We Keep Coding

How to get the visible width and height of a ThreeJS mesh object in pixel unit

I am facing the challenge of figuring out the visible view width and height of a ThreeJS mesh object in pixel units.
In the screenshot below you can see objects floating in 3D space, on mouse click I need to be able to figure out what view width and height they are occupying in pixels
As I am rather new to ThreeJS it is taking me rather long to find a solution, so I would welcome any kind of assistance.
The below function shows what kind of approaches I have been trying.
getObjectSizeInViewSpace(object){
const size = new THREE.Vector3()
const box = new THREE.Box3().setFromObject(object).getSize(size)
size.project(this.camera)
let halfWidth = window.innerWidth / 2;
let halfHeight = window.innerHeight / 2;
size.x = (size.x*halfWidth)
size.y = (size.y*halfHeight)
return new THREE.Vector2(size.x,size.y)
}

You're looking for Vector3.project(). This basically takes world-space (3D) coordinates, and uses the camera's viewport to convert into normalized device coordinates, which range from [-1, 1]. For example x: -1 is the left side of the screen, and x: 1 is the right side. So you'll have to take the 4 vectors (top left, top right, bottom left, bottom right)` of your plane to calculate their pixel dimensions in your browser:
// Get 3D positions of top left corner (assuming they're not rotated)
vec3 topLeft = new Vector3(
plane.position.x - planeWidth / 2,
plane.position.y - planeHeight / 2,
plane.positon.z
);
// This converts x, y, z to the [-1, 1] range
topLeft.project(camera);
// This converts from [-1, 1] to [0, windowWidth]
const topLeftX = (1 + topLeft.x) / 2 * window.innerWidth;
const topLeftY = (1 - topLeft.y) / 2 * window.innerHeight;
Notice the topLeftY value is inverted, since -y in 3D space goes +y in pixel coordinates. Do this 4 times (once for each corner), and then you can subtract (right - left) to get the width, and the same for the height.

THREE.js orthographic camera zoom to mouse point

I'm working on an orthographic camera for our THREE.js app. Essentially, this camera will present the scene to the user in 2D (users have the option of switching between the 2D and 3D camera). This camera will allow for panning and zooming to mouse point. I have the panning working, and I have zooming working, but not zooming to mouse point. Here's my code:
import React from 'react';
import T from 'three';
let panDamper = 0.15;
let OrthoCamera = React.createClass({
getInitialState: function () {
return {
distance: 150,
position: { x: 8 * 12, y: 2 * 12, z: 20 * 12 },
};
},
getThreeCameraObject: function () {
return this.camera;
},
applyPan: function (x, y) { // Apply pan by changing the position of the camera
let newPosition = {
x: this.state.position.x + x * -1 * panDamper,
y: this.state.position.y + y * panDamper,
z: this.state.position.z
};
this.setState({position: newPosition});
},
applyDirectedZoom: function(x, y, z) {
let zoomChange = 10;
if(z < 0) zoomChange *= -1;
let newDistance = this.state.distance + zoomChange;
let mouse3D = {
x: ( x / window.innerWidth ) * 2 - 1,
y: -( y / window.innerHeight ) * 2 + 1
};
let newPositionVector = new T.Vector3(mouse3D.x, mouse3D.y, 0.5);
newPositionVector.unproject(this.camera);
newPositionVector.sub(this.camera.position);
let newPosition = {
x: newPositionVector.x,
y: newPositionVector.y,
z: this.state.position.z
};
this.setState({
distance: newDistance,
position: newPosition
});
},
render: function () {
let position = new T.Vector3(this.state.position.x, this.state.position.y, this.state.position.z);
let left = (this.state.distance / -2) * this.props.aspect + this.state.position.x;
let right = (this.state.distance / 2) * this.props.aspect + this.state.position.x;
let top = (this.state.distance / 2) + this.state.position.y;
let bottom = (this.state.distance / -2) + this.state.position.y;
// Using react-three-renderer
// https://github.com/toxicFork/react-three-renderer
return <orthographicCamera
{...(_.pick(this.props, ['near', 'far', 'name']))}
position={position}
left={left}
right={right}
top={top}
bottom={bottom}
ref={(camera) => this.camera = camera}/>
}
});
module.exports = OrthoCamera;
Some zooming towards the mouse point happens but it seems erratic. I want to keep a 2D view, so as I zoom, I also move the camera (rather than having a non-perpendicular target, which kills the 2D effect).
I took cues from this question. As far as I can tell, I am successfully converting to THREE.js coordinates in mouse3D (see the answer to this question).
So, given this setup, how can I smoothly zoom to the mouse point (mouse3D) using the orthographic camera and maintaining a two dimensional view? Thanks in advance.

Assuming you have a camera that is described by a position and a look-at (or pivot) point in world coordinates, zooming at (or away from) a specific point is quite simple at its core.
Your representation seems to be even simpler: just a position/distance pair. I didn't see a rotation component, so I'll assume your camera is meant to be a top-down orthographic one.
In that case, your look-at point (which you won't need) is simply (position.x, position.y - distance, position.z).
In the general case, all you need to do is move both the camera position and the look-at point towards the zoom-at point while preserving the camera normal (i.e. direction). Note that this will work regardless of projection type or camera rotation. EDIT (2020/05/01): When using an orthographic projection, this is not all you need to do (see update at the bottom).
If you think about it, this is exactly what happens when you're zooming at a point in 3D. You keep looking at the same direction, but you move ever closer (without ever reaching) your target.
If you want to zoom by a factor of 1.1 for example, you can imagine scaling the vector connecting your camera position to your zoom-at point by 1/1.1.
You can do that by simply interpolating:
var newPosition = new THREE.Vector3();
newPosition.x = (orgPosition.x - zoomAt.x) / zoomFactor + zoomAt.x;
newPosition.y = (orgPosition.y - zoomAt.y) / zoomFactor + zoomAt.y;
newPosition.z = (orgPosition.z - zoomAt.z) / zoomFactor + zoomAt.z;
As I said above, in your case you won't really need to update a look-at point and then calculate the new distance. Your new distance will simply be:
var newDistance = newPosition.y
That should do it.
It only gets a little bit more sophisticated (mainly in the general case) if you want to set minimum and maximum distance limits both between the position/look-at and position/zoom-at point pairs.
UPDATE (2020/05/01):
I just realized that the above, although correct (except for missing one minor but very important step) is not a complete answer to OP's question. Changing the camera's position in orthographic mode won't of course change the scale of graphics being rendered. For that, the camera's projection matrix will have to be updated (i.e. the left, right, top and bottom parameters of the orthographic projection will have to be changed).
For this reason, many graphics libraries include a scaling factor in their orthographic camera class, which does exactly that. I don't have experience with ThreeJS, but I think that property is called 'zoom'.
So, summing everything up:
var newPosition = new THREE.Vector3();
newPosition.x = (orgPosition.x - zoomAt.x) / zoomFactor + zoomAt.x;
newPosition.y = (orgPosition.y - zoomAt.y) / zoomFactor + zoomAt.y;
newPosition.z = (orgPosition.z - zoomAt.z) / zoomFactor + zoomAt.z;
myCamera.zoom = myCamera.zoom * zoomFactor
myCamera.updateProjectionMatrix()
If you want to use your orthographic camera class code above instead, you will probably have to change the section that computes left, right, top and bottom and add a scaling factor in the calculation. Here's an example:
var aspect = this.viewportWidth / this.viewportHeight
var dX = (this.right - this.left)
var dY = (this.top - this.bottom) / aspect
var left = -dX / (2 * this.scale)
var right = dX / (2 * this.scale)
var bottom = -dY / (2 * this.scale)
var top = dY / (2 * this.scale)
mat4.ortho(this.mProjection, left, right, bottom, top, this.near, this.far)

Image jumping at the beginning of scrolling

I wrote some code to zoom in my image, but when I scroll at the very beginning this picture jumps a little. How to fix the problem?
Full page view.
Editor view.
HTML
<canvas id="canvas"></canvas>
JS
function draw(scroll) {
scroll = (window.scrollY || window.pageYOffset) / (document.body.clientHeight - window.innerHeight) * 3000;
canvas.setAttribute('width', window.innerWidth);
canvas.setAttribute('height', window.innerHeight);
//The main formula that draws and zooms the picture
drawImageProp(ctx, forest, 0, (-scroll * 3.9) / 4, canvas.width, canvas.height + (scroll * 3.9) / 2);
}

Not a bug fix
I had a look at the Codepen example and it does jump at the top (sometimes). I have a fix for you but I did not have the time to locate the source of your code problem. I did notice that the jump involved a aspect change so it must be in the scaling that your error is. (look out for negatives)
GPU is a better clipper
Also your code is actually doing unnecessary work, because you are calculating the image clipping region. Canvas context does the clipping for you and is especially good at clipping images. Even though you provide the clip area the image will still go through clip as that is part of the render pipeline. The only time you should be concerned about the clipped display of an image is whether or not any part of the image is visible so that you don't send a draw call, and it only really matters if you are pushing the image render count (ie game sprite counts 500+)
Code example
Anyway I digress. Below is my code. You can add the checks and balances. (argument vetting, scaling max min, etc).
Calling function.
// get a normalised scale 0-1 from the scroll postion
var scale = (window.scrollY || window.pageYOffset) / (document.body.clientHeight - window.innerHeight);
// call the draw function
// scale 0-1 where 0 is min scale and 1 is max scale (min max determined in function
// X and y offset are clamped but are ranged
// 0 - image.width and 0 - image.height
// where 0,0 shows top left and width,height show bottom right
drawImage(ctx, forest, scale, xOffset, yOffset);
The function.
The comments should cover what you need to know. You will notice that all I am concerned with is how big the image should be and where the top left corner will be. The GPU will do the clipping for you, and will not cost you processing time (even for unaccelerated displays). I personally like to work with normalised values 0-1, it is a little extra work but my brain likes the simplicity, it also reduces the need for magic numbers (magics number are a sign that code is not adaptable) . Function will work for any size display and any size image. Oh and I like divide rather than multiply, (a bad coding habit that comes from a good math habit) replacing the / 2 and needed brackets with * 0.5 will make it more readable.
function drawImage(ctx, img, scale, x, y){
const MAX_SCALE = 4;
const MIN_SCALE = 1;
var w = canvas.width; // set vars just for source clarity
var h = canvas.height;
var iw = img.width;
var ih = img.height;
var fit = Math.max(w / iw, h / ih); // get the scale to fill the avalible display area
// Scale is a normalised value from 0-1 as input arg Convert to range
scale = (MAX_SCALE - MIN_SCALE) * scale + MIN_SCALE;
var idw = iw * fit * scale; // get image total display size;
var idh = ih * fit * scale;
x /= iw; // normalise offsets
y /= ih; //
x = - (idw - w) * x; // transform offsets to display coords
y = - (idh - h) * y;
x = Math.min( 0, Math.max( - (idw - w), x) ); // clamp image to display area
y = Math.min( 0, Math.max( - (idh - h), y) );
// use set transform to scale and translate
ctx.setTransform(scale, 0, 0, scale, idw / 2 + x, idh / 2 + y);
// display the image to fit;
ctx.drawImage(img, ( - iw / 2 ) * fit, (- ih / 2 ) * fit);
// restore transform.
ctx.setTransform(1, 0, 0, 1, 0, 0)
}
Sorry I did not solve the problem directly, but hopefully this will help you redesign your approch.
I recently added a similar answer involving zooming and panning (and rotation) with the mouse which you may be interested in How to pan the canvas? Its a bit messy still "note to self (my clean it up)" and has no bounds clamping. But shows how to set a zoom origin, and convert from screen space to world space. (find where a screen pixel is on a pan/scale/rotated display).
Good luck with your project.

Particles "run away" from mouse

Im creating a simple particle experiment on canvas. Now i want them to "run away" from mouse coursor over canvas. detecting the distance from the mouse is not a problem, but how to code their behaviour?
each particle is created as following:
var particle = {
x: Math.floor(Math.random() * width),
y: Math.floor(Math.random() * height),
xVel: Math.random() * 10 - 5,
yVel: Math.random() * 10 - 5,
}
so i assume i should also save the direction somehow, and if the distance from pointer is < x, reverse the direction? maybe also save old speed, and decrease it slowly while moving away?
how to detect the direction?

Velocity (xVel, yVel, together) is a 2D vector. And so is the distance between the mouse and the particles. A vector contains both direction and magnitude. So you want a vector that is the difference between the mouse position and the particle position.
var posRelativeToMouse = {
x: particle.x - mousPosX,
y: particle.y - mousPosY
};
So small numbers of x and y mean the the particle is close to the mouse, and big mean it's far away.
Next we need to figure out how these numbers should affect the velocity of the particle. So we need 2 things.
What direction do we push them in?
We already have this, mostly. posRelativeToMouse is a vector that has the direction we want. We just normalize it, which means to set the length of the vector to 1. To do that, we divide each component by the current length of the vector. The length of this vector is always the distance to from the particle to the mouse.
var distance = Math.sqrt(
posRelativeToMouse.x * posRelativeToMouse.x +
posRelativeToMouse.y * posRelativeToMouse.y
);
var forceDirection = {
x: posRelativeToMouse.x / distance,
y: posRelativeToMouse.y / distance,
};
How hard do we push the particles?
This is an inverse of the distance. Close means a big push, far means a little push. So lets reuse our distance we calculated above.
// distance past which the force is zero
var maxDistance = 1000;
// convert (0...maxDistance) range into a (1...0).
// Close is near 1, far is near 0
// for example:
// 250 => 0.75
// 100 => 0.9
// 10 => 0.99
var force = (maxDistance - distance) / maxDistance;
// if we went below zero, set it to zero.
if (force < 0) force = 0;
Ok we have a direction, and we have the force. All that's left is to apply this to the particle velocity.
particle.xVel += forceDirection.x * force * timeElapsedSinceLastFrame;
particle.yVel += forceDirection.y * force * timeElapsedSinceLastFrame;
And assuming you are animating your position each frame by that xVel/yVel, you should now have particles being pushed away by the mouse.

you can obtain a vector v by subtracting the position of particle from position of mouse,
then you can find the magnitude of this vector my taking sqrt(x^2 + y^2)
by dividing v by magnitude, you obtain a unit vector in the direction you want your particles to go.
for instance.
suppose I have 10 particles in a list U, each has an x and y field.
I can obtain it's vector from each particle v by setting v = (xpart - mousepart, ypart - mousepart)
then you need to find the magnitude vmag by taking sqrt(vx^2 + vy^2)
then you obtain vunit = (vx / vmag, vy / vmag)
This is the vector "away from the mouse".
the rest can be left to detemining speed you want to move at, and ensuring you bounce of walls and such.
I have a similar project at github open source:
https://github.com/dmitrymakhnin/JavaParticleSystem/blob/master/Main.java

Calculate the bounding box's X, Y, Height and Width of a rotated element via JavaScript

Basically I'm asking this question for JavaScript: Calculate Bounding box coordinates from a rotated rectangle
In this case:
iX = Width of rotated (blue) HTML element
iY = Height of rotated (blue) HTML element
bx = Width of Bounding Box (red)
by = Height of Bounding Box (red)
x = X coord of Bounding Box (red)
y = Y coord of Bounding Box (red)
iAngle/t = Angle of rotation of HTML element (blue; not shown but
used in code below), FYI: It's 37 degrees in this example (not that it matters for the example)
How does one calculate the X, Y, Height and Width of a bounding box (all the red numbers) surrounding a rotated HTML element (given its width, height, and Angle of rotation) via JavaScript? A sticky bit to this will be getting the rotated HTML element (blue box)'s original X/Y coords to use as an offset somehow (this is not represented in the code below). This may well have to look at CSS3's transform-origin to determine the center point.
I've got a partial solution, but the calculation of the X/Y coords is not functioning properly...
var boundingBox = function (iX, iY, iAngle) {
var x, y, bx, by, t;
//# Allow for negetive iAngle's that rotate counter clockwise while always ensuring iAngle's < 360
t = ((iAngle < 0 ? 360 - iAngle : iAngle) % 360);
//# Calculate the width (bx) and height (by) of the .boundingBox
//# NOTE: See https://stackoverflow.com/questions/3231176/how-to-get-size-of-a-rotated-rectangle
bx = (iX * Math.sin(iAngle) + iY * Math.cos(iAngle));
by = (iX * Math.cos(iAngle) + iY * Math.sin(iAngle));
//# This part is wrong, as it's re-calculating the iX/iY of the rotated element (blue)
//# we want the x/y of the bounding box (red)
//# NOTE: See https://stackoverflow.com/questions/9971230/calculate-rotated-rectangle-size-from-known-bounding-box-coordinates
x = (1 / (Math.pow(Math.cos(t), 2) - Math.pow(Math.sin(t), 2))) * (bx * Math.cos(t) - by * Math.sin(t));
y = (1 / (Math.pow(Math.cos(t), 2) - Math.pow(Math.sin(t), 2))) * (-bx * Math.sin(t) + by * Math.cos(t));
//# Return an object to the caller representing the x/y and width/height of the calculated .boundingBox
return {
x: parseInt(x), width: parseInt(bx),
y: parseInt(y), height: parseInt(by)
}
};
I feel like I am so close, and yet so far...
Many thanks for any help you can provide!
TO HELP THE NON-JAVASCRIPTERS...
Once the HTML element is rotated, the browser returns a "matrix transform" or "rotation matrix" which seems to be this: rotate(Xdeg) = matrix(cos(X), sin(X), -sin(X), cos(X), 0, 0); See this page for more info.
I have a feeling this will enlighten us on how to get the X,Y of the bounding box (red) based solely on the Width, Height and Angle of the rotated element (blue).
New Info
Humm... interesting...
Each browser seems to handle the rotation differently from an X/Y perspective! FF ignores it completely, IE & Opera draw the bounding box (but its properties are not exposed, ie: bx & by) and Chrome & Safari rotate the rectangle! All are properly reporting the X/Y except FF. So... the X/Y issue seems to exist for FF only! How very odd!
Also of note, it seems that $(document).ready(function () {...}); fires too early for the rotated X/Y to be recognized (which was part of my original problem!). I am rotating the elements directly before the X/Y interrogation calls in $(document).ready(function () {...}); but they don't seem to update until some time after(!?).
When I get a little more time, I will toss up a jFiddle with the example, but I'm using a modified form of "jquery-css-transform.js" so I have a tiny bit of tinkering before the jFiddle...
So... what's up, FireFox? That ain't cool, man!
The Plot Thickens...
Well, FF12 seems to fix the issue with FF11, and now acts like IE and Opera. But now I am back to square one with the X/Y, but at least I think I know why now...
It seems that even though the X/Y is being reported correctly by the browsers for the rotated object, a "ghost" X/Y still exists on the un-rotated version. It seems as though this is the order of operations:
Starting with an un-rotated element at an X,Y of 20,20
Rotate said element, resulting in the reporting of X,Y as 15,35
Move said element via JavaScript/CSS to X,Y 10,10
Browser logically un-rotates element back to 20,20, moves to 10,10 then re-rotates, resulting in an X,Y of 5,25
So... I want the element to end up at 10,10 post rotation, but thanks to the fact that the element is (seemingly) re-rotated post move, the resulting X,Y differs from the set X,Y.
This is my problem! So what I really need is a function to take the desired destination coords (10,10), and work backwards from there to get the starting X,Y coords that will result in the element being rotated into 10,10. At least I know what my problem is now, as thanks to the inner workings of the browsers, it seems with a rotated element 10=5!

I know this is a bit late, but I've written a fiddle for exactly this problem, on an HTML5 canvas:
http://jsfiddle.net/oscarpalacious/ZdQKg/
I hope somebody finds it useful!
I'm actually not calculating your x,y for the upper left corner of the container. It's calculated as a result of the offset (code from the fiddle example):
this.w = Math.sin(this.angulo) * rotador.h + Math.cos(this.angulo) * rotador.w;
this.h = Math.sin(this.angulo) * rotador.w + Math.cos(this.angulo) * rotador.h;
// The offset on a canvas for the upper left corner (x, y) is
// given by the first two parameters for the rect() method:
contexto.rect(-(this.w/2), -(this.h/2), this.w, this.h);
Cheers

Have you tried using getBoundingClientRect() ?
This method returns an object with current values of "bottom, height, left, right, top, width" considering rotations

Turn the four corners into vectors from the center, rotate them, and get the new min/max width/height from them.
EDIT:
I see where you're having problems now. You're doing the calculations using the entire side when you need to be doing them with the offsets from the center of rotation. Yes, this results in four rotated points (which, strangely enough, is exactly as many points as you started with). Between them there will be one minimum X, one maximum X, one minimum Y, and one maximum Y. Those are your bounds.

My gist can help you
Bounding box of a polygon (rectangle, triangle, etc.):
Live demo https://jsfiddle.net/Kolosovsky/tdqv6pk2/
let points = [
{ x: 125, y: 50 },
{ x: 250, y: 65 },
{ x: 300, y: 125 },
{ x: 175, y: 175 },
{ x: 100, y: 125 },
];
let minX = Math.min(...points.map(point => point.x));
let minY = Math.min(...points.map(point => point.y));
let maxX = Math.max(...points.map(point => point.x));
let maxY = Math.max(...points.map(point => point.y));
let pivot = {
x: maxX - ((maxX - minX) / 2),
y: maxY - ((maxY - minY) / 2)
};
let degrees = 90;
let radians = degrees * (Math.PI / 180);
let cos = Math.cos(radians);
let sin = Math.sin(radians);
function rotatePoint(pivot, point, cos, sin) {
return {
x: (cos * (point.x - pivot.x)) - (sin * (point.y - pivot.y)) + pivot.x,
y: (sin * (point.x - pivot.x)) + (cos * (point.y - pivot.y)) + pivot.y
};
}
let boundingBox = {
x1: Number.POSITIVE_INFINITY,
y1: Number.POSITIVE_INFINITY,
x2: Number.NEGATIVE_INFINITY,
y2: Number.NEGATIVE_INFINITY,
};
points.forEach((point) => {
let rotatedPoint = rotatePoint(pivot, point, cos, sin);
boundingBox.x1 = Math.min(boundingBox.x1, rotatedPoint.x);
boundingBox.y1 = Math.min(boundingBox.y1, rotatedPoint.y);
boundingBox.x2 = Math.max(boundingBox.x2, rotatedPoint.x);
boundingBox.y2 = Math.max(boundingBox.y2, rotatedPoint.y);
});
Bounding box of an ellipse:
Live demo https://jsfiddle.net/Kolosovsky/sLc7ynd1/
let centerX = 350;
let centerY = 100;
let radiusX = 100;
let radiusY = 50;
let degrees = 200;
let radians = degrees * (Math.PI / 180);
let radians90 = radians + Math.PI / 2;
let ux = radiusX * Math.cos(radians);
let uy = radiusX * Math.sin(radians);
let vx = radiusY * Math.cos(radians90);
let vy = radiusY * Math.sin(radians90);
let width = Math.sqrt(ux * ux + vx * vx) * 2;
let height = Math.sqrt(uy * uy + vy * vy) * 2;
let x = centerX - (width / 2);
let y = centerY - (height / 2);