I am new to WebGL and Three.js. I am trying to visualize a large grid of circles changing colors at once.
As I increase the number of instances, it gets noticeably slower, where it takes seconds to update. What are some suggestions for improving my code? Can I update 4000 circles at once?
Here is my existing implementation:
<html>
<head>
<title>My first Three.js app</title>
<style></style>
</head>
<body>
<script src="./three.js"></script>
<script>
var ROWS = 40
var COLS = 100
var SEGMENTS = 10;
var windowWidth = window.innerWidth, windowHeight = window.innerHeight;
var camera, scene, renderer;
var group, text, plane;
function init() {
// create and append container/canvas
container = document.createElement( 'div' );
document.body.appendChild( container );
// create camera
camera = new THREE.PerspectiveCamera(100, windowWidth / windowHeight, 0.1, 1000 );
// set position of camera
camera.position.z = 500;
camera.position.x = windowWidth/2
camera.position.y = windowHeight/2
// Create a scene
scene = new THREE.Scene();
renderer = new THREE.CanvasRenderer();
renderer.setClearColor( 0xf0f0f0 );
renderer.setSize( windowWidth, windowHeight );
renderer.sortElements = false;
container.appendChild( renderer.domElement );
document.addEventListener( 'mousedown', onDocumentMouseDown, false );
}
function addCircle(color, x, y, z, s , radius) {
var geometry = new THREE.CircleGeometry(radius, SEGMENTS, SEGMENTS)
var material = new THREE.MeshBasicMaterial( { color: color, overdraw: true } );
var mesh = new THREE.Mesh( geometry, material );
mesh.position.set( x, y, z );
mesh.scale.set( s, s, s );
scene.add( mesh );
}
function toHex(d) {
var valueStr = d.toString(16);
valueStr = valueStr.length < 2 ? "0"+valueStr : valueStr;
var fillColor = "0x00" + valueStr + "00";
return parseInt(fillColor);
}
function drawData(data) {
var rows = data.length;
var cols = data[0].length;
distanceBetweenCircles = Math.min(windowWidth/(cols), windowHeight/(rows));
var radius = distanceBetweenCircles/2.0
for(var i = 0; i < data.length; i++) {
for (var j = 0; j < data[0].length; j++) {
var color = toHex(data[i][j])
var x = distanceBetweenCircles*j - radius
var y = distanceBetweenCircles*i - radius
addCircle( color, x, y, 0, 1 , radius-3);
}
}
}
function newData(){
var newData = []
for (var i = 0; i < ROWS; i++) {
var row = [];
for (var j = 0; j < COLS; j ++) {
row.push(Math.floor(Math.random()*255));
}
newData.push(row);
}
return newData;
}
function onDocumentMouseDown ( event ) {
event.preventDefault();
// Update circles
var randomData = newData()
drawData(randomData);
}
var render = function() {
requestAnimationFrame(render);
renderer.render(scene, camera);
}
init();
render();
</script>
</body>
</html>
To add a CircleGeometry object to a scene requires the circle to be drawn. Adding an image/texture of a circle to a scene requires the circle to be printed, so to speak.
Multiply that by 4000 and the drawing becomes quite expensive.
It would be faster to maintain an array of the meshes and update their properties, rather than creating a new set of geometries, materials and meshes every mouse click.
Memory management is extremely important in software design. There is a cost to every variable you introduce, especially those who's instantiation invokes a cascade of allocations behind some API call. This is the down side of using layerings like THREE which hide complexity but also hide ramifications of using their calls. Short of avoiding THREE and doing all the WebGL plumbing yourself (which is always a good first step before ignoring the plumbing and just using a shim like THREE), do some homework to identify what is getting created as you make calls to any API, like THREE. Rip out of inner loops variable creation for objects which should be reused across calls. To your question, yes you can easily update 4000 circles across each animation event loop time slice once your architecture is carefully thought through especially if you use shaders to craft your objects and avoid such computation back in the CPU
For pure speed I suggest you learn graphics by writing OpenGL/WebGL by hand instead of the higher level abstraction library Three.js ... the price of ease of use too often is higher computational load of unnecessary logic which can be cut out if written by hand
Here is a WebGL toy I built which has no Three.js ... it does real-time updates to geometry of 10's of thousands of objects as well as rendering audio using Web Audio API
https://github.com/scottstensland/webgl-3d-animation
Related
I have a basic scene in which I'm using each loops to add multiple meshes ( hundreds of simple cylinders ) to a group (for each line), and grouping the lines to cover the surfaces. The result is this:
This is the code to add these cylinders:
var base_material = new THREE.MeshPhongMaterial( {
color: 0x666666,
side: THREE.FrontSide,
});
var cylinderGeometry = new THREE.CylinderGeometry( 1, 1, 1, 4 );
var floor_geometry = new THREE.PlaneGeometry( 68, 10000, 10 );
var floor = new THREE.Mesh( floor_geometry, base_material );
floor.receiveShadow = true;
scene.add( floor );
floor.position.set(0,-15,-530);
floor.rotation.x = -Math.PI / 2;
// Add Floor Studs
for ( var i = 0; i < 15; i++ ) {
var lineGroup = new THREE.Group();
for ( var n = 0; n < 1000; n++ ) {
var cylinder = new THREE.Mesh( cylinderGeometry, base_material );
// cylinder.castShadow = true;
// cylinder.receiveShadow = true;
lineGroup.add( cylinder );
posZ = 0 - n*6;
cylinder.position.set(0,0, posZ);
}
scene.add( lineGroup );
posX = -28.4 + i*4.1;
lineGroup.position.set(posX,-14.7,0);
}
When I animate the camera to traverse through the scene the framerate is dire. Potential approaches I've come across include merging the geometry, possibly rendering out and loading in a single GLTF model with all of these cylinders, or duplicating them somehow. As you can see the geometry and material is created once and reused, however the mesh is recreated each time which I suspect is the culprit..
My question is, what is the most optimum of these approaches to do this, is there a standard best practice method?
Thanks in advance!
I'm using Cannon.js with Three.js. I have set a scene which has 5 columns of 4 blocks stacked on top of each other.
I want these to be interactable with other objects I'm planning on adding to the scene. However, the blocks in the columns seem to be causing lots of micro-collisions and over time, jitter out of position. I want them to stay exactly in line until they're interacted with.
If you view the codepen and wait for about 20/30 seconds you'll see the blocks start to move. Is there something specific I need to set on these blocks to prevent this from happening?
Here is an example I've put together - https://codepen.io/danlong/pen/XxZROj
As an aside, there's also quite a big performance drop when there are these blocks in the scene which I wasn't expecting. I plan to add more objects to the scene and not sure why the performance drops?
Is it something to do with the below in my animate() loop?
this.world.step(1 / 30);
Code specifically to set up my 'Cannon world' and 'columns' is below:
Cannon World:
this.world = new CANNON.World();
this.world.defaultContactMaterial.contactEquationStiffness = 1e6;
this.world.defaultContactMaterial.contactEquationRegularizationTime = 3;
this.world.solver.iterations = 20;
this.world.gravity.set(0,-25,0);
this.world.allowSleep = true;
this.world.broadphase = new CANNON.SAPBroadphase(this.world);
Columns:
var geometry = new THREE.BoxBufferGeometry(5,5,5);
var material = new THREE.MeshNormalMaterial();
var shape = new CANNON.Box(new CANNON.Vec3(5/2, 5/2, 5/2));
for (var rows = 0, yPos = 2.5; rows < 4; rows++, yPos+=5) {
for (var i = -20; i <= 20; i+=10) {
// physics
var body = new CANNON.Body({
mass: 0.5,
position: new CANNON.Vec3(i, yPos, 0),
friction: 0.1,
restitution: 0.3
});
body.allowSleep = true;
body.sleepSpeedLimit = 0.01;
body.sleepTimeLimit = 1.0;
body.addShape(shape);
this.world.addBody(body);
this.bodies.push(body);
// material
var mesh = new THREE.Mesh(geometry, material);
this.scene.add(mesh);
this.meshes.push(mesh);
}
}
Try this?
body.sleepSpeedLimit = 1.0;
I have been trying to implement the drag and drop functionality found here...
http://www.smartjava.org/tjscb/07-animations-physics/07.08-drag-n-drop-object-around-scene.html
Whenever I customise it slightly and use it in my project I get the following..
"Uncaught TypeError: Cannot read property 'point' of undefined"
whenever I try to drag a cube. The rotation isn't occurring so it must be recognising that I'm trying to drag an object and it relates to this line of code..
"selectedObject.position.copy(intersects[0].point.sub(offset))"
I assumed since I am new to all of this that I had messed up, so I copied all of the code from the link above into a new page (so should be identical) and ran it and I get the same thing (everything else works good)
Im probably missing something really stupid, I have searched for this and looked at other examples on how to achieve this, but since I was working my way through a book which explained everything I thought I would stick with this, and also it would be a good learning experience to figure out why its not working. If anyone could point me in the right direction I would appreciate it
<!DOCTYPE html>
<html>
<head>
<title>07.08 - Drag and drop object around scene</title>
<script type="text/javascript" src="js/threejs/three.min.js"></script>
<script type="text/javascript" src ="js/threejs/OrbitControls.js"></script>
<style>
body {
margin: 0;
overflow: hidden;
}
</style>
<script>
// global variables
var renderer;
var scene;
var camera;
var cube;
var control;
var orbit;
// used for drag and drop
var plane;
var selectedObject;
var offset = new THREE.Vector3();
var objects = [];
// based on http://mrdoob.github.io/three.js/examples/webgl_interactive_draggablecubes.html
function init() {
// create a scene, that will hold all our elements such as objects, cameras and lights.
scene = new THREE.Scene();
// create a camera, which defines where we're looking at.
camera = new THREE.PerspectiveCamera(45, window.innerWidth / window.innerHeight, 0.1, 1000);
// create a render, sets the background color and the size
renderer = new THREE.WebGLRenderer();
renderer.setClearColor(0xffffff, 1.0);
renderer.setSize(window.innerWidth, window.innerHeight);
plane = new THREE.Mesh(new THREE.PlaneGeometry(2000, 2000, 18, 18), new THREE.MeshBasicMaterial({
color: 0x00ff00,
opacity: 0.25,
transparent: true
}));
plane.visible = false;
scene.add(plane);
var dirLight = new THREE.DirectionalLight();
dirLight.position.set(25, 23, 15);
scene.add(dirLight);
var dirLight2 = new THREE.DirectionalLight();
dirLight2.position.set(-25, 23, 15);
scene.add(dirLight2);
for (var i = 0; i < 200; i++) {
// create a cube and add to scene
var cubeGeometry = new THREE.BoxGeometry(2, 2, 2);
var cubeMaterial = new THREE.MeshLambertMaterial({color: Math.random() * 0xffffff});
cubeMaterial.transparent = true;
cube = new THREE.Mesh(cubeGeometry, cubeMaterial);
objects.push(cube);
cube.scale.x = Math.random() + 0.5 * 2;
cube.scale.y = Math.random() + 0.5 * 2;
cube.scale.z = Math.random() + 0.5 * 2;
cube.position.x = Math.random() * 50 - 25;
cube.position.y = Math.random() * 50 - 25;
cube.position.z = Math.random() * 50 - 25;
cube.rotation.x = Math.random() * Math.PI * 2;
cube.rotation.y = Math.random() * Math.PI * 2;
cube.rotation.z = Math.random() * Math.PI * 2;
scene.add(cube);
}
// position and point the camera to the center of the scene
camera.position.x = 35;
camera.position.y = 35;
camera.position.z = 53;
camera.lookAt(scene.position);
// add some controls so we can rotate
orbit = new THREE.OrbitControls(camera);
// add the output of the renderer to the html element
document.body.appendChild(renderer.domElement);
// call the render function
render();
}
function render() {
renderer.render(scene, camera);
orbit.update();
requestAnimationFrame(render);
}
document.onmousemove = function (event) {
// make sure we don't access anything else
event.preventDefault();
// get the mouse positions
var mouse_x = ( event.clientX / window.innerWidth ) * 2 - 1;
var mouse_y = -( event.clientY / window.innerHeight ) * 2 + 1;
// get the 3D position and create a raycaster
var vector = new THREE.Vector3(mouse_x, mouse_y, 0.5);
vector.unproject(camera);
var raycaster = new THREE.Raycaster(camera.position,
vector.sub(camera.position).normalize());
// first check if we've already selected an object by clicking
if (selectedObject) {
// check the position where the plane is intersected
var intersects = raycaster.intersectObject(plane);
// reposition the selectedobject based on the intersection with the plane
selectedObject.position.copy(intersects[0].point.sub(offset));
} else {
// if we haven't selected an object, we check if we might need
// to reposition our plane. We need to do this here, since
// we need to have this position before the onmousedown
// to calculate the offset.
var intersects = raycaster.intersectObjects(objects);
if (intersects.length > 0) {
// now reposition the plane to the selected objects position
plane.position.copy(intersects[0].object.position);
// and align with the camera.
plane.lookAt(camera.position);
}
}
};
document.onmousedown = function (event) {
// get the mouse positions
var mouse_x = ( event.clientX / window.innerWidth ) * 2 - 1;
var mouse_y = -( event.clientY / window.innerHeight ) * 2 + 1;
// use the projector to check for intersections. First thing to do is unproject
// the vector.
var vector = new THREE.Vector3(mouse_x, mouse_y, 0.5);
// we do this by using the unproject function which converts the 2D mouse
// position to a 3D vector.
vector.unproject(camera);
// now we cast a ray using this vector and see what is hit.
var raycaster = new THREE.Raycaster(camera.position,
vector.sub(camera.position).normalize());
// intersects contains an array of objects that might have been hit
var intersects = raycaster.intersectObjects(objects);
if (intersects.length > 0) {
orbit.enabled = false;
// the first one is the object we'll be moving around
selectedObject = intersects[0].object;
// and calculate the offset
var intersects = raycaster.intersectObject(plane);
offset.copy(intersects[0].point).sub(plane.position);
}
};
document.onmouseup = function (event) {
orbit.enabled = true;
selectedObject = null;
}
// calls the init function when the window is done loading.
window.onload = init;
</script>
</head>
<body>
</body>
</html>
"Uncaught TypeError: Cannot read property 'point' of undefined"
"selectedObject.position.copy(intersects[0].point.sub(offset))"
This means, intersects[0] is undefined which means the array intersects has no element (length = 0). You are using raycasting and it isn't working properly.
You should share your modified code so that we can check what is going wrong in your raycasting.
Update: I think your three.js version is greater than 71 while three.js version of this website is 71 or less. In the 72th version, there is an update in the raycaster -
Ignore invisible objects. (#mrdoob, #tschw)
So, the problem is here -
var intersects = raycaster.intersectObject(plane);
Since the plane is invisible, the intersectObject is returning empty array.
Workaround: I found a workaround. You can remove the following line -
plane.visible = false;
You can hide the material of the plane instead in the following way -
plane = new THREE.Mesh(new THREE.PlaneGeometry(2000, 2000, 18, 18), new THREE.MeshBasicMaterial({
color: 0xffff00,
opacity: 0.50,
transparent: true,
visible: false
}));
In this way, the raycaster will work properly and the plane will be invisible as well.
UPDATE Cause of problem has been found - see Update section end of question.
I have a complex app using THREE.js (r60) which adds a special plane object to the main scene. The plane geometry is determined by heightmapping from an internally-supplied base64 uri image (size 16x16, 32x32 or 64x64 pixels). The scene has two static lights (ambient and directional) and one moveable point light which switches on and off.
In the complex app the point light is not reflected by the plane object. (Point light is toggled by pressing "R" key or button).
I have made a first JSFiddle example using THREE.js latest version (r70) where the lights work fine.
[Update] I have now made a second JSFiddle example using the older THREE.js library (r60) it also works OK.
I suspect the problem in the complex app (r60) may have something to do with system capacity and or timing/sequencing. Capacity is definitely an issue because other simpler scene objects (boxes and cylinders) show individual responses or non-responses to the point light which vary from one run of the app to the next, seemingly depending on the overall level of system activity (cpu, memory usage). These simpler objects may reflect in one run but not in the next. But the heightmapped plane object is consistently non-reflective to the point light. These behaviors are observed on (i) a Win7 laptop and (ii) an Android Kitkat tablet.
The heightmapping process may be part of the cause. I say this because when I comment out the heightmapped plane and activate a simple similar plane object (with randomly assigned z-levels) the latter plane behaves as expected (i.e. it reflects point light).
I guess that the usual approach now would be to upgrade my complex app to r70 (not a trivial step) and then start disabling chunks of the app to narrow down the cause. However it may be that the way in which heightmapping is implemented (e.g. with a callback) is a factor in explaining the failure of the heightmapped plane to reflect point light.
[RE-WRITTEN] So I would be grateful if anyone could take a look at the code in the correctly-working, previously-cited, (r70) JSFiddle example and point out any glaring design faults which (if applied in more complex, heavilly-loaded apps) might lead to failure of the height-mapped plane to reflect point light.
Full code (javascript, not html or css) of the (r70) JSFiddle:-
//... Heightmap from Image file
//... see http://danni-three.blogspot.co.uk/2013/09/threejs-heightmaps.html
var camera, scene, renderer;
var lpos_x = -60,lpos_y = 20,lpos_z = 100;
var mz = 1;
var time = 0, dt = 0;
var MyPlane, HPlane;
base64_imgData = "data:image/jpeg;base64,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";
init();
animate();
//==================================================================
function init() {
scene = new THREE.Scene();
camera = new THREE.PerspectiveCamera(45, window.innerWidth / window.innerHeight, 10);
camera.position.x = 1300;
camera.position.y = 400;
camera.position.z = 0;
camera.lookAt(0, 0, 0);
scene.add(camera);
scene.add(new THREE.AmbientLight(0x001900));
SunLight = new THREE.DirectionalLight(0xff0000,.3,20000);//...color, intensity, range.
SunLight.position.set(0, 3000, -8000);
scene.add(SunLight);
//POINT LIGHT
PL_color = 0x0000ff;
PL_intensity = 10;
PL_range_to_zero_intensity = 1200;
PL = new THREE.PointLight(PL_color, PL_intensity, PL_range_to_zero_intensity);
scene.add(PL);
PL_pos_x = -100;
PL_pos_y = -100;
PL_pos_z = 120;
PL.position.set(PL_pos_x, PL_pos_y, PL_pos_z);
//INDICATOR SPHERE
var s_Geometry = new THREE.SphereGeometry(5, 20, 20);
var s_Material = new THREE.MeshBasicMaterial({
color: 0xaaaaff
});
i_Sphere = new THREE.Mesh(s_Geometry, s_Material);
i_Sphere.position.set(PL_pos_x, PL_pos_y, PL_pos_z);
scene.add(i_Sphere);
//Plane02
var Plane02Geo = new THREE.PlaneGeometry(50, 50); //...
var Plane02Material = new THREE.MeshPhongMaterial({
side: THREE.DoubleSide
}, {
color: 0xaaaaaa
});
Plane02 = new THREE.Mesh(Plane02Geo, Plane02Material);
Plane02.position.set(0, 0, -120);
scene.add(Plane02);
//PEAS
xxx = SOW_F_Make_peas();
//RENDERER
renderer = new THREE.WebGLRenderer({
antialias: true
});
renderer.setSize(window.innerWidth, window.innerHeight);
renderer.shadowMapEnabled = true;
renderer.shadowMapSoft = false;
document.body.appendChild(renderer.domElement);
// controls
controls = new THREE.OrbitControls(camera, renderer.domElement);
xxx = SOW_F_Make_Heightmap_Object_from_Image_File(scene, camera);
} //...EOFunction Init
//==================================================================
function animate() {
dt = 0.1;
time += dt;
if (time < 10000) {
requestAnimationFrame(animate);
//move point light & indicator sphere
speed = 16;
if (Math.abs(PL_pos_z) > 400) mz = (-1)* mz;
PL_pos_x += 0.01 * speed * mz;
PL_pos_y += 0.05 * speed * mz;
PL_pos_z -= 0.2 * speed * mz;
PL.position.set(PL_pos_x, PL_pos_y, PL_pos_z);
i_Sphere.position.set(PL_pos_x, PL_pos_y, PL_pos_z);
renderer.render(scene, camera);
} else alert("Time=" + time + "Finished");
}
//==================================================================
function SOW_F_Make_Heightmap_Object_from_Image_File(givenScene, givenCamera) {
//... Read a Heightmap from a coloured image file
//... into a (pre-defined global) plane object called HPlane
MyImage = new Image();
MyImage.onload = function () {
var MyPlane_width = 1000;//6000; //...MyPlane width or height are in scene units and do not have to match image width or height
var MyPlane_height = 1000;//6000;
var MyPlane_w_segs = MyImage.naturalWidth - 1; //... important that this mapping is correct for texture 1 pixel :: 1 segment.
var MyPlane_h_segs = MyImage.naturalHeight - 1; //... important that this mapping is correct for texture 1 pixel :: 1 segment.
var Hgeometry = new THREE.PlaneGeometry(MyPlane_width, MyPlane_height, MyPlane_w_segs, MyPlane_h_segs);
//var texture = THREE.ImageUtils.loadTexture( '/images/Tri_VP_Texturemap.jpg' );
var texture = THREE.ImageUtils.loadTexture( base64_imgData );
//... Choose texture or color
//var Hmaterial = new THREE.MeshLambertMaterial( { map: texture, side: THREE.DoubleSide} );//....fails
var Hmaterial = new THREE.MeshPhongMaterial( {
color: 0x111111 , side: THREE.DoubleSide } ); //... works OK
HPlane = new THREE.Mesh(Hgeometry, Hmaterial);
//...get Height Data from Image
var scale = 0.6;//1//6; //0.25;
var Height_data = DA_getHeightData(MyImage, scale);
//... set height of vertices
X_offset = 0;
Y_offset = 0;
Z_offset = -100; //...this will (after rotation) add to the vertical height dimension (+ => up).
for (var iii = 0; iii < HPlane.geometry.vertices.length; iii++) {
//HPlane.geometry.vertices[iii].x = X_offset;
//HPlane.geometry.vertices[iii].y = Y_offset;
HPlane.geometry.vertices[iii].z = Z_offset + Height_data[iii];
}
//----------------------------------------------------------------------
//... Must do it in this order...Faces before Vertices
//... see WestLangley's response in http://stackoverflow.com/questions/13943907/my-object-isnt-reflects-the-light-in-three-js
HPlane.rotation.x = (-(Math.PI) / 2); //... rotate MyPlane -90 degrees on X
//alert("Rotated");
HPlane.geometry.computeFaceNormals(); //... for Lambert & Phong materials
HPlane.geometry.computeVertexNormals(); //... for Lambert & Phong materials
/*
HPlane.updateMatrixWorld();
HPlane.matrixAutoUpdate = false;
HPlane.geometry.verticesNeedUpdate = true;
*/
givenScene.add(HPlane);
HPlane.position.set(0, -150, 0);//... cosmetic
//return HPlane; //... not necessary, given that HPlane is global.
} ; //... End of MyImage.onload = function ()
//===============================================================
//... *** IMPORTANT ***
//... Only NOW do we command the script to actually load the image source
//... This .src statement will load the image from file into MyImage object
//... and invoke the pre-associated MyImage.OnLoad function
//... cause cross-origin problem: MyImage.src = '/images/Tri_VP_Heightmap_64x64.jpg'; //...if image file is local to this html file.
MyImage.src = base64_imgData;//... uses image data provided in the script to avoid Cross-origin file source restrictions.
} //... End of function SOW_F_Make_Heightmap_Object_from_Image_File
//===========================================================================
function DA_getHeightData(d_img, scale) {
//... This is used by function SOW_F_Make_Heightmap_Object_from_Image_File.
//if (scale == undefined) scale=1;
var canvas = document.createElement('canvas');
canvas.width = d_img.width; //OK
canvas.height = d_img.height;
var context = canvas.getContext('2d');
var size = d_img.width * d_img.height;
var data = new Float32Array(size);
context.drawImage(d_img, 0, 0);
for (var ii = 0; ii < size; ii++) {
data[ii] = 0;
}
var imgData = context.getImageData(0, 0, d_img.width, d_img.height);
var pix = imgData.data; //... Uint(8) UnClamped Array[1024] for a 16x16 = 256 pixel image = 4 slots per pixel.
var jjj = 0;
//... presumably each pix cell can have value 0 to 255
for (var iii = 0; iii < pix.length; iii += 4) {
var all = pix[iii] + pix[iii + 1] + pix[iii + 2];
//... I guess RGBA and we don't use the fourth cell (A, = Alpha channel)
jjj++;
data[jjj] = all * scale / 3; //...original code used 12 not 3 ??? and divided by scale.
//console.log (iii, all/(3*scale), data[jjj]);
}
return data;
} //... end of function DA_getHeightData(d_img,scale)
//==================================================================================================
function SOW_F_Get_A_Plane(givenScene, givenCamera) {
//...MyPlane width or height are in scene units and do not have to match image width or height
var MyPlane_width = 1000;
var MyPlane_height = 1000;
var MyPlane_w_segs = 64; //...
var MyPlane_h_segs = 64; //...
geometry = new THREE.PlaneGeometry(MyPlane_width, MyPlane_height, MyPlane_w_segs, MyPlane_h_segs);
//var material = new THREE.MeshLambertMaterial( { color: 0xeeee00, side: THREE.DoubleSide} );
var material = new THREE.MeshPhongMaterial({
color: 0xeeee00,side: THREE.DoubleSide
}); //... OK
MyPlane = new THREE.Mesh(geometry, material);
givenScene.add(MyPlane);
MyPlane.rotation.x = (-(Math.PI) / 2); // rotate it -90 degrees on X
MyPlane.position.set(0, 100, 0);
MyPlane.geometry.computeFaceNormals(); //...for Lambert & Phong materials
MyPlane.geometry.computeVertexNormals(); //...for Lambert & Phong materials
/*
MyPlane.geometry.verticesNeedUpdate = true;
MyPlane.updateMatrixWorld();
MyPlane.matrixAutoUpdate = false;
*/
} //... EOF SOW_F_Get_A_Plane
//====================================================================
function SOW_F_Make_peas()
{
//----------------- Make an array of spheres -----------------------
Pea_geometry = new THREE.SphereGeometry(5,16,16);
//Pea_material = new THREE.MeshNormalMaterial({ shading: THREE.SmoothShading});
Pea_material = new THREE.MeshPhongMaterial({ color: 0xaa5522});
// global...
num_peas = 1200;
for (var iii = 0; iii < num_peas; iii++)
{
//...now global
ob_Pea = new THREE.Mesh(Pea_geometry, Pea_material);
ob_Pea.position.set(
400 * Math.random() - 150,
300 * Math.random() - 150,
1200 * Math.random() - 150);
scene.add(ob_Pea);//TEST
}
}
UPDATE
It appears the problem is a result of phasing. See this new JSFiddle(r70). Pointlight is created in function init() but not added to scene, or is immediately removed from scene after being added. Then various graphical mesh objects are created. When pointlight is added back to the scene (in the animate loop) it is too late - the mesh objects will not be illuminated by the pointlight.
A procedural solution is simply to not remove pointlights from the scene if they are to be used later. If they need to be "extinguished" temporarilly then just turn down the intensity and turn it up later: e.g.
myPointLight.intensity = 0.00
I've made a simple scene in blender which contains a simple box and two shape deformation keyframes.
My exported .js file contains lots of morph targets (each one for each animation frame I suppose?), but still no animation shown in production, just a static box.
Here's the way I'm trying to get this working:
<script src="three.js" type="text/javascript"></script>
<script type="text/javascript">
var size_width = window.innerWidth;
var size_height = window.innerHeight;
var player;
var scene = new THREE.Scene();
var camera = new THREE.PerspectiveCamera(75, size_width/size_height, 0.1, 1000);
var renderer = new THREE.WebGLRenderer();
var animation;
var animOffset = 1;
var duration = 1000;
var keyframes = 101;
var interpolation = duration / keyframes;
var lastKeyframe = 0;
var currentKeyframe = 0;
renderer.setSize(size_width, size_height);
document.body.appendChild(renderer.domElement);
camera.position.x = 10;
camera.position.y = -20;
camera.position.z = 10;
camera.rotation.x = 1.4;
var player_loader = new THREE.JSONLoader();
player_loader.load( "boxy.js", function(geo) {
player = new THREE.Mesh(geo);
scene.add(player);
});
function animate() {
requestAnimationFrame(animate);
render();
}
function render() {
if ( player ) {
var time = Date.now() % duration;
var keyframe = Math.floor( time / interpolation );
if ( keyframe != currentKeyframe ) {
player.morphTargetInfluences[ lastKeyframe ] = 0;
player.morphTargetInfluences[ currentKeyframe ] = 1;
player.morphTargetInfluences[ keyframe ] = 0;
lastKeyframe = currentKeyframe;
currentKeyframe = keyframe;
}
player.morphTargetInfluences[ keyframe ] = ( time % interpolation ) / interpolation;
player.morphTargetInfluences[ lastKeyframe ] = 1 - player.morphTargetInfluences[ keyframe ];
}
renderer.render(scene, camera);
}
animate();
</script>
here's my export:
http://touhou.ru/upload/7b7513a903963b0804b0be763b8cc67c.js
Also no errors were reported to the console.
You need to render your mesh with a material that expects morph targets. You can do this by instantiating a material with the morphTargets boolean set to true in the constructor options.
I'm not sure how familiar you are with three.js, but in most cases people create mesh objects with both a geometry object and a material object passed to the mesh constructor as arguments. You only gave the constructor a geometry object.
To get the animation running in your code change the line where you instantiate a new mesh in the loader callback:
player = new THREE.Mesh(geo);
to instantiate a mesh with a material with morph targets enable:
player = new THREE.Mesh( geo, new THREE.MeshLambertMaterial({ morphTargets: true }) );
When I ran your code with that change I saw one corner of a cube deforming outward and then back in.