Camera component

Camera-as-a-component is the new way of structuring a game, an approach that allows more flexibility in placing the camera, or even having more than one camera simultaneously.

In order to understand how this approach works, imagine that your game world is an entity that exists somewhere independently from your application. Imagine that your game is merely a window through which you can look into that world. That you can close that window at any moment, and the game world would still be there. Or, on the contrary, you can open multiple windows that all look at the same world (or different worlds) at the same time.

With this mindset, we can now understand how camera-as-a-component works.

First, there is the World class, which contains all components that are inside your game world. The World component can be mounted anywhere, for example at the root of your game class, like the built-in World is.

Then, a CameraComponent class that “looks at” the World. The CameraComponent has a Viewport and a Viewfinder inside, allowing both the flexibility of rendering the world at any place on the screen, and also control the viewing location and angle. The CameraComponent also contains a backdrop component which is statically rendered below the world.


This component should be used to host all other components that comprise your game world. The main property of the World class is that it does not render through traditional means – instead it is rendered by one or more CameraComponents to “look at” the world. In the FlameGame class there is one World called world which is added by default and paired together with the default CameraComponent called camera.

A game can have multiple World instances that can be rendered either at the same time, or at different times. For example, if you have two worlds A and B and a single camera, then switching that camera’s target from A to B will instantaneously switch the view to world B without having to unmount A and then mount B.

Just like with most Components, children can be added to World by using the children argument in its constructor, or by using the add or addAll methods.

For many games you want to extend the world and create your logic in there, such a game structure could look like this:

void main() {
  runApp(GameWidget(FlameGame(world: MyWorld())));

class MyWorld extends World {
  Future<void> onLoad() async {
    // Load all the assets that are needed in this world
    // and add components etc.


This is a component through which a World is rendered. It requires a reference to a World instance during construction; however later the target world can be replaced with another one. Multiple cameras can observe the same world at the same time.

There is a default CameraComponent called camera on the FlameGame class which is paired together with the default world, so you don’t need to create or add your own CameraComponent if your game doesn’t need to.

A CameraComponent has two other components inside: a Viewport and a Viewfinder. Unlike the World object, the camera owns the viewport and the viewfinder, which means those components are children of the camera.

There is also a static property CameraComponent.currentCamera which is not null only during the rendering stage, and it returns the camera object that currently performs rendering. This is needed only for certain advanced use cases where the rendering of a component depends on the camera settings. For example, some components may decide to skip rendering themselves and their children if they are outside of the camera’s viewport.

The FlameGame class has a camera field in its constructor, so you can set what type of default camera that you want like this for example:

void main() {
        camera: CameraComponent.withFixedResolution(width: 800, height: 600),


This factory constructor will let you pretend that the user’s device has a fixed resolution of your choice. For example:

final camera = CameraComponent.withFixedResolution(
  world: myWorldComponent,
  width: 800,
  height: 600,

This will create a camera with a viewport centered in the middle of the screen, taking as much space as possible while still maintaining the 800:600 aspect ratio, and showing a game world region of size 800 x 600.

A “fixed resolution” is very simple to work with, but it will underutilize the user’s available screen space, unless their device happens to have the same pixel ratio as your chosen dimensions.


The Viewport is a window through which the World is seen. That window has a certain size, shape, and position on the screen. There are multiple kinds of viewports available, and you can always implement your own.

The Viewport is a component, which means you can add other components to it. These children components will be affected by the viewport’s position, but not by its clip mask. Thus, if a viewport is a “window” into the game world, then its children are things that you can put on top of the window.

Adding elements to the viewport is a convenient way to implement “HUD” components.

The following viewports are available:

  • MaxViewport (default) – this viewport expands to the maximum size allowed by the game, i.e. it will be equal to the size of the game canvas.

  • FixedResolutionViewport – keeps the resolution and aspect ratio fixed, with black bars on the sides if it doesn’t match the aspect ratio.

  • FixedSizeViewport – a simple rectangular viewport with predefined size.

  • FixedAspectRatioViewport – a rectangular viewport which expands to fit into the game canvas, but preserving its aspect ratio.

  • CircularViewport – a viewport in the shape of a circle, fixed size.

If you add children to the Viewport they will appear as static HUDs in front of the world.


This part of the camera is responsible for knowing which location in the underlying game world we are currently looking at. The Viewfinder also controls the zoom level, and the rotation angle of the view.

The anchor property of the viewfinder allows you to designate which point inside the viewport serves as a “logical center” of the camera. For example, in side-scrolling action games it is common to have the camera focused on the main character who is displayed not in the center of the screen but closer to the lower-left corner. This off-center position would be the “logical center” of the camera, controlled by the viewfinder’s anchor.

If you add children to the Viewfinder they will appear will appear in front of the world, but behind the viewport and with the same transformations as are applied to the world, so these components are not static.

You can also add behavioral components as children to the viewfinder, for example effects or other controllers. If you for example would add a ScaleEffect you would be able to achieve a smooth zoom in your game.


To add static components behind the world you can add them to the backdrop component, or replace the backdrop component. This is for example useful if you want to have a static ParallaxComponent beneath a world that you can move around it.




camera.backdrop = MyStaticBackground();

Camera controls

There are several ways to modify camera’s settings at runtime:

  1. Do it manually. You can always override the CameraComponent.update() method (or the same method on the viewfinder or viewport) and within it change the viewfinder’s position or zoom as you see fit. This approach may be viable in some circumstances, but in general it is not recommended.

  2. Apply effects and/or behaviors to the camera’s Viewfinder or Viewport. The effects and behaviors are special kinds of components whose purpose is to modify over time some property of a component that they attach to.

  3. Use special camera functions such as follow(), moveBy() and moveTo(). Under the hood, this approach uses the same effects/behaviors as in (2).

Camera has several methods for controlling its behavior:

  • Camera.follow() will force the camera to follow the provided target. Optionally you can limit the maximum speed of movement of the camera, or allow it to move horizontally/vertically only.

  • Camera.stop() will undo the effect of the previous call and stop the camera at its current position.

  • Camera.moveBy() can be used to move the camera by the specified offset. If the camera was already following another component or moving towards, those behaviors would be automatically cancelled.

  • Camera.moveTo() can be used to move the camera to the designated point on the world map. If the camera was already following another component or moving towards another point, those behaviors would be automatically cancelled.

  • Camera.setBounds() allows you to add limits to where the camera is allowed to go. These limits are in the form of a Shape, which is commonly a rectangle, but can also be any other shape.


The camera exposes property visibleWorldRect, which is a rect that describes the world’s region which is currently visible through the camera. This region can be used in order to avoid rendering components that are out of view, or updating objects that are far away from the player less frequently.

The visibleWorldRect is a cached property, and it updates automatically whenever the camera moves or the viewport changes its size.

Check if a component is visible from the camera point of view

The CameraComponent has a method called canSee which can be used to check if a component is visible from the camera point of view. This is useful for example to cull components that are not in view.

if (!camera.canSee(component)) {
   component.removeFromParent(); // Cull the component