Bevy asset loader

This Bevy plugin reduces boilerplate for handling game assets. The crate offers the derivable AssetCollection trait and can automatically load structs that implement it. Asset collections contain handles to your game assets and are available to your systems as resources after loading.

In most cases you will want to load your asset collections during loading states (think loading screens). During such a state, all assets are loaded and their loading process is observed. Only when asset collections can be build with fully loaded asset handles, the collections are inserted as resources. If you do not want to use a loading state, asset collections can still result in cleaner code and improved maintainability (see the "usage without a loading state" section).

The main branch and the latest release support Bevy version 0.11 (see version table)

Loading states

A loading state is responsible for managing the loading process during a configurable Bevy state (see the cheatbook on states).

If your LoadingState is set up, you can start your game logic from the next state and use the asset collections as resources in your systems. The loading state guarantees that all handles in your collections are fully loaded by the time the next state starts.

app
    .add_state::<GameState>()
    .add_loading_state(
        LoadingState::new(GameState::Loading)
            .continue_to_state(GameState::Next)
    )
    .add_collection_to_loading_state::<_, MyAssets>(GameState::Loading)

Your Bevy state needs to be added to the application before you can add a loading state.

You can add collections to a loading state in multiple places (e.g. in different plugins). All collections added anywhere in your application will be loaded. Important is, that the loading state itself is added to the application before you try to add any collections to it.

Compile time vs. Run time (dynamic) assets

Asset configurations, like their file path or dimensions of sprite sheets, can be given at compile time (through derive macro attributes), or at run time ("Dynamic assets"). The second, allows managing asset configurations as assets. That means you can keep a list of your asset files and their properties in asset files. The main benefit of using dynamic assets is a cleaner split of code and data leading to less recompiles while working on your assets. It also makes your game more approachable for people that want to contribute without touching code.

The derive macro for AssetCollection supports multiple attributes. They configure how the asset is loaded.

The following code sets up a loading state with a collection that has all it's configuration in derive macro attributes:

use bevy::prelude::*;
use bevy_asset_loader::prelude::*;

fn main() {
    App::new()
        .add_state::<GameState>()
        .add_plugins(DefaultPlugins)
        .add_loading_state(
            LoadingState::new(GameState::AssetLoading)
                .continue_to_state(GameState::Next)
        )
        .add_collection_to_loading_state::<_, MyAssets>(GameState::AssetLoading)
        .add_systems(OnEnter(GameState::Next), use_my_assets)
        .run();
}

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(path = "images/player.png")]
    player: Handle<Image>,
    #[asset(path = "walking.ogg")]
    walking: Handle<AudioSource>,
}

fn use_my_assets(_my_assets: Res<MyAssets>) {
    // do something using the asset handles from the resource
}

#[derive(Clone, Eq, PartialEq, Debug, Hash, Default, States)]
enum GameState {
    #[default]
    AssetLoading,
    Next,
}

The full_collection example showcases all the different kinds of fields that an asset collection can contain using only derive macro attributes.

Dynamic assets

Dynamic assets are configured through the derive macro attribute key and are not allowed to have a path or paths attribute:

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct ImageAssets {
  #[asset(key = "player")]
  player: Handle<Image>,
  #[asset(key = "tree")]
  tree: Handle<Image>,
}

The keys player and tree in the example above should either be set manually in the DynamicAssets resource prior to the loading state (see the manual_dynamic_asset example), or be part of a dynamic assets file (see dynamic_asset). A dynamic assets file for the collection above might look like this:

({
    "player": File (
        path: "images/player.png",
    ),
    "tree": File (
        path: "images/tree.png",
    ),
})

Using dynamic assets like File and loading ron files requires the standard_dynamic_assets feature to be enabled.

The file ending is .assets.ron by default, but can be configured via LoadingState::set_standard_dynamic_asset_collection_file_endings.

Dynamic assets can be optional. This requires the derive attribute optional on the field and the type to be an Option. The value of the field will be None in case the given key cannot be resolved at run time.

The example full_dynamic_collection shows all supported field types for dynamic assets. Note that adding a dynamic asset file to a loading state requires the AssetServer resource to be available. In most cases that means that you should add the DefaultPlugins before configuring your loading state.

Custom dynamic assets

You can define your own types to load as dynamic assets. Take a look at the custom_dynamic_assets.rs example for some code.

Supported asset fields

The simplest field is of the type Handle<T> and is loaded from a single file without any special processing. One example might be audio sources, but any asset type that has an asset loader registered with Bevy can be used like this.

The field should only have the path attribute set. The path is relative to your assets directory.

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(path = "my-background.ogg")]
    background: Handle<AudioSource>,
}

The dynamic version of the same collection looks like this:

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(key = "background")]
    background: Handle<AudioSource>,
}

({
    "background": File (
        path: "my-background.ogg",
    ),
})

The following sections describe more types of asset fields that you can load through asset collections.

Collections

Folders

This asset field type is not supported in web builds. See Files for a web compatible way of loading a collection of files.

You can load all files in a folder as a vector of untyped handles. This field requires the additional derive macro attribute collection:

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(path = "images", collection)]
    folder: Vec<HandleUntyped>,
}

Just like Bevy's load_folder, this will also recursively load sub folders.

If all assets in the folder have the same (known) type, you can load the folder as Vec<Handle<T>> by setting typed in the collection attribute. Don't forget to adapt the type of the struct field:

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(path = "images", collection(typed))]
    folder: Vec<Handle<Image>>,
}

Folders are also supported as a dynamic asset. The path attribute is replaced by the key attribute:

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(key = "my.images", collection(typed))]
    images: Vec<Handle<Image>>,
}

({
    "my.images": Folder (
        path: "images",
    ),
})

Loading folders is not supported for web builds. If you want to be compatible with Wasm, load you handles from a list of paths instead (see next section).

List of paths

If you want to load a list of asset files with the same type into a vector of Handle<T>, you can list their paths in an attribute:

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(paths("images/player.png", "images/tree.png"), collection(typed))]
    files_typed: Vec<Handle<Image>>,
}

In case you do not know their types, or they might have different types, the handles can also be untyped:

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(paths("images/player.png", "sound/background.ogg"), collection)]
    files_untyped: Vec<HandleUntyped>,
}

As dynamic assets, these two fields replace their paths attribute with key. This is the same as for folders.

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(key = "files_untyped", collection)]
    files_untyped: Vec<HandleUntyped>,
    #[asset(key = "files_typed", collection(typed))]
    files_typed: Vec<Handle<Image>>,
}

The corresponding assets file differs from the folder example:

({
    "files_untyped": Files (
        paths: ["images/tree.png", "images/player.png"],
    ),
    "files_typed": Files (
        paths: ["images/tree.png", "images/player.png"],
    ),
})

Collections as maps

Collections can be loaded as maps using their file paths as the keys. This is only a change in derive attributes and asset field type. Some examples from the sections above would look like this:

use bevy::prelude::*;
use bevy::utils::HashMap;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(path = "images", collection(mapped))]
    folder: HashMap<String, HandleUntyped>,
    #[asset(paths("images/player.png", "images/tree.png"), collection(typed, mapped))]
    files_typed: HashMap<String, Handle<Image>>,
    #[asset(key = "files_untyped", collection(mapped))]
    dynamic_files_untyped: HashMap<String, HandleUntyped>,
    #[asset(key = "files_typed", collection(typed, mapped))]
    dynamic_files_typed: HashMap<String, Handle<Image>>,
}

Standard materials

You can directly load standard materials if you enable the feature 3d. For a complete example please take a look at standard_material.rs.

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(standard_material)]
    #[asset(path = "images/player.png")]
    player: Handle<StandardMaterial>,
}

This is also supported as a dynamic asset:

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(key = "image.player")]
    player: Handle<StandardMaterial>,
}

({
    "image.player": StandardMaterial (
        path: "images/player.png",
    ),
})

Texture atlases

You can directly load texture atlases from sprite sheets if you enable the feature 2d. For a complete example please take a look at atlas_from_grid.rs.

use bevy::prelude::*;
use bevy_asset_loader::asset_collection::AssetCollection;

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(texture_atlas(tile_size_x = 64., tile_size_y = 64., columns = 8, rows = 1, padding_x = 12., padding_y = 12., offset_x = 6., offset_y = 6.))]
    #[asset(path = "images/sprite_sheet.png")]
    sprite: Handle<TextureAtlas>,
}

As a dynamic asset this example becomes:

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(key = "image.player")]
    sprite: Handle<TextureAtlas>,
}

({
    "image.player": TextureAtlas (
        path: "images/sprite_sheet.png",
        tile_size_x: 100.,
        tile_size_y: 64.,
        columns: 8,
        rows: 1,
        padding_x: 12.,
        padding_y: 12.,
        offset_x: 6.,
        offset_y: 6.,
    ),
})

The four padding & offset fields/attributes are optional, and default to 0..

Types implementing FromWorld

Any field in an asset collection without any attribute is required to implement the FromWorld trait. When the asset collection is build, the FromWorld implementation is called to get the value for the field.

Initializing FromWorld resources

In situations where you would like to prepare other resources based on your loaded asset collections you can use App::init_resource_after_loading_state to initialize FromWorld resources. See init_resource.rs for an example that loads two images and then combines their pixel data into a third image.

App::init_resource_after_loading_state does the same as Bevy's App::init_resource, but at a different point in time. While Bevy inserts your resources at the very beginning, bevy_asset_loader will initialize them only after your loaded asset collections are inserted. That means you can use your asset collections in the FromWorld implementation.

Progress tracking

With the feature progress_tracking, you can integrate with iyes_progress to track asset loading during a loading state. This, for example, enables progress bars.

See progress_tracking for a complete example.

A note on system ordering

The loading state runs in a base set between CoreSet::StateTransitions and CoreSet::Update. This means that systems running in CoreSet::Update can already see the reported progress of all tracked asset collections for the current frame.

Failure state

You can configure a failure state in case some asset in a collection fails to load by calling on_failure_continue_to with a state (see failure_state example). If no failure state is configured and some asset fails to load, your application will be stuck in the loading state.

In most cases this happens, an asset file is missing or a certain file ending does not have a corresponding asset loader. In both of these cases the application log should help since Bevy prints warnings about those issues.

Usage without a loading state

Although the pattern of a loading state is quite nice, you might have reasons not to use it. In this case bevy_asset_loader can still be helpful. Deriving AssetCollection on a resource can significantly reduce the boilerplate for managing assets.

You can directly initialise asset collections on the bevy App or World. See no_loading_state.rs for a complete example.

use bevy::prelude::*;
use bevy_asset_loader::prelude::*;

fn main() {
    App::new()
        .add_plugins(DefaultPlugins)
        .init_collection::<MyAssets>()
        .run();
}

#[derive(AssetCollection, Resource)]
struct MyAssets {
    #[asset(texture_atlas(tile_size_x = 100., tile_size_y = 96., columns = 8, rows = 1, padding_x = 12., padding_y = 12.))]
    #[asset(path = "images/sprite_sheet.png")]
    sprite: Handle<TextureAtlas>,
}

Unloading assets

Bevy unloads an asset when there are no strong asset handles left pointing to the asset. An AssetCollection stores strong handles and ensures that assets contained in it are not removed from memory. If you want to unload assets, you need to remove any AssetCollection resource that holds handles pointing to those assets. You, for example, could do this when leaving the state that needed the collection.

Compatible Bevy versions

The main branch is compatible with the latest Bevy release, while the branch bevy_main tries to track the main branch of Bevy (PRs updating the tracked commit are welcome).

Compatibility of bevy_asset_loader versions:

License

Dual-licensed under either of

• Apache License, Version 2.0, (LICENSE-APACHE or https://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or https://opensource.org/licenses/MIT)

at your option.

Assets in the examples might be distributed under different terms. See the readme in the bevy_asset_loader/examples directory.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.