react-native-babylonjs
v1.1.2
Published
React Native for Babylon.js
Downloads
61
Maintainers
Readme
react-native-babylonjs
There are 3 ways to use babylonjs with react-native
- Official @babylonjs/react-native
react-babylonjs
+@babylonjs/react-native
as described in https://github.com/brianzinn/react-babylonjs#react-nativereact-native-babylonjs
+@flyskywhy/react-native-gcanvas
Performance: @babylonjs/react-native
with JSI should be better.
MultiView: react-native-babylonjs
can open many View
, while @babylonjs/react-native said only one EngineView can be active
.
Convenient: With react-native-babylonjs
, the react-babylonjs multiple touches demo Drag 'n' Drop just ported a few codes to react -> react-native, here is the result.
Install react-native-babylonjs
npm install react-native-babylonjs @flyskywhy/react-native-gcanvas @babylonjs/core @babylonjs/gui
If your textures are stored in APP not on Web, you should also npm install react-native-fs [email protected]
and write codes ref to this commit react -> react-native: babylonjs Non-Declarative
works well on Android with new version @flyskywhy/react-native-gcanvas support 3d game engine babylonjs.
Use react-native-babylonjs
Ref to react-babylonjs
described below and babylonjs examples in https://github.com/flyskywhy/GCanvasRNExamples.
react-native-babylonjs
is touchable by default, If don't want it be touchable, need set props touchActionNone={true}
e.g. <Engine touchActionNone={true} />
.
react-native-babylonjs
support .glb
not .gltf
.
import dataUri from 'data-uri.macro';
import {SceneLoader} from '@babylonjs/core';
import '@babylonjs/loaders/glTF';
const christmasTreeFile = dataUri('../images/LightTree/christmas-tree.glb');
onSceneMount = async (e) => {
const {canvas, scene} = e;
...
SceneLoader.ImportMeshAsync(
'',
// 'https://raw.githubusercontent.com/orbitgw/Chirstmas-Tree/master/src/model/',
// 'christmas-tree.glb',
// use above or below
'',
christmasTreeFile,
scene,
).then((result) => {
const christmasTree = result.meshes[1];
christmasTree.position.set(0, 250, 0);
christmasTree.scalingDeterminant = 25;
const standardMaterial = new StandardMaterial('redMat', scene);
standardMaterial.diffuseColor = new Color4(0.4, 0.4, 0.4, 1);
standardMaterial.specularColor = new Color4(0.4, 0.4, 0.4, 1);
standardMaterial.emissiveColor = new Color4(0.0, 0.16, 0.0, 1);
christmasTree.material = standardMaterial;
});
};
and need below hack to avoid Cannot create URL for blob
sed -i -e "s/this.skipMaterials = false/this.skipMaterials = true/" node_modules/@babylonjs/loaders/glTF/glTFFileLoader.js
Feature of react-native-babylonjs
Not support Post Processes yet.
Memo when debug Post Processes
On Android, if uncomment pipeline.depthOfFieldEnabled = true;
in /src/nonDeclarative.js
,will got black screen (and got glGetError()=GL_INVALID_FRAMEBUFFER_OPERATION after glDrawElements(GL_TRIANGLES, 6, GL_UNSIGNED_SHORT, 0)), then if put if(postProcess.name === 'circleOfConfusion')
before this._postProcesses.push(postProcess);
in @babylonjs v5.1.0/core/Cameras/camera.js
, will got pink screen. This maybe bug of@babylonjs v5.1.0
ref to https://developer.vuforia.com/forum/unity/black-screen-android-8010
, or maybe Mali GPU ref to https://stackoverflow.com/questions/46483852/gl-invalid-framebuffer-operation-framebuffer-is-not-complete-or-incompatible-wi
, or maybe WebGL/OpenGLES version ref to https://forum.unity.com/threads/webgl-on-android-only-works-in-fullscreen.914435/
, or maybe gl.getExtension("OES_texture_float")
is false so that ThinEngine.prototype._canRenderToFramebuffer
is false, or maybe FrameBuffer size should be power of 2 (I tried to set GCanvasView to 256x256
and change this._gl.drawingBufferWidth
to this._gl.canvas.width
in @babylonjs v5.1.0/core/Engines/thinEngine.js
so that width, height
in ThinEngine.prototype._setupFramebufferDepthAttachments
will be 256, 256
but no effect, and needPOTTextures in @babylonjs v5.1.0/core/Engines/thinEngine.js
should be true, and if not change this._gl.drawingBufferWidth
to this._gl.canvas.width
will GL_INVALID_FRAMEBUFFER_OPERATION after glDrawElements), or cause by WebGL can GL_DEPTH_STENCIL
but OpenGL ES will getError so use GL_DEPTH24_STENCIL8
instead by renderbufferStorage()
in https://github.com/flyskywhy/react-native-gcanvas/blob/master/packages/gcanvas/src/context/webgl/RenderingContext.js , or since https://blenderartists.org/t/purple-material-texture-is-driving-me-nuts/1371196/9 and https://stackoverflow.com/questions/72931171/in-unity-custom-shader-materials-appear-pink said pink generally means the material (or corresponding shader) has problem, maybe need test glFramebufferTexture2D in @flyskywhy/react-native-gcanvas
, I have no clue, so just not use post processes.
Memo when debug babylonjs
if (!reflectionTexture.isReadyOrNotBlocking())
in @babylonjs v5.1.0/core/Materials/PBR/pbrBaseMaterial.js
means whether url
in CubeTexture.CreateFromPrefilteredData()
is downloaded.
_useFinalCode()
in @babylonjs v5.1.0/core/Materials/effect.js
shows the shaders name that babylonjs used.
_this._hdr
in @babylonjs v5.1.0/core/PostProcesses/RenderPipeline/Pipelines/defaultRenderingPipeline.js
, on Web is true, on Android is false, thus cause there is no 'imageProcessing' when enable some Post Process on Android ref to DefaultRenderingPipeline.prototype._buildPipeline
.
PostProcessManager.prototype._prepareFrame
and PostProcessManager.prototype._finalizeFrame
in @babylonjs v5.1.0/core/PostProcesses/postProcessManager.js
are called by Scene.prototype._renderForCamera
in @babylonjs v5.1.0/core/scene.js
to process camera as targetTexture?
Forked from react-babylonjs, below is its README.
react-babylonjs
'react-babylonjs' integrates the Babylon.js real time 3D engine with React
react-babylonjs
lets you build your scene and components using a familiar
declarative syntax with the benefits of reusable components and hooks. The
Babylon.js API is mostly covered declaratively thanks to code generation and
even custom props allow you to declaratively add shadows, physics, 3D models,
attach 2D/3D UI to meshes, etc.
Fully supports hooks. Full support for TypeScript with auto-completion on elements and compile time checks. Context API and hooks provide easy access to Scene/Engine/Canvas.
How to Install
$ npm i react-babylonjs @babylonjs/core @babylonjs/gui
OR
$ yarn add react-babylonjs @babylonjs/core @babylonjs/gui
No third party dependencies outside of React + babylon.js If you are upgrading from 2.x please follow the breaking changes guide:
Models
If you are using 3D models ensure you have added the @babylonjs/loaders
NPM.
It is not a direct dependency, but registers loaders as plugins via imports with
side effects:
- Register all model types
import @babylonjs/loaders;
- OBJ
import '@babylonjs/loaders/OBJ';
- glTF
import '@babylonjs/loaders/glTF';
(more instructions on model loading in ES6 here )
Usage Styles
react-babylonjs
tries to remain unopinionated about how you integrate
BabylonJS with React. This module provides a 100% declarative option and/or you
can customise by adding code. There are lots of escape hatches where you can
switch to imperative coding and direct access to objects.
Connecting the pieces
If you are new to React or babylon.js (or both) there is some learning ahead. The babylon.js documentation site is really useful for understanding the basics of lighting, cameras, etc. This project aims to make easy to integrate those into React using JSX.
Here we re-use a SpinningBox
component that can be clicked or hovered. These
reusable components can be used to compose a declarative scene. We are using
hooks for the clicking, hovering and spinning.
import React, { useRef, useState } from 'react'
import {
Engine,
Scene,
useBeforeRender,
useClick,
useHover,
} from 'react-babylonjs'
import { Vector3, Color3 } from '@babylonjs/core'
const DefaultScale = new Vector3(1, 1, 1)
const BiggerScale = new Vector3(1.25, 1.25, 1.25)
const SpinningBox = (props) => {
// access Babylon scene objects with same React hook as regular DOM elements
const boxRef = useRef(null)
const [clicked, setClicked] = useState(false)
useClick(() => setClicked((clicked) => !clicked), boxRef)
const [hovered, setHovered] = useState(false)
useHover(
() => setHovered(true),
() => setHovered(false),
boxRef
)
// This will rotate the box on every Babylon frame.
const rpm = 5
useBeforeRender((scene) => {
if (boxRef.current) {
// Delta time smoothes the animation.
var deltaTimeInMillis = scene.getEngine().getDeltaTime()
boxRef.current.rotation.y +=
(rpm / 60) * Math.PI * 2 * (deltaTimeInMillis / 1000)
}
})
return (
<box
name={props.name}
ref={boxRef}
size={2}
position={props.position}
scaling={clicked ? BiggerScale : DefaultScale}
>
<standardMaterial
name={`${props.name}-mat`}
diffuseColor={hovered ? props.hoveredColor : props.color}
specularColor={Color3.Black()}
/>
</box>
)
}
export const SceneWithSpinningBoxes = () => (
<div>
<Engine antialias adaptToDeviceRatio canvasId="babylonJS">
<Scene>
<arcRotateCamera
name="camera1"
target={Vector3.Zero()}
alpha={Math.PI / 2}
beta={Math.PI / 4}
radius={8}
/>
<hemisphericLight
name="light1"
intensity={0.7}
direction={Vector3.Up()}
/>
<SpinningBox
name="left"
position={new Vector3(-2, 0, 0)}
color={Color3.FromHexString('#EEB5EB')}
hoveredColor={Color3.FromHexString('#C26DBC')}
/>
<SpinningBox
name="right"
position={new Vector3(2, 0, 0)}
color={Color3.FromHexString('#C8F4F9')}
hoveredColor={Color3.FromHexString('#3CACAE')}
/>
</Scene>
</Engine>
</div>
)
Hooks, Shadows and Physics (and optionally TypeScript, too)
You can declaratively use many features together - here only the button click handler actually has any code - and we have declarative Physics, GUI, Lighting and Shadows. demo: Bouncy demo
import React, { useRef } from 'react';
// full code at https://github.com/brianzinn/create-react-app-typescript-babylonjs
const App: React.FC = () => {
let sphereRef = useRef<Nullable<Mesh>>();
const onButtonClicked = () => {
if (sphereRef.current) {
sphereRef.current.physicsImpostor!.applyImpulse(
Vector3.Up().scale(10),
sphereRef.current.getAbsolutePosition()
);
}
};
return (
<Engine antialias={true} adaptToDeviceRatio={true} canvasId="sample-canvas">
<Scene enablePhysics={[gravityVector, new CannonJSPlugin()]}>
<arcRotateCamera name="arc" target={ new Vector3(0, 1, 0) }
alpha={-Math.PI / 2} beta={(0.5 + (Math.PI / 4))}
radius={4} minZ={0.001} wheelPrecision={50}
lowerRadiusLimit={8} upperRadiusLimit={20} upperBetaLimit={Math.PI / 2}
/>
<hemisphericLight name='hemi' direction={new Vector3(0, -1, 0)} intensity={0.8} />
<directionalLight name="shadow-light" setDirectionToTarget={[Vector3.Zero()]} direction={Vector3.Zero()} position = {new Vector3(-40, 30, -40)}
intensity={0.4} shadowMinZ={1} shadowMaxZ={2500}>
<shadowGenerator mapSize={1024} useBlurExponentialShadowMap={true} blurKernel={32}
shadowCasters={["sphere1", "dialog"]} forceBackFacesOnly={true} depthScale={100}
/>
</directionalLight>
<sphere ref={sphereRef} name="sphere1" diameter={2} segments={16} position={new Vector3(0, 2.5, 0)}>
<physicsImpostor type={PhysicsImpostor.SphereImpostor} _options={{
mass: 1,
restitution: 0.9
}} />
<plane name="dialog" size={2} position={new Vector3(0, 1.5, 0)}>
<advancedDynamicTexture name="dialogTexture" height={1024} width={1024} createForParentMesh={true} hasAlpha={true}>
<rectangle name="rect-1" height={0.5} width={1} thickness={12} cornerRadius={12}>
<rectangle>
<babylon-button name="close-icon" background="green" onPointerDownObservable={onButtonClicked}>
<textBlock text={'\uf00d click me'} fontFamily="FontAwesome" fontStyle="bold" fontSize={200} color="white" />
</babylon-button>
</rectangle>
</rectangle>
</advancedDynamicTexture>
</plane>
</sphere>
<ground name="ground1" width={10} height={10} subdivisions={2} receiveShadows={true}>
<physicsImpostor type={PhysicsImpostor.BoxImpostor} _options={{
mass: 0,
restitution: 0.9
}} />
</ground>
<vrExperienceHelper webVROptions={{ createDeviceOrientationCamera: false }} enableInteractions={true} />
</Scene>
</Engine>
);
}
Developer Experience and Fast Refresh
With declarative (TSX/JSX) coding and fast refresh, you experience the same development workflow in 3D - ie: save changes in your editor and see them immediately in the browser. Note in this capture when the light is dimmed that the state changes persist even after code updates and scene refresh.
API
This project uses code generation, which allows fast reconciliation and excellent typings support.
Release History and changes
Storybook
~50 sample web pages with viewable source code on github pages.
Example Projects
- Create React App (JavaScript) CRA JavaScript implementation. Github pages has live examples of typical and declarative usage some with Redux.
- Create React App (TypeScript) CRA 3 TypeScript. Examples of physics, shadows, etc.
Contributors
- Huge shout out to Konsumer that helped bring this project to the next level. The ideas and code sandboxes from issue #6 inspired the code generation and HOC + Context API integration.
- seacloud9 for adding storybook, GSAP demo, dynamic terrain (extension) & PIXI demo.
- hookex has made the largest contribution :)
Proper texture handling
demo,
Node parenting,
demo
Full Screen GUI
demo,
Effect Layers
glow demo,
behaviors
demo,
useHover & useClick hooks
demo
and react-spring integration
demo.
Author of
react-babylon-spring
- https://github.com/hookex/react-babylon-spring. - dennemark add support for
CascadedShadowGenerator
,createPortal
andHtml
(project HTML in scene - ported fromdrei
). Also, stories for Snippet Material and "tunnel" zustand integration. - kencyke created a cool multi-canvas + cloud
point repo that insipired creation of
<pointsCloudSystem .../>
as host element. - flostellbrink fixed the
react-babylon-spring
integration and added GH action for Storybook (github pages)
Thanks also to all the people who have contributed with issues/questions/discussions. All the great ideas and requests are why this project evolved beyond an experiment.
Made with ♥