Skip to content
Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

nubAll Contributors


nub is a simple, expressive, language-agnostic, and extensible visual computing library, featuring interaction, visualization and animation frameworks and supporting advanced (onscreen/offscreen) rendering techniques, such as view frustum culling.

nub is meant to be coupled with third party real and non-real time renderers. Our current release supports the 2D and 3D PGraphicsOpenGL (a.k.a. P2D and P3D, respectively) Processing renderers.

If looking for the API docs, check them here.

Readers unfamiliar with geometry transformations may first check the great Processing 2D transformations tutorial by J David Eisenberg and the affine transformations and scene-graphs presentations that discuss some related formal foundations.


Instantiate your on-screen scene at the setup():

// import all nub classes
import nub.primitives.*;
import nub.core.*;
import nub.processing.*;

Scene scene;

void setup() {
  scene = new Scene(this);

The Scene.context() corresponds to the PApplet main PGraphics instance.

Off-screen scenes should be instantiated upon a PGraphics object:

import nub.primitives.*;
import nub.core.*;
import nub.processing.*;

Scene offScreenScene;

void setup() {
  offScreenScene = new Scene(createGraphics(w, h / 2, P3D));

In this case, the offScreenScene.context() corresponds to the PGraphics instantiated with createGraphics() (which is of course different than the PApplet main PGraphics instance).


A node may be translated, rotated and scaled (the order is important) and be rendered when it has a shape. Node instances define each of the nodes comprising the scene tree. To illustrate their use, suppose the following scene hierarchy is being implemented:

 n1 eye
 n2 n3

To setup the scene hierarchy of nodes use code such as the following:

import nub.primitives.*;
import nub.core.*;
import nub.processing.*;

Scene scene;
Node n1, n2, n3;

void setup() {
  size(720, 480, P3D);
  // the scene object creates a default eye node
  scene = new Scene(this);
  // Create a top-level node (i.e., a node whose reference is null) with:
  n1 = new Node();
  // whereas for the remaining nodes we pass any constructor taking a
  // reference node parameter, such as Node(Node referenceNode), or
  // Node(Node referenceNode, shape). Here the shape parameter defines
  // an immediate mode rendering procedure node visual representation
  n2 = new Node(n1, (PGraphics pg) -> {
    pg.fill(255, 0, 255);, 10, 30);
  // here the shape paramenter defines a retained mode rendering
  // procedure node visual representation
  n3 = new Node(n1, createShape(BOX, 30));
  // translate the node to make it visible
  n3.translate(50, 50, 50);

Note that the hierarchy of nodes may be modified with setReference(Node) and the scene eye() set from an arbitrary node instance with setEye(Node).


A node position, orientation and magnitude may be set with the following methods:

Node localization Position Orientation Magnitude
Globally setWorldPosition(vector) setWorldOrientation(quaternion) setWorldMagnitude(scalar)
Locally setPosition(vector) setOrientation(quaternion) setMagnitude(scalar)
Incrementally translate(vector, [inertia]) rotate(quaternion, [inertia]), scale(scalar, [inertia])

Note that the optional inertia parameter should be a value in [0..1], 0 no inertia (which is the default value) & 1 no friction. Its implementation was inspired by the great PeasyCam damped actions.

Motion filters

Calling setTranslationFilter(filter params), setRotationFilter(filter, params) and/or setScalingFilter(filter, params) will apply a filter which is a function used to limit the node motion when calling any of the methods found in the previous section. The node provides the following default filters: translationalAxisFilter, translationalPlaneFilter and rotationalAxisFilter (see the NodeFilters and the Luxo examples).

Visual hints

The node visual representation may be configured using the following hints:

  • SHAPE which displays the node shape set with setShape(shape) which is either an immediate-mode rendering Processing procedure or a retained-mode rendering Processing PShape.
  • HUD which displays the node Heads-Up-Display set with setHUD(shape).
  • BOUNDS which displays the bounding volume of each scene for which this node is the eye. Only meaningful if there's a second scene perspective to look at this eye node from.
  • KEYFRAMES which displays the Catmull Rom spline defining the current node animation path.
  • AXES which displays an axes hint centered at the node worldPosition() an oriented according to the node worldOrientation().
  • BULLSEYE which displays a bullseye centered at the node worldPosition() screen projection. Call setBullsEyeSize(value) to set the size of the hint.
  • TORUS which displays a torus solenoid.
  • CAMERA which displays a camera hint centered at the node screen projection.


  1. The actual node visual hint() is a bitwise-or mask of a subset of the above hints enabled at a given time, and so it does the hint parameter in methods to enable (node.enableHint(hint)), disable (node.disableHint(hint)) or toggle (node.toggleHint(hint)) them, e.g., node.enableHint(Node.SHAPE | Node.KEYFRAMES) enables altogether the SHAPE and KEYFRAMES node hints.
  2. Displaying the hint requires first to enabling it (enableHint(mask)) and then calling a scene rendering algorithm.
  3. Those enabled visual hints are used to pick the node with ray casting (see picking). Fine tune the mask picking hint with enablePicking(mask), disablePicking(mask) and togglePicking(mask), where mask is also a bitwise-or of the above hints.

Hints configuration

The node configHint(hint, params) configures hint using varargs params as follows:

  • BOUNDS: configHint(Node.BOUNDS, boundsWeight).
  • KEYFRAMES: configHint(Node.KEYFRAMES, keyframesMask) or configHint(Node.KEYFRAMES, keyframesMask, steps) or configHint(Node.KEYFRAMES, keyframesMask, steps, splineStroke), or configHint(Node.KEYFRAMES, keyframesMask, steps, splineStroke, splineWeight).
  • AXES: configHint(Node.AXES, axesLength).
  • BULLSEYE: configHint(Node.BULLSEYE, bullseyeStroke), configHint(Node.BULLSEYE, bullseyeShape), or configHint(Node.BULLSEYE, bullseyeStroke, bullseyeShape).
  • TORUS: configHint(Node.TORUS, torusStroke), or configHint(Node.TORUS, torusStroke, torusFaces).
  • CAMERA: configHint(Node.CAMERA, cameraStroke) or configHint(Node.CAMERA, cameraStroke, cameraLength).

e.g., node.configHint(Node.BULLSEYE, color(255, 0, 0)) colors red the BULLSEYE hint.

Space transformations

The following Scene methods transform points (locations) and vectors (displacements) between screen space (a box of width * height * 1 dimensions where user interaction takes place), NDC and nodes (including the world, i.e., the null node):

Space transformation Points Vectors
NDC to Screen ndcToScreenLocation(point) ndcToScreenDisplacement(vector)
Screen to NDC screenToNDCLocation(pixel) screenToNDCDisplacement(vector)
Screen to Node location(pixel, node) displacement(vector, node)
Node to Screen screenLocation(point, node) screenDisplacement(vector, node)
Screen to World location(pixel) displacement(vector)
World to Screen screenLocation(point) screenDisplacement(vector)

Note that point, pixel and vector are Vector instances.

The following Node methods transform points (locations) and scalars / vectors/ quaternions (displacements) between different node instances (including the world):

Space transformation Points Scalars / Vectors / Quaternions
Node to (this) Node location(point, node) displacement(element, node)
(this) Node to Reference referenceLocation(point) referenceDisplacement(element)
Reference to (this) Node localLocation(point) localDisplacement(element)
World to (this) Node location(point) displacement(element)
(this) Node to World worldLocation(point) worldDisplacement(element)

Note that point is a Vector instance and element is either a float (scalar), Vector or Quaternion one.


Keyframes allow to define the position, orientation and magnitude a node (including the eye) should have at a particular moment in time. The node may then be animated through a Catmull-Rom spline, matching in space-time the key-frames which defines it. Use code such as the following:

Scene scene;
PShape pshape;
Node shape;

void setup() {
  // ...
  shape = new Node(pshape);
  for (int i = 0; i < random(4, 10); i++) {
    // addKeyFrame(hint, elapsedTime) where elapsedTime is defined respect
    // to the previously added key-frame and expressed in milliseconds.
    shape.addKeyFrame(Node.AXES | Node.SHAPE, i % 2 == 1 ? 1000 : 4000);

which will create a shape interpolator containing [4..10] random key-frames. See the KeyFrames example.


Custom node behaviors to be executed every iteration of the scene render() algorithm may be added with code such as the following:

Scene scene;
Node node;

void setup() {
  Scene scene = new Scene();
  Node node = new Node();
  scene.addBehavior(node, this::behavior);

void behavior(Scene scene, Node node) {
  // Custom node behavior for the scene.render() algorithm

See the CajasOrientadas and the the ViewFrustumCulling examples.


Render the scene node hierarchy from its eye point-of-view with render(subtree) and display it with image(cornerX, cornerY), e.g.,

void draw() {
  // drawing commands here take place at the world coordinate system
  // use scene.context() to access the scene Processing PGraphics instance,
  // e.g., reset the background with:
  scene.render(subtree); // the subtree param is optional
  // drawing commands here also take place at the world coordinate system
  scene.image(cornerX, cornerY);

Onscreen scenes may just call render(subtree), e.g.,

void draw() {
  scene.render(subtree); // the subtree param is optional
  // drawing commands here take place at the world coordinate system

Or use the higher level display() algorithm to achive the same results for both of the above onscreen and offscreen scenes:

void draw() {
  // the offscreen scene subtree is rendered at (cornerX, cornerY)
  scene.display(color(100), subtree, cornerX, cornerY);
void draw() {
  // the onscreen scene subtree is rendered with the world axes
  scene.display(color(125), true, subtree);


  1. Enclose 2D screen space with scene.beginHUD() and scene.endHUD() stuff (such as gui elements and text) with to render it on top of a 3D scene.
  2. Customize the rendering traversal algorithm by setting a node custom behavior with scene.addBehavior(node, behavior), see the ViewFrustumCulling example.

Drawing functionality

The Scene implements several static drawing functions that complements those already provided by Processing, such as: drawCylinder(PGraphics, int, float, float), drawHollowCylinder(PGraphics, int, float, float, Vector, Vector), drawCone(PGraphics, int, float, float, float, float), drawCone(PGraphics, int, float, float, float, float, float) and drawTorusSolenoid(PGraphics, int, int, float, float).

Drawing functions that take a PGraphics parameter (including the above static ones), such as beginHUD(PGraphics), endHUD(PGraphics), drawAxes(PGraphics, float), drawCross(PGraphics, float, float, float) and drawGrid(PGraphics) among others, can be used to set a node shape.

Another scene's eye (different than this one) can be drawn with drawFrustum(Scene). Typical usage include interactive minimaps and visibility culling visualization and debugging.


Node interaction methods

To directly interact with a given node (or the eye when no node param is given), call any of the following screen space scene methods:

Scene screen space method
shift([node], [dx], [dy], [dz], [inertia])
turn([node], roll, pitch, yaw, [inertia])
zoom([node], delta, [inertia])
spin([node], [pixel1X], [pixel1Y], [pixel2X], [pixel2Y], [inertia])

Customize node behaviors by registering a user gesture data parser function_object with the node setInteraction(function_object) method, and then send gesture data to the node by calling one of the scene custom interaction invoking methods: interact(gesture), interact(tag, gesture) or interact(node, gesture), where gesture type is Object..... See the CustomNodeInteraction example.

Eye mouse and keyboard code snippets

// define a mouse-dragged eye interaction
void mouseDragged() {
  if (mouseButton == LEFT)
  else if (mouseButton == RIGHT)
// define a mouse-moved eye interaction
void mouseMoved(MouseEvent event) {
  if (event.isShiftDown())
    scene.turn(scene.mouseRADY(), scene.mouseRADX(), 0);
// define a mouse-wheel eye interaction
void mouseWheel(MouseEvent event) {
  if (scene.is3D())
    scene.moveForward(event.getCount() * 20);
    scene.zoom(event.getCount() * 20);
// define a mouse-click eye interaction
void mouseClicked(MouseEvent event) {
  if (event.getCount() == 1)
// define a key-pressed eye interaction
void keyPressed() {
  // roll with 'x' key
  scene.turn(key == 'x' ? QUARTER_PI / 2 : -QUARTER_PI / 2, 0, 0);

The SpaceNavigator and CustomEyeInteraction examples illustrate how to set up other hardware such as a keyboard or a full fledged 6-DOFs device like the space-navigator.

Node mouse and keyboard code snippets

void mouseDragged() {
  // spin n1
  if (mouseButton == LEFT)
  // shift n3
  else if (mouseButton == RIGHT)
  // zoom n1
    scene.zoom(n1, scene.mouseDX());
void keyPressed() {
  if (key == CODED)
    if(keyCode == UP)
      // shift n2 up
      scene.shift(n2, 0, 10);
    if(keyCode == DOWN)
      // shift n2 down
      scene.shift(n2, 0, -10);


Picking a node (which should be different than the scene eye) to interact with it is a two-step process:

  1. Tag the node using an arbitrary tag either with tag(tag, node) or ray-casting:

    Synchronously 🔹 Asynchronously 🔸
    updateTag([tag], [pixelX], [pixelY]) tag([tag], [pixelX], [pixelY])

    🔹 The tagged node is returned immediately 🔸 The tagged node is returned during the next call to the render() algorithm

  2. Interact with your tagged nodes by calling any scene method implementing the interact(tag, gesture...) which resolves the node param in the methods above (using the scene node(tag) method):

    Scene screen space method
    shift([tag], [dx], [dy], [dz], [inertia])
    turn([tag], roll, pitch, yaw, [inertia])
    zoom([tag], delta, [inertia])
    spin([tag], [pixel1X], [pixel1Y], [pixel2X], [pixel2Y], [inertia])


  1. A node can have multiple tags but a given tag cannot be assigned to more than one node, and since the null tag is allowed, signatures of all the above methods lacking the tag parameter are provided for convenience.
  2. Refer to the picking() and enablePicking(int) for the different ray-casting node picking modes.
  3. To check if a given node would be picked with a ray cast at a given screen position, call tracks(node, [pixelX], [pixelY]).
  4. To tag the nodes in a given array with ray casting use updateTag([tag], [pixelX], [pixelY], Node[]).
  5. Set Scene.inertia in [0..1] (0 no inertia & 1 no friction) to change the default inertia value globally. It is initially set to 0.8. See the CajasOrientadas example.
  6. Invoke custom node behaviors by either calling the scene interact(node, gesture), interact(gesture) or interact(tag, gesture) methods. See the CustomNodeInteraction example.

Mouse and keyboard code snippets

// pick with mouse-moved
void mouseMoved() {

// interact with mouse-dragged
void mouseDragged() {
  if (mouseButton == LEFT)
    // spin the picked node or the eye if no node has been picked
  else if (mouseButton == RIGHT)
    // shift the picked node or the eye if no node has been picked
    // zoom the picked node or the eye if no node has been picked
    scene.zoom(mouseX - pmouseX);
// pick with mouse-clicked
void mouseClicked(MouseEvent event) {
  if (event.getCount() == 1)
    // use the null tag to manipulate the picked node with mouse-moved
  if (event.getCount() == 2)
    // use the "key" tag to manipulate the picked node with key-pressed

// interact with mouse-moved
void mouseMoved() {
  // spin the node picked with one click

// interact with key-pressed
void keyPressed() {
  // focus the node that has been picked with two clicks


Import/update it directly from your PDE. Otherwise download your release and extract it to your sketchbook libraries folder.


Thanks goes to these wonderful people (emoji key):

Jean Pierre Charalambos
Jean Pierre Charalambos

📝 🐛 💻 🎨 📖 📋 💡 💵 🔍 🤔 📦 🔌 💬 👀 📢 ⚠️ 📹

This project follows the all-contributors specification. Contributions of any kind welcome!