## Bind an action

Declare an action in `inputActions`, then add bindings under `inputBindings.actions`. This document fragment binds `interact` to physical E and a touch button:

```json
{
  "inputActions": ["interact"],
  "inputBindings": {
    "actions": {
      "interact": [{"kind": "key", "code": "KeyE"},
                   {"kind": "touchButton", "label": "Use"}]
    }
  }
}
```

Keep the rest of the game's declared actions when applying this fragment. `key` matches physical key position. `keyValue` follows the keyboard's character value. Actions also accept mouse buttons, gamepad buttons, thresholded gamepad axes, and touch-stick directions.

A listed action uses exactly its listed bindings. An empty array unbinds it. An omitted action retains its name-based defaults. In 2D, arrows and WASD map directions and Space maps `space`. In 3D, defaults include Space for `jump`, R for `respawn`, and mouse button 0 or F for `fire`.

## Add axes and look

3D axes use key pairs, gamepad axes or buttons, and touch-stick axes. When several bindings contribute, the largest magnitude wins. Default `moveX` and `moveZ` use WASD, arrows, and gamepad/touch movement.

Look accepts mouse motion, a gamepad stick, or touch drag, with sensitivity and inversion controls. Click the 3D player to capture the mouse. Input pressed while paused is discarded. Player rebinding is not persisted automatically.

## Frame the scene

A 2D camera defines width, height, and zoom. Its world dimensions combine with pixels-per-unit to determine HUD coordinates.

A 3D camera uses perspective field of view or orthographic size, plus near and far planes. A fixed camera follows its authored transform. A follow camera names a target and provides position and look-at offsets, yaw, pitch limits, sensitivity, and collision radius. Keep near/far values appropriate to the scene to avoid clipping and depth precision problems.

The editor orbit camera is an inspection tool. The scene's active camera controls play and capture. Verify framing with the active camera and exported aspect ratio, not only the editor viewport.