Physically based renderer (PBR) and scene graph for PEX.
This is an experimental API and it's likely to change in the future.
- pex-context modern WebGL wrapper (buffers, textures, pipelines, commands etc)
- pex-math array based math (vec3, mat4, quat etc)
- Usage
- Examples
- API
- Renderer
- Entities
- Components
- Scene Components
- transform, camera, postProcessing, overlay
- Geometry Components
- animation, geometry, material, morph, skin
- Lighting Components
- ambientLight, directionalLight, areaLight, spotLight, skybox, reflectionProbe
- Loaders
- Creating Custom Components
PEX Renderer v3 is currently in beta. You can install the latest version via npm:
npm i pex-renderer@next
This will install v3 with the beta release number after the dash e.g. [email protected].
PEX Renderer is a CommonJS module and you will need a bundler (e.g. Browserify) to run it in the browser.
Open live examples here.
const createContext = require('pex-context')
const createRenderer = require('pex-renderer')
const createSphere = require('primitive-sphere')
const ctx = createContext({ width: 800, height: 600 })
const renderer = createRenderer({
ctx: ctx
})
const camera = renderer.entity([
renderer.transform({ position: [0, 0, 3] }),
renderer.camera({
fov: Math.PI / 2,
aspect: ctx.gl.drawingBufferWidth / ctx.gl.drawingBufferHeight,
near: 0.1,
far: 100
})
])
renderer.add(camera)
const cube = renderer.entity([
renderer.transform({ position: [0, 0, 0] }),
renderer.geometry(createSphere(1)),
renderer.material({
baseColor: [1, 0, 0, 1]
})
])
renderer.add(cube)
const skybox = renderer.entity([
renderer.skybox({
sunPosition: [1, 1, 1]
})
])
renderer.add(skybox)
const reflectionProbe = renderer.entity([renderer.reflectionProbe()])
renderer.add(reflectionProbe)
ctx.frame(() => {
renderer.draw()
})
You can find runnable examples in the /examples
folder in this repository. To run an example install Node.js, clone or download this repository and then:
# go to the example folder
cd examples
# install dependencies
npm install
# start a webpack-dev-server to run all the examples
npm start
Main class responsible for managing scene hierarchy and rendering. You add your entities to the renderer and call draw every frame.
Note: PEX Renderer doesn't currently have a concept of a scene. This can be simulated by creating multiple root entities with their own scene hierarchies and adding / removing them as necessary.
const createRenderer = require('pex-renderer')
const renderer = createRenderer({
ctx,
shadowQuality: 2,
rgbm: false,
profile: false,
pauseOnBlur: true
})
- renderer.paused
- renderer.profiler
property | info | type | default |
---|---|---|---|
ctx |
rendering context | pex-context.Context | null |
shadowQuality |
shadow smoothness | Integer 0-4 | 2 |
rgbm |
use RGBM color packing for rendering pipeline | Boolean | false |
profile |
enable profiling | Boolean | false |
pauseOnBlur |
stop rendering when window looses focus | Boolean | false |
entities * |
list of entities in the scene | Array of Entity | [] |
 _ required  _ read only
function frame() {
renderer.draw()
requestAnimationFrame(frame)
}
requestAnimationFrame(frame)
// or using built-in frame() from pex-context
ctx.frame(() => {
renderer.draw()
})
Updates transforms, shadow-maps, reflection probes, materials, shaders, renders the scene and applies post-processing. Should be called every frame.
Entities are collection of components representing an object in the scene graph.
NOTE: It's worth mentioning that in its current form PEX Renderer doesn't implement Entity-Component-System architecture. Components are self contained and fully functional not merely buckets of data to be processed by a collection of systems. In that regard it's comparable to Unity and its GameObject and MonoBehaviour implementation.
Creates an entity from a list of components.
components
: Array of Component - list of components that the entity is made oftags
- Array of String - list of tags
Note: entities are not added to the scene graph automatically.
Note on tagging: Camera component also accepts tags. Only entities matching one or more camera tags will be rendered. If camera doesn't have any tags only untagged entities will be rendered.
const entity = renderer.entity(
[
renderer.transform({ position: [0, 1, 0] }),
renderer.geometry({ positions: [], normals: [], cells: [] }),
renderer.material({ baseColor: [1, 0, 0, 1] })
],
['opaque', 'debug-only']
)
Adds entity to the scene graph and attaches to a parent as a child.
Removes entity from the scene graph.
Adds component to an entity.
const entity = renderer.entity([renderer.pointLight()])
entity.getComponent('PointLight')
Gets component by it's class name.
type
- upper camel case name of the component class
Removes entity from the scene and disposes all the components and their resources.
Components are bits of functionality (transform, light type, geometry, material etc) that are added to an entity.
property | info | type | default |
---|---|---|---|
type * |
component class name | String | '' |
entity * |
entity the component is attached to | Entity | null |
changed * |
event emitted whenever component's property changes | Signal | null |
* read only
const entity = renderer.entity([renderer.transform()])
function onParamChange(name) {
console.log(`param ${name} has changed`)
}
// start listening
entity.transform.changed.add(onParamChange)
// done internaly by transform whenever position changes
entity.transform.dispatch('position')
// stop listening
entity.transform.changed.remove(onParamChange)
transformComponent.set({
position: [Math.cos(time), 0, 0]
})
- TODO: #75
const transform = renderer.transform({
position: [0, 0, 0],
scale: [1, 1, 1],
rotation: [0, 0, 0, 1]
})
property | info | type | default |
---|---|---|---|
position |
entity position relatively to it's parent | Vec3 / [x, y, z] | [0, 0, 0] |
scale |
entity scale relatively to it's parent | Vec3 / [x, y, z] | [1, 1, 1] |
rotation |
entity rotation relatively to it's parent | Quat / [x, y, z, w] | [0, 0, 0, 1] |
parent |
entity's parent entity | Entity | null |
enabled |
should the entity be rendered | Boolean | true |
children * |
Array of Entity | false | |
bounds * |
|||
worldBounds * |
|||
localModelMatrix * |
|||
modelMatrix * |
* read only
Defines rendering viewport and projection.
Note: camera
position/rotation
are derived from entity.transform.position/rotation
. It's probably easier to use Orbiter
component at the moment.
const camera = renderer.camera({
fov: Math.PI / 4,
aspect: ctx.gl.drawingBufferWidth / ctx.gl.drawingBufferHeight,
near: 0.1,
far: 100
})
property | info | type | default |
---|---|---|---|
projection |
camera projection type | 'perspective' | 'orthographic' | 'perspective' |
viewport |
camera viewport | Array [x, y, width, height] | [0, 0, gl.drawingBufferWidth, gl.drawingBufferHeight] |
near |
near plane distance | Number | 0.1 |
far |
far plane distance | Number | 100 |
aspect |
aspect ratio | Number | 1 |
exposure |
exposure value | Number | 1 |
fov |
perspective vertical field of view (yfov) | Number [rad] | Math.PI / 41 |
focalLength |
focal length of the camera lens [10mm - 200mm] | Number [mm] | 50 |
fStop |
ratio of camera lens opening, f-number, f/N, aperture [1.2 - 32] | Number | 2.8 |
sensorSize |
physical camera sensor or film size [sensorWidth, sensorHeight] | Vec2 [mm, mm] | [36, 24] |
sensorFit |
how camera frame matches sensor frame | 'vertical' | 'horizontal' | 'fit' | 'overscan' | 'vertical' |
left , right , top , bottom |
orthographic frustum bounds | Number | 1 |
zoom |
orthographic zoom | Number | 1 |
projectionMatrix * |
|||
viewMatrix * |
|||
inverseViewMatrix * |
* read only 1 depends on viewport aspect ratio, focalLength and sensorFit
Defines rendering post-processing.
const postProcessing = renderer.postProcessing({
fxaa: true,
ssao: true,
dof: true,
bloom: true
})
property | info | type | default |
---|---|---|---|
fxaa |
FXX antaliasing on/off | Boolean | false |
property | info | type | default |
---|---|---|---|
ssao |
SSAO on/off | Boolean | false |
ssaoIntensity |
SSAO shadows | Number | 5 |
ssaoRadius |
SSAO shadows | Number | 12 |
ssaoBias |
SSAO shadows | Number | 0.01 |
ssaoBlurRadius |
SSAO shadows | Number | 2 |
ssaossaoBlurSharpnessBias |
SSAO shadows | Number | 10 |
property | info | type | default |
---|---|---|---|
dof |
DoF on/off | Boolean | false |
dofFocusDistance |
Distance to focus plane | Number [meters] | 5 |
property | info | type | default |
---|---|---|---|
bloom |
Bloom on/off | Boolean | false |
bloomRadius |
Amount of bloom blur | Number | 1 |
bloomThreshold |
Bloom color cut off (default 1 = only "hdr" colors will bloom) | Number | 1 |
bloomIntensity |
Amount of the bloom to add to the scene | Number | 0.1 |
TODO: fog, fogColor, fogStart, fogDensity, inscatteringCoeffs, sunPosition, sunColor, sunDispertion, sunIntensity
Orbiter controller for camera component.
Note: orbiter actually doesn't modify the camera but the entity's transform therefore both Orbiter and Camera should be attached to the same entity.
const orbiter = renderer.orbiter({
target: [0, 0, 0],
position: [1, 1, 1],
lat: 0,
lon: Math.PI / 2,
easing: 0.1
})
Flat 2D overlay, useful for tex and logos.
const overlay = renderer.overlay({
x: 0,
y: 0,
width: 1,
height: 1,
texture: ctx.Texture
})
Represents 3d mesh geometry attributes.
const geometry = renderer.geometry({
positons: [[0, 0, 1], [1, 2, 3], ...[]],
normals: [[0, 0, 1], [0, 0, 1], ...[]],
uvs: [[0, 0], [0, 1], ...[]],
indices: [[0, 1, 2], [3, 4, 5], ...[]],
offsets: { data: [[0, 0, 0], [0, 1, 0], ...[]], divisor: 1 }
})
property | info | type | default |
---|---|---|---|
positions |
vertex positions | Array of Vec3 [x, y, z] | null |
normals |
vertex normals | Array of Vec3 [x, y, z] | null |
texCoords |
vertex tex coords | Array of Vec2 [u, v] | null |
uvs 1 |
alias of texCoords |
Array of Vec2 [u, v] | null |
colors |
vertex colors | Array of Vec4 [r, g, b, a] | null |
indices |
indices |
Array of Vec3 | null |
cells 1 |
geometry faces | Array of Vec3 of Int [i, j, j] | null |
offsets 2 |
instances offsets | Array of Vec3 [x, y, z] | null |
rotations 2 |
instances rotations | Array of Quat/Vec4 [x, y, z, w] | null |
scales 2 |
instances scales | Array of Vec3 [x, y, z] | null |
tints 2 |
instanced rotations | Array of Color/Vec4 [r, g, b, a] | null |
1 write only aliases, uvs
data will be stored in texCoords
, cells
data will be stored in indices
2 those attributes are always instanced and need to be defined with a divisor and additionally number of instances needs to be specified:
const offsets = [[x, y, z], ...[]]
const g = renderer.geometry({
positions: [[x, y, z], ...[]],
offsets: { data: offsets, divisor: 1 },
instances: offsets.length
})
Physically based material description. Default to a Metallic Roughness workflow but can also use a Specular Glossiness workflow or even be unlit.
const material = renderer.material({
baseColor: [1, 1, 1, 1],
emissiveColor: [0, 0, 0, 1],
metallic: 0.8,
roughness: 0.2,
castShadows: false,
receiveShadows: false,
alphaTest: 0.5,
alphaMap: ctx.Texture2D
})
property | info | type | default |
---|---|---|---|
baseColor |
albedo | Color/Vec4 [r, g, b, a] | [1, 1, 1, 1] |
baseColorMap |
base color texture. Multiplied by baseColor . |
ctx.Texture | TextureMap | null |
unlit |
no lighting / shadowing. Use baseColor . |
Boolean | false |
metallic |
metallic factor. Used if no metallicMap is provided. |
Number | 1 |
metallicMap |
metallic texture. Used if no metallicRoughnessMap is provided. |
ctx.Texture | TextureMap | null |
roughness |
roughness factor. Used if no roughnessMap is provided. |
Number | 1 |
roughnessMap |
roughness texture. Used if no metallicRoughnessMap is provided. |
ctx.Texture | TextureMap | null |
metallicRoughnessMap |
metallic (b channel) and roughness (g channel) combined in a texture. | ctx.Texture | TextureMap | null |
useSpecularGlossinessWorkflow |
use a specular/glossiness PBR workflow instead of above Metallic/Roughness | Boolean | false |
diffuse |
diffuse color. Used if no uDiffuseMap is provided. |
Color/Vec4 [r, g, b, a] | 1 |
diffuseMap |
specular (b channel) and roughness (g channel) combined in a texture. | ctx.Texture | TextureMap | null |
specular |
specular color. Used if no specularGlossinessMap is provided. |
Color/Vec3 [r, g, b] | 1 |
glossiness |
glossiness or smoothness. Used if no specularGlossinessMap is provided. |
Number | 1 |
specularGlossinessMap |
specular and glossiness combined in a texture. | ctx.Texture | TextureMap | null |
normalMap |
normal texture. Doesn't modify vertices positions, only impacts lighting. | ctx.Texture | TextureMap | null |
normalScale |
normal factor. Control how much the normalMap affects lighting. |
Number | 1 |
displacementMap |
displacement texture. Modifies vertices positions (r channel). | ctx.Texture | TextureMap | null |
displacement |
displacement factor. Control how much the displacementMap affects vertices. |
Number | 0 |
emissiveColor |
light emitted | Color/Vec4 [r, g, b, a] | null |
emissiveIntensity |
emissive factor | Number | 1 |
emissiveColorMap |
base color texture. Multiplied by emissiveColor and emissiveIntensity . |
ctx.Texture | TextureMap | null |
occlusionMap |
occlusion texture. Indicates areas of indirect lighting. | ctx.Texture | TextureMap | null |
reflectance |
control specular intensity on non-metallic surfaces. | Number 0-1 | 0.5 |
clearCoat |
strength of the clear coat layer. | Number 0-1 | null |
clearCoatRoughness |
roughness of the clear coat layer. | Number 0-1 | null |
clearCoatNormalMap |
normal texture for the clear coat layer. | ctx.Texture | TextureMap | null |
clearCoatNormalMapScale |
clear coat normal factor. | Number | 1 |
alphaMap |
alpha texture. Impacts opacity (r channel). | ctx.Texture | TextureMap | null |
alphaTest |
value against which to test alpha. | Number 0-1 | true |
depthWrite |
depth write mask | Boolean | true |
depthTest |
depth test on/off | Boolean | true |
depthFunc |
depth test function | ctx.DepthFunc | ctx.DepthFunc.LessEqual |
blend |
blending on/off | Boolean | false |
blendSrcRGBFactor |
blending source color factor | ctx.BlendFactor | ctx.BlendFactor.One |
blendSrcAlphaFactor |
blending source alpha factor | ctx.BlendFactor | ctx.BlendFactor.One |
blendDstRGBFactor |
blending destination color factor | ctx.BlendFactor | ctx.BlendFactor.One |
blendDstAlphaFactor |
blending destination alpha factor | ctx.BlendFactor | ctx.BlendFactor.One |
cullFace |
face culling on/off | Boolean | false |
cullFaceMode |
face culling mode | ctx.Face | ctx.Face.Back |
pointSize |
set gl_PointSize for ctx.Primitive.Points |
Number | 1 |
castShadows |
impact shadow casting | Boolean | false |
receiveShadows |
receive potential shadowing | Boolean | false |
Texture transforms are achieved by optionally passing a TextureMap object with offset, rotation and/or scale alongside the texture itself: { texture: ctx.Texture, offset?: Vec2 [x, y], rotation?: Radians, scale?: Vec2 [x, y] }
_The reflectance value represents a remapping of a percentage of reflectance (with a default of 4%: 0.16 _ pow(0.5, 2) = 0.04) and replaces an explicit index of refraction (IOR)*
Geometry attribute animations based on glTF 2.0 Spec / Animations.
const animation = renderer.animation({
channels: [], // Array of Channels
autoplay: true,
loop: true
})
// TODO
// const Channel = {
// input: null,
// output: null,
// interpolation: null,
// target: null,
// path: null,
// }
Geometry morph targets based on glTF 2.0 Spec / Morph Targets.
const morph = renderer.morph({
sources: { positions, normals, tangents, ...attributes },
targets: { positions, normals, tangents, ...attributes },
weights: [0.0, 0.0, ...weights]
})
Geometry vertex skin based on glTF 2.0 Spec / Skin.
const skin = renderer.skin({
joints: [entity, entity, ...entities],
inverseBindMatrices: [mat4, mat4, ...mat4]
})
Components representing light sources used for rendering of the scene.
Note on position and orientation of lights: Similar as camera light components position and orientation is controlled via transform component of the entity the light is attached to.
const directionalLightEnity = renderer.entity([
renderer.transform({
rotation: quat.fromAxisAngle(quat.create(), [0, 0, 1], Math.PI / 2)
}),
renderer.directionalLight({
color: [1, 1, 1, 1],
intensity: 1,
castShadows: true
})
])
const ambientLight = renderer.ambientLight({
color: [1, 1, 1, 1],
intensity: 1
})
const directionalLight = renderer.directionalLight({
color: [1, 1, 1, 1],
intensity: 1,
castShadows: true
})
Note: directionalLight
direction
is derived from entity.transform.rotation
Rectangular area light.
const areaLight = renderer.areaLight({
color: [1, 1, 1, 1],
intensity: 1
})
Note: areaLight
position/rotation/size
are derived from entity.transform.position/rotation/scale
const spotLight = renderer.spotLight({
color: [1, 1, 1, 1],
intensity: 1
})
Note: spotLight
direction
is derived from entity.transform.rotation
const skybox = renderer.skybox({
sunPosition: [1, 1, 1], // sky gradient used for reflections
texture: ctx.texture2D(), // used for reflections instad of sky
backgroundTexture: ctx.texture2D(), // used for background rendering, not reflections,
backgroundBlur: 0 // if set to 1, blurs texture for background rendering, not reflections
})
Note: By default a sky background is rendered unless hdr equirect panorama texture is provided. Note: Skybox orientation differ from engine to engine; to update it, set the entity's transform component rotation and set any reflection probe to dirty.
Captures environmental map of the scene for Image Based Lighting (IBL) specular reflection and irradiance diffuse. Currently requires Skybox component to be present in the scene as only Skybox background is captured.
const reflectionProbe = renderer.reflectionProbe({})
Note: Due to the cost of updating and pre-filtering environment map the ReflectionProbe is no updated automatically and requires reflectionProbe.set({ dirty: true })
whenever Skybox changes. The dirty flag is true by default so the Reflection Probe will get updated once on init.
Load a 3D model as a scene: an object containing a root entity hierarchy that you can add to the renderer like any other entity.
const scene = await renderer.loadScene('model.gltf')
renderer.add(scene.root)
Note: Currently only glTF is supported (JSON, binary and Embedded).
Start by creating new class as follows:
// MyComponent.js
const Signal = require('signals')
function MyComponent(opts) {
this.type = 'MyComponent'
this.entity = null
this.numberParameter = 1
this.stringParameter = 'some text'
this.changed = new Signal()
this.dirty = false
this.set(opts)
}
// this function gets called when the component is added
// to an enity
MyComponent.prototype.init = function(entity) {
this.entity = entity
}
MyComponent.prototype.set = function(opts) {
Object.assign(this, opts)
this.dirty = true
Object.keys(opts).forEach((prop) => this.changed.dispatch(prop))
}
MyComponent.prototype.update = function() {
if (!this.dirty) return
this.dirty = false
const transform = this.entity.transform
// do sth with transform
const geom = this.entity.getComponent('Geometry')
// do sth with geom
}
// by pex-renderer convention we export factory function
// instead of the class type
module.exports = function createMyComponent(opts) {
return new MyComponent(opts)
}
Create instance of your component and add it to an entity.
const createMyComponent = require('/path/to/MyComponent')
const myComponent = createMyComponent({ numberParameter: 1 })
const entity = renderer.entity([myComponent])
MIT, see LICENSE.md for details.