I have a CubeGeometry which the camera is looking at, and I want the camera to zoom so that the cube is entirely visible, but no larger.
My initial attempt was to convert the cube verticies to the camera coordinate system,
function toScreenXY(position, camera) {
var pos = position.clone();
var projScreenMat = new THREE.Matrix4();
projScreenMat.multiply(camera.projectionMatrix, camera.matrixWorldInverse);
projScreenMat.multiplyVector3( pos );
return pos;
}
function ScaleInView() {
camera.fov = 0.0;
for (var i=0; i<8; i++) {
proj2d = toScreenXY(cube.geometry.vertices[i],camera);
angle = 57.296 * Math.max(Math.atan(proj2d.x/proj2d.z), Math.atan(proj2d.y/proj2d.z));
camera.fov = Math.max(camera.fov,angle);
}
camera.updateProjectionMatrix();
}
I thought this would work, but sometimes it's too small, and other times too large (depending on the position of the camera).
I also need to do this for Orthographic Camera.
Edit:
I know how to do this when the cube is facing the camera, I'm looking for a way to do it when the camera is moved to some arbitrary (r, theta, phi) position (spherical polar coordinates; r is actually constant for my purposes).
Perspective Camera. If the camera is centered and viewing the cube head-on, define
dist = distance from the camera to the front face ( important ! ) of the cube
and
height = height of the cube.
If you set the camera field-of-view as follows
fov = 2 * Math.atan( height / ( 2 * dist ) ) * ( 180 / Math.PI );
then the cube height will match the visible height.
Orthographic Camera. If the camera is centered and viewing the cube head-on, define
aspect = the aspect ratio of your window (i.e., width / height)
and
height = height of the cube.
Then construct your camera this way:
camera = new THREE.OrthographicCamera( -aspect * height/2, aspect * height/2, height/2, -height/2, near, far );
The cube height will match the visible height.
In either case, if the camera is not centered, or is otherwise viewing the cube at an angle, then the problem is more complicated.
Also, if the window is narrower than it is high, then the width is the constraining factor, and the problem is more complicated.
Multiplying by camera.matrixWorldInverse gives a vector in the camera's coordinates, but importantly does not apply perspective.
function toCameraCoords(position) {
return camera.matrixWorldInverse.multiplyVector3(position.clone());
}
We can then find the smallest angle that will fit all the box corners in the scene. arctan(D.x / D.z) gives the angle BCD where B is what the camera is looking at, C is the camera's position, and D the position of an object that you want to be visible in the camera coordinates.
In my case, the following ensures that the the cube boundbox is fully visible.
function ScaleInView() {
var tmp_fov = 0.0;
for (var i=0; i<8; i++) {
proj2d = toCameraCoords(boundbox.geometry.vertices[i]);
angle = 114.59 * Math.max( // 2 * (Pi / 180)
Math.abs(Math.atan(proj2d.x/proj2d.z) / camera.aspect),
Math.abs(Math.atan(proj2d.y/proj2d.z))
);
tmp_fov = Math.max(tmp_fov, angle);
}
camera.fov = tmp_fov + 5; // An extra 5 degrees keeps all lines visible
camera.updateProjectionMatrix();
}
Related
I am adding a plane to the scene like this:
// Camera
this.three.camera = new THREE.PerspectiveCamera(45, window.innerWidth/window.innerHeight, 0.1, 60);
// Plane
const planeGeometry = new THREE.PlaneBufferGeometry(1,1,this.options.planeSegments,this.options.planeSegments);
const planeMat = new THREE.ShaderMaterial( ... )
this.three.plane = new THREE.Mesh(planeGeometry,planeMat);
this.three.scene.add(this.three.plane);
Pretty basic. I am than trying to find out how I have to move the plane in the Z axis for it to fill the browser-viewport. For that,
// See attachment "solving for this" is closeZ
const closeZ = 0.5 / Math.tan((this.three.camera.fov/2.0) * Math.PI / 180.0);
this.uniforms.uZMax = new THREE.Uniform(this.three.camera.position.z - closeZ);
So now I know in my shader how much I can add to Z to make the plane fill the viewport. Vertex Shader looks like this:
uniform float uZMax;
void main() {
vec3 pos = (position.xy, uZMax);
gl_Position = projectionMatrix * modelViewMatrix * vec4( pos, 1 );
}
This actually zoom the plane to fill the viewport, but in Y-Axis, not in X-Axis.
I would like to discover why my math is referring to the Y-Axis and how I need to transform it, so the plane will fill the viewport width instead of it's height?
Edit:
I'm trying to achieve something like this https://tympanus.net/Tutorials/GridToFullscreenAnimations/index4.html - But in the given example they're just scaling the x- and y-pixels to fill the screen and therefore no actual 3d - and therefore again no lighting is going on.
I want to actually move the plane towards the camera using different z-values so I can calculate surface normals to then again calculate lighting in the fragment shader by how aligned the normal is with the light direction - like it's done in raymarching.
You can easily achieve such a fullscreen effect by using the following setup:
const camera = new THREE.OrthographicCamera( - 1, 1, 1, - 1, 0, 1 );
const geometry = new THREE.PlaneBufferGeometry( 2, 2 );
When creating a mesh with this geometry and a custom shader material, the orthographic camera will ensure the intended fullscreen effect. This approach is used in all post-processing example where the entire viewport has to be filled with a single quad.
I figured it out, and as suspected it has to do with the aspect ratio passed to the camera. For anyone looking for a solution after me, here is how it works:
I wrongly assumed that the field-of-value for the camera is the same in all directions. But the FOV is referring to the Y-Axis FOV, so we have to convert the camera-fov to the x-axis also:
function getXFOV() {
// Convert angle to radiant
const FOV = this.three.camera.fov;
let yFovRadiant = FOV * Math.PI/180;
// Calculate X-FOV Radiant
let xFovRadiant = 2 * Math.atan( Math.tan(yFovRadiant/2) * (window.innerWidth / window.innerHeight));
// Convert back to angle
let xFovAngle = xFovRadiant * 180/Math.PI;
return xFovAngle;
}
And then we simply use that angle in in the closeZ-calculation instead of the camera's fov. Now it snaps to the window-width.
const closeZ = 0.5 / Math.tan((this.getXFOV()) * Math.PI / 180.0);
this.uniforms.uZMax = new THREE.Uniform(this.three.camera.position.z - closeZ);
I have some code that converts a perspective camera to an orthographic camera. The problem is that when I make the conversion, the model becomes very tiny and hard to see.
I have calculated the zoom factor for the orthographic camera, based on the distance and the FOV. Are there any other properties that I need to set on the orthographic camera (e.g. clipping plane, etc..)?
I believe the position remains the same. I'm not sure what else I need to calculate.
fieldOfView = viewInfo.fov;
var getCameraPosition = function() {
return viewer._viewport._implementation.getCamera()._nativeCamera.position;
};
// Calculate the delta position between the camera and the object
var getPositionDelta = function(position1, position2) {
return {
x: position1.x - position2.x,
y: position1.y - position2.y,
z: position1.z - position2.z
}
};
var getDistance = function(positionDelta, cameraDirection) {
return dot(positionDelta, cameraDirection);
};
distance = getDistance(positionDelta, cameraDirection),
var depth = distance;
var viewportWidth = view.getDomRef().getBoundingClientRect().width;
var viewportHeight = view.getDomRef().getBoundingClientRect().height;
var aspect = viewportWidth / viewportHeight;
var height_ortho = depth * 2 * Math.atan( fieldOfView * (Math.PI/180) / 2 )
var width_ortho = height_ortho * aspect;
var near = viewInfo.near, far = viewInfo.far;
var newCamera = new THREE.OrthographicCamera(
width_ortho / -2, width_ortho / 2,
height_ortho / 2, height_ortho / -2,
near, far );
newCamera.position.copy( viewInfo.position );
var sCamera = new vk.threejs.OrthographicCamera(); //framework creatio of threejs cam
sCamera.setZoomFactor(orthoZoomFactor);
sCamera.setCameraRef(newCamera);
view.getViewport().setCamera(sCamera);
I also tried setting the same camera properties (e.g. clipping planes etc) of the perspective for the orthographic and I still had the same problem.
I guess I am missing some property or calculation required to put the object in the same position as when it was in perspective camera view.
Let's assume you have a perspective view with a given vertical field of view angle fov_y (in degrees) and you know the size of the viewport width and height. Furthermore, you have the near and far plane. These are the values which you use to setup the THREE.PerspectiveCamera:
perspCamera = new THREE.PerspectiveCamera( fov_y, width / height, near, far );
Also, you know the position of the object and the position of the camera. An object doesn't have only a single position, but you have to choose a representative position for its depth.
First you have to calculate the depth of the object.
var v3_object = .... // THREE.Vector3 : positon of the object
var v3_camera = perspCamera.position;
var line_of_sight = new THREE.Vector3();
perspCamera.getWorldDirection( line_of_sight );
var v3_distance = v3_object.clone().sub( v3_camera );
depth = v3_distance.dot( line_of_sight );
Then you have to calculate the "size" of the rectangle which is projected to the viewport at the depth:
aspect = width / height;
height_ortho = depth * 2 * Math.atan( fov_y*(Math.PI/180) / 2 )
width_ortho = height_ortho * aspect;
With these values the THREE.OrthographicCamera can be setup like this:
var orthoCamera = new THREE.OrthographicCamera(
width_ortho / -2, width_ortho / 2,
height_ortho / 2, height_ortho / -2,
near, far );
orthoCamera.position.copy( perspCamera.position );
The positon and direction of the perspective camera can be committed to the orthographic camera like this:
orthoCamera.position.copy( perspCamera.position );
orthoCamera.quaternion.copy( perspCamera.quaternion );
See also stackoverflow question Three.js - Find the current LookAt of a camera?
I am making this program where you can click on an object, zoom to it, then look at it from all angles by holding the right mouse button and dragging. I need the camera to be going around the object, not rotate the object with the camera looking at it. I honestly just have no idea how to math it out!
For testing there is already a game object with an xyz we have selected and are looking at
var g = new GameObject(500, 0, 0);//The game object with xyz
this.selected = g;//set selected to g
//Create and set the camera
this.camera = new THREE.PerspectiveCamera(45, w/h, 1, 10000);
this.camera.position.x = 0;
this.camera.position.y = 0;
this.camera.position.z = 0;
//set camera to look at the object which is 500 away in the x direction
this.camera.lookAt(new THREE.Vector3(this.selected.x, this.selected.y, this.selected.z));
So the radius between the camera and the object is 500 and while selected and rotating, the camera should always be 500 away.
I update the scene here:
Main.prototype.update = function(){
this.renderer.render(this.scene, this.camera);//scene is just some ambient lighting
//what to do when mouse right is held down
if(this.rightMouseDown){
//placeholder functionality, needs to rotate around object based on mouse movements
this.camera.position.x -= 5;
}
}
How do I rotate this camera around g with a radius of 500?!?!
As gaitat mentioned, trackball controls are the best place to start with many configurable parameters to make camera rotation/revolution easy. One enormous potential benefit of this method ( especially for your project ) is avoiding "gimbal lock" which is the source of much frustration when working with rotations. Here's a link that might help you with Trackball controls and Orbitcontrols:
Rotate camera in Three.js with mouse
Another option would be setting camera coordinates yourself in the animation loop which is actually quite simple:
var angle = 0;
var radius = 500;
function animate() {
...
// Use Math.cos and Math.sin to set camera X and Z values based on angle.
camera.position.x = radius * Math.cos( angle );
camera.position.z = radius * Math.sin( angle );
angle += 0.01;
...
}
Another option would be to connect the camera to a pivot object and just rotate the pivot:
var camera_pivot = new THREE.Object3D()
var Y_AXIS = new THREE.Vector3( 0, 1, 0 );
scene.add( camera_pivot );
camera_pivot.add( camera );
camera.position.set( 500, 0, 0 );
camera.lookAt( camera_pivot.position );
...
camera_pivot.rotateOnAxis( Y_AXIS, 0.01 ); // radians
If you pursue this option, be aware that the camera object is in "camera pivot space", and might be more challenging to manipulate further.
I've seen a lot of posts people want to have the camera position to screen position. My question is how to do the contrary.
What I currently want to achieve is set the "door" position to a % of the screen, this calculation is ready, and I do have the final screen X, Y (px) position. The current Z offset = 250 of the camera.
I've found this code to convert camera position to screen position:
var vector = projector.projectVector(door.position.clone(), camera);
vector.x = (vector.x + 1) / 2 * window.innerWidth;
vector.y = -(vector.y - 1) / 2 * window.innerHeight;
To do the reversie I tried this, but does not give the result I expect:
var mouseX = ((perc.x) / window.innerWidth) * 2 - 1,
mouseY = -((perc.y) / window.innerHeight) * 2 + 1;
var vector = new THREE.Vector3(mouseX, mouseY, 1);
projector.unprojectVector(vector, camera);
return vector.sub(camera.position).normalize()
I have tried several ways and tried to Google, but didn't found an answer.
Q: How to convert screen position to camera position?
Well, as posted by Gero3, what you want to do is to create a plane that will cover all of your visible camera area (something like new THREE.PlaneGeometry(5000,5000); and then use raycaster to locate screen coordinated on that plane.
Here is an example from another similar question:
http://jsfiddle.net/PwWbT/
Hope that helps.
Use a large invisble plane that defines in which direction the door can expand and then use a raycaster on the plane to search for the position to where the door object should extend.
I've working on an app which displays some 3D models. We load the models, create the meshes, add them to the scene...standard procedure. After the last mesh is added, we compute the bounding box in order to move the camera and cover all the scene, using the size of the total geometry and the size of the viewport to do the math.
if (bounds.bx / bounds.by < camera.aspect) {
/* Vertical max */
r = bounds.by / (2 * Math.tan(Math.PI / 8));
} else {
/* Horizontal max */
hFOV = 2 * Math.atan(Math.tan(Math.PI / 8) * camera.aspect);
r = bounds.bx / (2 * Math.tan((hFOV / 2)));
}
bounds is an object containing the width and height of the bounding box. After this calculation, we move the camera(plus a little ratio, just aesthetics, we want a little room between the geometry and the screen border :) ) and render
camera.position.z = r * 1.05;
So far this is implemented and runs ok. This has been done with PerspectiveCamera. Now we want to change that and use OrthographicCamera...turns out to be a mess. Models are too small, we lose the mousewheel zoom from the TrackBall Controls and the algorithm to move the camera is not working anymore. Also I don't understand the parameters of the constructor for the camera...these width and height are for the geometry or the viewport?
The pattern for instantiating an orthographic camera in three.js is:
var camera = new THREE.OrthographicCamera( width / - 2, width / 2, height / 2, height / - 2, near, far );
where width and height are the width and height of the camera's cuboid-shaped frustum measured in world-space units.
near and far are the world-space distances to the near and far planes of the frustum. Both near and far should be greater than zero.
To prevent distortion, you will typically want the aspect ratio of the orthographic camera ( width / height ) to match the aspect ratio of the render's canvas. (see *Note below)
It is unfortunate that many of the three.js examples pass window.innerWidth and window.innerHeight as args to this constructor. Doing so only makes sense if the orthographic camera is used for rendering to a texture, or if the world units for your orthographic scene are in pixels.
*Note: Actually, the camera aspect ratio should match the aspect ratio of the renderer's viewport. The viewport can be a sub-region of the canvas. If you do not set the renderer's viewport directly using renderer.setViewport(), the viewport will be the same size as the canvas, and hence have the same aspect ratio as the canvas.
three.js r.73
For future reference:
Updated video
var w = container.clientWidth;
var h = container.clientHeight;
var viewSize = h;
var aspectRatio = w / h;
_viewport = {
viewSize: viewSize,
aspectRatio: aspectRatio,
left: (-aspectRatio * viewSize) / 2,
right: (aspectRatio * viewSize) / 2,
top: viewSize / 2,
bottom: -viewSize / 2,
near: -100,
far: 100
}
_camera = new THREE.OrthographicCamera (
_viewport.left,
_viewport.right,
_viewport.top,
_viewport.bottom,
_viewport.near,
_viewport.far
);
In my specific case, my world units are pixels. Hence, I am using container.clientWidth and container.clientHeight as width and height. You probably don't want to do this.
camera.top = (.95*camera.top);
camera.bottom = (.95*camera.bottom);
camera.left = (.95*camera.left);
camera.right = (.95*camera.right);