rotating a sphere a to b point on itself - javascript

Im trying to figure out how to rotate a sphere from point A on itself to point b on itself. I found some Unity3d code:
Quaternion rot = Quaternion.FromToRotation (pointA, pointB);
sphere.transform.rotation *= rot; //Multiply rotations to get the resultant rotation
via http://answers.unity3d.com/questions/21921/rotate-point-a-on-sphere-to-point-b-on-sphere.html but I can't figure out how to implement it in Three.js.
Here's my code:
var s = sphere mesh
var va = 1st start vector
var vb = 2nd end vector;
var qa = new THREE.Quaternion(va.x, va.y, va.z, 1);
var qb = new THREE.Quaternion(vb.x, vb.y, vb.z, 1);
var qc = new THREE.Quaternion();
THREE.Quaternion.slerp(qa, qb, qc, 1);
s.useQuaternion = true;
s.quaternion = qc;
Thanks!

Assume the sphere is centered at the origin and A and B are normalized (i.e. unit length). Then compute the cross product C = A×B. This is your vector representing your rotation axis. The angle for your rotation is given by θ = cos-1(A∙B) where A∙B is the dot product of the unit vectors A and B. Note that the angle in this case is typically in radians, not degrees.
If the sphere is centered at some point P not at the origin or its radius is not unit length, you will have to translate and scale A and B before derive the rotation. This looks like:
A' ← (A-P)/|A-P| // Normalized version of A
B' ← (B-P)/|B-P| // Normalized version of B
V ← A'×B' // Axis about which to rotate the system
θ ← cos-1(A'∙B') // Angle by which to rotate
You can then use V and θ as your arguments for constructing the quaternion you will use for defining your rotation. This rotation will be centered at the origin, so before applying it, translate by -P, then apply the rotation, and translate back by P.
One note: There may be a sign error in here. If it doesn't work, it's because the sign of the cross product doesn't match with the sign of the dot product. Just reverse the order of the arguments in the cross product to fix it if this is a problem.

var c = group.rotation.y;
var d = -b * (Math.PI / 180)%(2 * Math.PI);
var e = Math.PI / 2 * -1;
group.rotation.y = c % (2 * Math.PI);
group.rotation.x = a * (Math.PI / 180) % Math.PI;
group.rotation.y= d+e;
where a= latitude, b= longitude,group=Object3D(or sphere)

Related

Finding point in a plane at fixed distances from 2 other points in the plane

I have three 3D points defining a plane, P1(x1,y1,z1) P2(x2,y2,z3) and P3(x3,y3,z3). I need to find another point A on that plane which is L1 length from P1 and L2 length from P2 and closest to P3 (out of the 2 possible positions). I have found a solution for this in 2D (https://www.hindawi.com/journals/jr/2010/984823) but I am having difficulty extending it to 3D. I will be implementing this in JS. Any help would be appreciated!
function calculatePointFromLengths(p1,p2,l1,l2) {
let L = sqrt((p2.x-p1.x)*(p2.x-p1.x)+(p2.y-p1.y)*(p2.y-p1.y));
let angleP1P2 = Math.atan((p2.y-p1.y)/(p2.x-p1.x));
let theta1 = Math.acos(l1*l1+l2*l2-l2*l2)/(2*l1*L) + angleP1P2;
// now calculate the point using the angle
return( {
x: p1.x + l1 * Math.cos(theta1);
y: p1.y + l1 * Math.sin(theta1);
});
}
Because you already have angle between P1A and P1P2 vectors
theta = Math.acos(l1*l1+l2*l2-l2*l2)/(2*l1*L)
You can take vector with direction of P1P2 and length l1
V = P1P2 * l1 / L (do the same for every component)
and rotate it by angle theta around axis defined by vector product of P1P2 and P2P3
axis = P1P2 x P2P3
and apply Rodrigues' rotation formula, then get A point as P1 + V_rotated
Alternative way: write A in parametric coordinates
A = P1 + P1P2*u + P1P3*v
Substitute these coordinates into distance equations
l1^2 = (P1P2*u + P1P3*v)^2
l2^2 = (-P2+ P1 + P1P2*u + P1P3*v)^2
and solve this equation system for unknows u,v, choosing solution with 0<u,v<1, u+v<1 (to provide A inside triangle)
I have used an answer from stack exchange in the mathematics group (https://math.stackexchange.com/questions/4548418/finding-point-in-a-plane-at-fixed-distances-from-2-other-points-in-the-plane) and coded it here with a demonstration:
https://playground.babylonjs.com/#UEQUG8#46

How to have isometric projection be a counter-clockwise rotation of flat 2d view

I am making this isometric projection of a 2D table in a Javascript project. I'm calculating my isometric coordinates (from cartesian) with the following function:
var isoX = x - y;
var isoY = (x + y) / 2;
The problem I have is that this rotates everything clock-wise:
AB
CD
becomes
A
C B
D
Instead, I would like it to become
B
A D
C
Can I adjust the function above somehow to achieve this? If not, is there another way to do this?
One idea is to rotate the table before the projection step.
The 2D rotation matrix for 90 degrees counter clockwise is
[ 0 1]
[-1 0]
Putting it in code, this transformation rotates everything around (0,0) 90 degrees ccw:
var transformedX = y;
var transformedY = -x;
Then, simply use the transformed coordinates for the projection.
var isoX = transformedX - transformedY;
var isoY = (transformedX + transformedY) / 2;

Find intersection coordinates of two circles on earth?

I'm trying to find a second intersection point of two circles. One of the points that I already know was used to calculate a distance and then used as the circle radius (exemple). The problem is that im not getting the know point, im getting two new coordinates, even thou they are similar. The problem is probably related to the earth curvature but I have searched for some solution and found nothing.
The circles radius are calculated with the earth curvature. And this is the code I have:
function GET_coordinates_of_circles(position1,r1, position2,r2) {
var deg2rad = function (deg) { return deg * (Math.PI / 180); };
x1=position1.lng;
y1=position1.lat;
x2=position2.lng;
y2=position2.lat;
var centerdx = deg2rad(x1 - x2);
var centerdy = deg2rad(y1 - y2);
var R = Math.sqrt(centerdx * centerdx + centerdy * centerdy);
if (!(Math.abs(r1 - r2) <= R && R <= r1 + r2)) { // no intersection
console.log("nope");
return []; // empty list of results
}
// intersection(s) should exist
var R2 = R*R;
var R4 = R2*R2;
var a = (r1*r1 - r2*r2) / (2 * R2);
var r2r2 = (r1*r1 - r2*r2);
var c = Math.sqrt(2 * (r1*r1 + r2*r2) / R2 - (r2r2 * r2r2) / R4 - 1);
var fx = (x1+x2) / 2 + a * (x2 - x1);
var gx = c * (y2 - y1) / 2;
var ix1 = fx + gx;
var ix2 = fx - gx;
var fy = (y1+y2) / 2 + a * (y2 - y1);
var gy = c * (x1 - x2) / 2;
var iy1 = fy + gy;
var iy2 = fy - gy;
// note if gy == 0 and gx == 0 then the circles are tangent and there is only one solution
// but that one solution will just be duplicated as the code is currently written
return [[iy1, ix1], [iy2, ix2]];
}
The deg2rad variable it is suppose to adjust the other calculations with the earth curvature.
Thank you for any help.
Your calculations for R and so on are wrong because plane Pythagorean formula does not work for spherical trigonometry (for example - we can have triangle with all three right angles on the sphere!). Instead we should use special formulas. Some of them are taken from this page.
At first find big circle arcs in radians for both radii using R = Earth radius = 6,371km
a1 = r1 / R
a2 = r2 / R
And distance (again arc in radians) between circle center using haversine formula
var R = 6371e3; // metres
var φ1 = lat1.toRadians();
var φ2 = lat2.toRadians();
var Δφ = (lat2-lat1).toRadians();
var Δλ = (lon2-lon1).toRadians();
var a = Math.sin(Δφ/2) * Math.sin(Δφ/2) +
Math.cos(φ1) * Math.cos(φ2) *
Math.sin(Δλ/2) * Math.sin(Δλ/2);
var ad = 2 * Math.atan2(Math.sqrt(a), Math.sqrt(1-a));
And bearing from position1 to position 2:
//where φ1,λ1 is the start point, φ2,λ2 the end point
//(Δλ is the difference in longitude)
var y = Math.sin(λ2-λ1) * Math.cos(φ2);
var x = Math.cos(φ1)*Math.sin(φ2) -
Math.sin(φ1)*Math.cos(φ2)*Math.cos(λ2-λ1);
var brng = Math.atan2(y, x);
Now look at the picture from my answer considering equal radii case.
(Here circle radii might be distinct and we should use another approach to find needed arcs)
We have spherical right-angle triangles ACB and FCB (similar to plane case BD is perpendicular to AF in point C and BCA angle is right).
Spherical Pythagorean theorem (from the book on sph. trig) says that
cos(AB) = cos(BC) * cos(AC)
cos(FB) = cos(BC) * cos(FC)
or (using x for AC, y for BC and (ad-x) for FC)
cos(a1) = cos(y) * cos(x)
cos(a2) = cos(y) * cos(ad-x)
divide equations to eliminate cos(y)
cos(a1)*cos(ad-x) = cos(a2) * cos(x)
cos(a1)*(cos(ad)*cos(x) + sin(ad)*sin(x)) = cos(a2) * cos(x)
cos(ad)*cos(x) + sin(ad)*sin(x) = cos(a2) * cos(x) / cos(a1)
sin(ad)*sin(x) = cos(a2) * cos(x) / cos(a1) - cos(ad)*cos(x)
sin(ad)*sin(x) = cos(x) * (cos(a2) / cos(a1) - cos(ad))
TAC = tg(x) = (cos(a2) / cos(a1) - cos(ad)) / sin(ad)
Having hypotenuse and cathetus of ACB triangle we can find angle between AC and AB directions (Napier's rules for right spherical triangles) - note we already know TAC = tg(AC) and a1 = AB
cos(CAB)= tg(AC) * ctg(AB)
CAB = Math.acos(TAC * ctg(a1))
Now we can calculate intersection points - they lie at arc distance a1 from position1 along bearings brng-CAB and brng+CAB
B_bearing = brng - CAB
D_bearing = brng + CAB
Intersection points' coordinates:
var latB = Math.asin( Math.sin(lat1)*Math.cos(a1) +
Math.cos(lat1)*Math.sin(a1)*Math.cos(B_bearing) );
var lonB = lon1.toRad() + Math.atan2(Math.sin(B_bearing)*Math.sin(a1)*Math.cos(lat1),
Math.cos(a1)-Math.sin(lat1)*Math.sin(lat2));
and the same for D_bearing
latB, lonB are in radians
I had a similar need ( Intersection coordinates (lat/lon) of two circles (given the coordinates of the center and the radius) on earth ) and hereby I share the solution in python in case it might help someone:
'''
FINDING THE INTERSECTION COORDINATES (LAT/LON) OF TWO CIRCLES (GIVEN THE COORDINATES OF THE CENTER AND THE RADII)
Many thanks to Ture Pålsson who directed me to the right source, the code below is based on whuber's brilliant work here:
https://gis.stackexchange.com/questions/48937/calculating-intersection-of-two-circles
The idea is that;
1. The points in question are the mutual intersections of three spheres: a sphere centered beneath location x1 (on the
earth's surface) of a given radius, a sphere centered beneath location x2 (on the earth's surface) of a given radius, and
the earth itself, which is a sphere centered at O = (0,0,0) of a given radius.
2. The intersection of each of the first two spheres with the earth's surface is a circle, which defines two planes.
The mutual intersections of all three spheres therefore lies on the intersection of those two planes: a line.
Consequently, the problem is reduced to intersecting a line with a sphere.
Note that "Decimal" is used to have higher precision which is important if the distance between two points are a few
meters.
'''
from decimal import Decimal
from math import cos, sin, sqrt
import math
import numpy as np
def intersection(p1, r1_meter, p2, r2_meter):
# p1 = Coordinates of Point 1: latitude, longitude. This serves as the center of circle 1. Ex: (36.110174, -90.953524)
# r1_meter = Radius of circle 1 in meters
# p2 = Coordinates of Point 2: latitude, longitude. This serves as the center of circle 1. Ex: (36.110174, -90.953524)
# r2_meter = Radius of circle 2 in meters
'''
1. Convert (lat, lon) to (x,y,z) geocentric coordinates.
As usual, because we may choose units of measurement in which the earth has a unit radius
'''
x_p1 = Decimal(cos(math.radians(p1[1]))*cos(math.radians(p1[0]))) # x = cos(lon)*cos(lat)
y_p1 = Decimal(sin(math.radians(p1[1]))*cos(math.radians(p1[0]))) # y = sin(lon)*cos(lat)
z_p1 = Decimal(sin(math.radians(p1[0]))) # z = sin(lat)
x1 = (x_p1, y_p1, z_p1)
x_p2 = Decimal(cos(math.radians(p2[1]))*cos(math.radians(p2[0]))) # x = cos(lon)*cos(lat)
y_p2 = Decimal(sin(math.radians(p2[1]))*cos(math.radians(p2[0]))) # y = sin(lon)*cos(lat)
z_p2 = Decimal(sin(math.radians(p2[0]))) # z = sin(lat)
x2 = (x_p2, y_p2, z_p2)
'''
2. Convert the radii r1 and r2 (which are measured along the sphere) to angles along the sphere.
By definition, one nautical mile (NM) is 1/60 degree of arc (which is pi/180 * 1/60 = 0.0002908888 radians).
'''
r1 = Decimal(math.radians((r1_meter/1852) / 60)) # r1_meter/1852 converts meter to Nautical mile.
r2 = Decimal(math.radians((r2_meter/1852) / 60))
'''
3. The geodesic circle of radius r1 around x1 is the intersection of the earth's surface with an Euclidean sphere
of radius sin(r1) centered at cos(r1)*x1.
4. The plane determined by the intersection of the sphere of radius sin(r1) around cos(r1)*x1 and the earth's surface
is perpendicular to x1 and passes through the point cos(r1)x1, whence its equation is x.x1 = cos(r1)
(the "." represents the usual dot product); likewise for the other plane. There will be a unique point x0 on the
intersection of those two planes that is a linear combination of x1 and x2. Writing x0 = ax1 + b*x2 the two planar
equations are;
cos(r1) = x.x1 = (a*x1 + b*x2).x1 = a + b*(x2.x1)
cos(r2) = x.x2 = (a*x1 + b*x2).x2 = a*(x1.x2) + b
Using the fact that x2.x1 = x1.x2, which I shall write as q, the solution (if it exists) is given by
a = (cos(r1) - cos(r2)*q) / (1 - q^2),
b = (cos(r2) - cos(r1)*q) / (1 - q^2).
'''
q = Decimal(np.dot(x1, x2))
if q**2 != 1 :
a = (Decimal(cos(r1)) - Decimal(cos(r2))*q) / (1 - q**2)
b = (Decimal(cos(r2)) - Decimal(cos(r1))*q) / (1 - q**2)
'''
5. Now all other points on the line of intersection of the two planes differ from x0 by some multiple of a vector
n which is mutually perpendicular to both planes. The cross product n = x1~Cross~x2 does the job provided n is
nonzero: once again, this means that x1 and x2 are neither coincident nor diametrically opposite. (We need to
take care to compute the cross product with high precision, because it involves subtractions with a lot of
cancellation when x1 and x2 are close to each other.)
'''
n = np.cross(x1, x2)
'''
6. Therefore, we seek up to two points of the form x0 + t*n which lie on the earth's surface: that is, their length
equals 1. Equivalently, their squared length is 1:
1 = squared length = (x0 + t*n).(x0 + t*n) = x0.x0 + 2t*x0.n + t^2*n.n = x0.x0 + t^2*n.n
'''
x0_1 = [a*f for f in x1]
x0_2 = [b*f for f in x2]
x0 = [sum(f) for f in zip(x0_1, x0_2)]
'''
The term with x0.n disappears because x0 (being a linear combination of x1 and x2) is perpendicular to n.
The two solutions easily are t = sqrt((1 - x0.x0)/n.n) and its negative. Once again high precision
is called for, because when x1 and x2 are close, x0.x0 is very close to 1, leading to some loss of
floating point precision.
'''
if (np.dot(x0, x0) <= 1) & (np.dot(n,n) != 0): # This is to secure that (1 - np.dot(x0, x0)) / np.dot(n,n) > 0
t = Decimal(sqrt((1 - np.dot(x0, x0)) / np.dot(n,n)))
t1 = t
t2 = -t
i1 = x0 + t1*n
i2 = x0 + t2*n
'''
7. Finally, we may convert these solutions back to (lat, lon) by converting geocentric (x,y,z) to geographic
coordinates. For the longitude, use the generalized arctangent returning values in the range -180 to 180
degrees (in computing applications, this function takes both x and y as arguments rather than just the
ratio y/x; it is sometimes called "ATan2").
'''
i1_lat = math.degrees( math.asin(i1[2]))
i1_lon = math.degrees( math.atan2(i1[1], i1[0] ) )
ip1 = (i1_lat, i1_lon)
i2_lat = math.degrees( math.asin(i2[2]))
i2_lon = math.degrees( math.atan2(i2[1], i2[0] ) )
ip2 = (i2_lat, i2_lon)
return [ip1, ip2]
elif (np.dot(n,n) == 0):
return("The centers of the circles can be neither the same point nor antipodal points.")
else:
return("The circles do not intersect")
else:
return("The centers of the circles can be neither the same point nor antipodal points.")
'''
Example: The output of below is [(36.989311051533505, -88.15142628069133), (38.2383796094578, -92.39048549120287)]
intersection_points = intersection((37.673442, -90.234036), 107.5*1852, (36.109997, -90.953669), 145*1852)
print(intersection_points)
'''
Any feedback is appreciated.

Cesiumjs calculating pitch, yaw, heading from vector

I have one of Cesium's models loaded into the scene, and I have two points which I want to use in order to calculate the orientation of the model and this is the function I created.
// calculate the direction which the model is facing
calculateOrientation({ position, nextPosition }) {
let dir = new Cesium.Cartesian3();
let normalizedDir = new Cesium.Cartesian3();
Cesium.Cartesian3.subtract(nextPosition, position, dir);
Cesium.Cartesian3.normalize(dir, normalizedDir);
var heading = Math.acos(normalizedDir.x);
var pitch = Math.acos(normalizedDir.y);
var roll = 0;
var hpr = new Cesium.HeadingPitchRoll(heading, pitch, roll);
var orientation = Cesium.Transforms.headingPitchRollQuaternion(position, hpr);
return orientation;
}
But the rotations I get don't make any sense. Is my math wrong?
UPDATE
After the first answer by #Keshet, I looked up how to find an angle between a plane and a vector. I figured if I find the angle between the normal of each plane and -90, I should get the correct angle, but I am not sure if this is correct.
Also I don't know how Cesium Axis work, and I cant find any document describing it. For example the XY plane and etc.
let dir = new Cesium.Cartesian3();
let xyNormal = new Cesium.Cartesian3(0,0,1);
let xzNormal = new Cesium.Cartesian3(0,1,0);
let yzNormal = new Cesium.Cartesian3(1,0,0);
Cesium.Cartesian3.subtract(nextPosition, position, dir);
let xyAngle = Cesium.Math.PI_OVER_TWO - Cesium.Cartesian3.angleBetween(dir, xyNormal);
let xzAngle = Cesium.Math.PI_OVER_TWO - Cesium.Cartesian3.angleBetween(dir, xzNormal);
let yzAngle = Cesium.Math.PI_OVER_TWO - Cesium.Cartesian3.angleBetween(dir, yzNormal);
UPDATE 2
Following #IIan suggestion using atan2, here is the code:
Cesium.Cartesian3.subtract(position, nextPosition, dir);
// create the mapped to plane vectors, and get the
// normalized versions
let xyMappedVectorNormalized = new Cesium.Cartesian3(0, 0, 0);
let xyMappedVector = new Cesium.Cartesian3(dir.x, dir.y, 0);
let xzMappedVectorNormalized = new Cesium.Cartesian3(0, 0, 0);
let xzMappedVector = new Cesium.Cartesian3(dir.x, 0, dir.z);
let yzMappedVectorNormalized = new Cesium.Cartesian3(0, 0, 0);
let yzMappedVector = new Cesium.Cartesian3(0, dir.y, dir.z);
Cesium.Cartesian3.normalize(xyMappedVector, xyMappedVectorNormalized);
Cesium.Cartesian3.normalize(xzMappedVector, xzMappedVectorNormalized);
Cesium.Cartesian3.normalize(yzMappedVector, yzMappedVectorNormalized);
// calculate the angles
let xyAngle = Math.atan2(xyMappedVectorNormalized.y, xyMappedVectorNormalized.x);
let xzAngle = Math.atan2(xzMappedVectorNormalized.z, xzMappedVectorNormalized.x);
let yzAngle = Math.atan2(yzMappedVectorNormalized.z, yzMappedVectorNormalized.y);
First, we need to explain what the heading, pitch, and roll angles represent.
The heading angle represents the angle in radians with respect to the XY - plane
The pitch angle represents the angle in radians with respect to the XZ - plane
The roll angle respresents the angle in radians with respect to the YZ plane
You cannot simply use the x / y to compute the heading / pitch
var heading = Math.acos(normalizedDir.x);
var pitch = Math.acos(normalizedDir.y);
You need to get the total angle on each of these planes.
On the XY plane, you will use the normalized |(X, Y)|
on the XZ plane, you will use the normalized |(X, Z)|
on the YZ plane you will use the normalized |(Y, Z)|
Update
|(X, Y)| represents a point on the Unit circle.
where (sin(theta), cos(theta)) = |(X, Y)|
IE when theta = 0, |(X, Y)| = (1, 0)
when theta = PI/2, |(X, Y)| = (0, 1)
this theta would be then the angle you use for the heading
then subsequently, you can call the arctan2 function to calculate the angle with respect to the plane.
atan2(y, x) = theta where [x, y] are calculated from above using the corresponding normalized 2D vector
To note: atan2 gives angles bounded from (-PI, PI]
Example
If your 3D vector is (1, 2, 3) -- on the XY plane X = 1, Y = 2.
Then if you normalize (1, 2) => (1 / sqrt(5), 2 / sqrt(5))
Then you can use atan2(2 / sqrt(5), 1 / sqrt(5) to calculate the angle in radians for the heading

Convert 2:1 equirectangular panorama to cube map

I'm currently working on a simple 3D panorama viewer for a website. For mobile performance reasons I'm using the Three.js CSS 3 renderer. This requires a cube map, split up into six single images.
I'm recording the images on the iPhone with Google Photo Sphere, or similar apps that create 2:1 equirectangular panoramas. I then resize and convert these to a cubemap with this website: http://gonchar.me/panorama/ (Flash)
Preferably, I'd like to do the conversion myself, either on the fly in Three.js, if that's possible, or in Photoshop. I found Andrew Hazelden's Photoshop actions, and they seem kind of close, but no direct conversion is available. Is there a mathematical way to convert these, or some sort of script that does it? I'd like to avoid going through a 3D application like Blender, if possible.
Maybe this is a long shot, but I thought I'd ask. I have okay experience with JavaScript, but I'm pretty new to Three.js. I'm also hesitant to rely on the WebGL functionality, since it seems either slow or buggy on mobile devices. Support is also still spotty.
If you want to do it server side there are many options. ImageMagick has a bunch of command-line tools which could slice your image into pieces. You could put the command to do this into a script and just run that each time you have a new image.
It's hard to tell quite what algorithm is used in the program. We can try and reverse engineer quite what is happening by feeding a square grid into the program. I've used a grid from Wikipedia:
Which gives:
This gives us a clue as to how the box is constructed.
Imaging a sphere with lines of latitude and longitude on it, and a cube surrounding it. Now projecting from the point at center of the sphere produces a distorted grid on the cube.
Mathematically, take polar coordinates r, θ, ø, for the sphere r=1, 0 < θ < π, -π/4 < ø < 7π/4
x= r sin θ cos ø
y= r sin θ sin ø
z= r cos θ
centrally project these to the cube. First we divide into four regions by the latitude -π/4 < ø < π/4, π/4 < ø < 3π/4, 3π/4 < ø < 5π/4, 5π/4 < ø < 7π/4. These will either project to one of the four sides the top or the bottom.
Assume we are in the first side -π/4 < ø < π/4. The central projection of
(sin θ cos ø, sin θ sin ø, cos θ) will be (a sin θ cos ø, a sin θ sin ø, a cos θ) which hits the x=1 plane when
a sin θ cos ø = 1
so
a = 1 / (sin θ cos ø)
and the projected point is
(1, tan ø, cot θ / cos ø)
If | cot θ / cos ø | < 1, this will be on the front face. Otherwise, it will be projected on the top or bottom and you will need a different projection for that. A better test for the top uses the fact that the minimum value of cos ø will be cos π/4 = 1/√2, so the projected point is always on the top if cot θ / (1/√2) > 1 or tan θ < 1/√2. This works out as θ < 35º or 0.615 radians.
Put this together in Python:
import sys
from PIL import Image
from math import pi,sin,cos,tan
def cot(angle):
return 1/tan(angle)
# Project polar coordinates onto a surrounding cube
# assume ranges theta is [0,pi] with 0 the north poll, pi south poll
# phi is in range [0,2pi]
def projection(theta,phi):
if theta<0.615:
return projectTop(theta,phi)
elif theta>2.527:
return projectBottom(theta,phi)
elif phi <= pi/4 or phi > 7*pi/4:
return projectLeft(theta,phi)
elif phi > pi/4 and phi <= 3*pi/4:
return projectFront(theta,phi)
elif phi > 3*pi/4 and phi <= 5*pi/4:
return projectRight(theta,phi)
elif phi > 5*pi/4 and phi <= 7*pi/4:
return projectBack(theta,phi)
def projectLeft(theta,phi):
x = 1
y = tan(phi)
z = cot(theta) / cos(phi)
if z < -1:
return projectBottom(theta,phi)
if z > 1:
return projectTop(theta,phi)
return ("Left",x,y,z)
def projectFront(theta,phi):
x = tan(phi-pi/2)
y = 1
z = cot(theta) / cos(phi-pi/2)
if z < -1:
return projectBottom(theta,phi)
if z > 1:
return projectTop(theta,phi)
return ("Front",x,y,z)
def projectRight(theta,phi):
x = -1
y = tan(phi)
z = -cot(theta) / cos(phi)
if z < -1:
return projectBottom(theta,phi)
if z > 1:
return projectTop(theta,phi)
return ("Right",x,-y,z)
def projectBack(theta,phi):
x = tan(phi-3*pi/2)
y = -1
z = cot(theta) / cos(phi-3*pi/2)
if z < -1:
return projectBottom(theta,phi)
if z > 1:
return projectTop(theta,phi)
return ("Back",-x,y,z)
def projectTop(theta,phi):
# (a sin θ cos ø, a sin θ sin ø, a cos θ) = (x,y,1)
a = 1 / cos(theta)
x = tan(theta) * cos(phi)
y = tan(theta) * sin(phi)
z = 1
return ("Top",x,y,z)
def projectBottom(theta,phi):
# (a sin θ cos ø, a sin θ sin ø, a cos θ) = (x,y,-1)
a = -1 / cos(theta)
x = -tan(theta) * cos(phi)
y = -tan(theta) * sin(phi)
z = -1
return ("Bottom",x,y,z)
# Convert coords in cube to image coords
# coords is a tuple with the side and x,y,z coords
# edge is the length of an edge of the cube in pixels
def cubeToImg(coords,edge):
if coords[0]=="Left":
(x,y) = (int(edge*(coords[2]+1)/2), int(edge*(3-coords[3])/2) )
elif coords[0]=="Front":
(x,y) = (int(edge*(coords[1]+3)/2), int(edge*(3-coords[3])/2) )
elif coords[0]=="Right":
(x,y) = (int(edge*(5-coords[2])/2), int(edge*(3-coords[3])/2) )
elif coords[0]=="Back":
(x,y) = (int(edge*(7-coords[1])/2), int(edge*(3-coords[3])/2) )
elif coords[0]=="Top":
(x,y) = (int(edge*(3-coords[1])/2), int(edge*(1+coords[2])/2) )
elif coords[0]=="Bottom":
(x,y) = (int(edge*(3-coords[1])/2), int(edge*(5-coords[2])/2) )
return (x,y)
# convert the in image to out image
def convert(imgIn,imgOut):
inSize = imgIn.size
outSize = imgOut.size
inPix = imgIn.load()
outPix = imgOut.load()
edge = inSize[0]/4 # the length of each edge in pixels
for i in xrange(inSize[0]):
for j in xrange(inSize[1]):
pixel = inPix[i,j]
phi = i * 2 * pi / inSize[0]
theta = j * pi / inSize[1]
res = projection(theta,phi)
(x,y) = cubeToImg(res,edge)
#if i % 100 == 0 and j % 100 == 0:
# print i,j,phi,theta,res,x,y
if x >= outSize[0]:
#print "x out of range ",x,res
x=outSize[0]-1
if y >= outSize[1]:
#print "y out of range ",y,res
y=outSize[1]-1
outPix[x,y] = pixel
imgIn = Image.open(sys.argv[1])
inSize = imgIn.size
imgOut = Image.new("RGB",(inSize[0],inSize[0]*3/4),"black")
convert(imgIn,imgOut)
imgOut.show()
The projection function takes the theta and phi values and returns coordinates in a cube from -1 to 1 in each direction. The cubeToImg takes the (x,y,z) coordinates and translates them to the output image coordinates.
The above algorithm seems to get the geometry right using an image of buckingham palace. We get:
This seems to get most of the lines in the paving right.
We are getting a few image artefacts. This is due to not having a one-to-one map of pixels. We need to do is use an inverse transformation. Rather than a loop through each pixel in the source and finding the corresponding pixel in the target, we loop through the target images and find the closest corresponding source pixel.
import sys
from PIL import Image
from math import pi,sin,cos,tan,atan2,hypot,floor
from numpy import clip
# get x,y,z coords from out image pixels coords
# i,j are pixel coords
# face is face number
# edge is edge length
def outImgToXYZ(i,j,face,edge):
a = 2.0*float(i)/edge
b = 2.0*float(j)/edge
if face==0: # back
(x,y,z) = (-1.0, 1.0-a, 3.0 - b)
elif face==1: # left
(x,y,z) = (a-3.0, -1.0, 3.0 - b)
elif face==2: # front
(x,y,z) = (1.0, a - 5.0, 3.0 - b)
elif face==3: # right
(x,y,z) = (7.0-a, 1.0, 3.0 - b)
elif face==4: # top
(x,y,z) = (b-1.0, a -5.0, 1.0)
elif face==5: # bottom
(x,y,z) = (5.0-b, a-5.0, -1.0)
return (x,y,z)
# convert using an inverse transformation
def convertBack(imgIn,imgOut):
inSize = imgIn.size
outSize = imgOut.size
inPix = imgIn.load()
outPix = imgOut.load()
edge = inSize[0]/4 # the length of each edge in pixels
for i in xrange(outSize[0]):
face = int(i/edge) # 0 - back, 1 - left 2 - front, 3 - right
if face==2:
rng = xrange(0,edge*3)
else:
rng = xrange(edge,edge*2)
for j in rng:
if j<edge:
face2 = 4 # top
elif j>=2*edge:
face2 = 5 # bottom
else:
face2 = face
(x,y,z) = outImgToXYZ(i,j,face2,edge)
theta = atan2(y,x) # range -pi to pi
r = hypot(x,y)
phi = atan2(z,r) # range -pi/2 to pi/2
# source img coords
uf = ( 2.0*edge*(theta + pi)/pi )
vf = ( 2.0*edge * (pi/2 - phi)/pi)
# Use bilinear interpolation between the four surrounding pixels
ui = floor(uf) # coord of pixel to bottom left
vi = floor(vf)
u2 = ui+1 # coords of pixel to top right
v2 = vi+1
mu = uf-ui # fraction of way across pixel
nu = vf-vi
# Pixel values of four corners
A = inPix[ui % inSize[0],clip(vi,0,inSize[1]-1)]
B = inPix[u2 % inSize[0],clip(vi,0,inSize[1]-1)]
C = inPix[ui % inSize[0],clip(v2,0,inSize[1]-1)]
D = inPix[u2 % inSize[0],clip(v2,0,inSize[1]-1)]
# interpolate
(r,g,b) = (
A[0]*(1-mu)*(1-nu) + B[0]*(mu)*(1-nu) + C[0]*(1-mu)*nu+D[0]*mu*nu,
A[1]*(1-mu)*(1-nu) + B[1]*(mu)*(1-nu) + C[1]*(1-mu)*nu+D[1]*mu*nu,
A[2]*(1-mu)*(1-nu) + B[2]*(mu)*(1-nu) + C[2]*(1-mu)*nu+D[2]*mu*nu )
outPix[i,j] = (int(round(r)),int(round(g)),int(round(b)))
imgIn = Image.open(sys.argv[1])
inSize = imgIn.size
imgOut = Image.new("RGB",(inSize[0],inSize[0]*3/4),"black")
convertBack(imgIn,imgOut)
imgOut.save(sys.argv[1].split('.')[0]+"Out2.png")
imgOut.show()
The results of this are:
If anyone want to go in the reverse, see this JS Fiddle page.
Given the excellent accepted answer, I wanted to add my corresponding C++ implementation, based on OpenCV.
For those not familiar with OpenCV, think of Mat as an image. We first construct two maps that remap from the equirectangular image to our corresponding cubemap face. Then, we do the heavy lifting (i.e., remapping with interpolation) using OpenCV.
The code can be made more compact, if readability is not of concern.
// Define our six cube faces.
// 0 - 3 are side faces, clockwise order
// 4 and 5 are top and bottom, respectively
float faceTransform[6][2] =
{
{0, 0},
{M_PI / 2, 0},
{M_PI, 0},
{-M_PI / 2, 0},
{0, -M_PI / 2},
{0, M_PI / 2}
};
// Map a part of the equirectangular panorama (in) to a cube face
// (face). The ID of the face is given by faceId. The desired
// width and height are given by width and height.
inline void createCubeMapFace(const Mat &in, Mat &face,
int faceId = 0, const int width = -1,
const int height = -1) {
float inWidth = in.cols;
float inHeight = in.rows;
// Allocate map
Mat mapx(height, width, CV_32F);
Mat mapy(height, width, CV_32F);
// Calculate adjacent (ak) and opposite (an) of the
// triangle that is spanned from the sphere center
//to our cube face.
const float an = sin(M_PI / 4);
const float ak = cos(M_PI / 4);
const float ftu = faceTransform[faceId][0];
const float ftv = faceTransform[faceId][1];
// For each point in the target image,
// calculate the corresponding source coordinates.
for(int y = 0; y < height; y++) {
for(int x = 0; x < width; x++) {
// Map face pixel coordinates to [-1, 1] on plane
float nx = (float)y / (float)height - 0.5f;
float ny = (float)x / (float)width - 0.5f;
nx *= 2;
ny *= 2;
// Map [-1, 1] plane coords to [-an, an]
// thats the coordinates in respect to a unit sphere
// that contains our box.
nx *= an;
ny *= an;
float u, v;
// Project from plane to sphere surface.
if(ftv == 0) {
// Center faces
u = atan2(nx, ak);
v = atan2(ny * cos(u), ak);
u += ftu;
} else if(ftv > 0) {
// Bottom face
float d = sqrt(nx * nx + ny * ny);
v = M_PI / 2 - atan2(d, ak);
u = atan2(ny, nx);
} else {
// Top face
float d = sqrt(nx * nx + ny * ny);
v = -M_PI / 2 + atan2(d, ak);
u = atan2(-ny, nx);
}
// Map from angular coordinates to [-1, 1], respectively.
u = u / (M_PI);
v = v / (M_PI / 2);
// Warp around, if our coordinates are out of bounds.
while (v < -1) {
v += 2;
u += 1;
}
while (v > 1) {
v -= 2;
u += 1;
}
while(u < -1) {
u += 2;
}
while(u > 1) {
u -= 2;
}
// Map from [-1, 1] to in texture space
u = u / 2.0f + 0.5f;
v = v / 2.0f + 0.5f;
u = u * (inWidth - 1);
v = v * (inHeight - 1);
// Save the result for this pixel in map
mapx.at<float>(x, y) = u;
mapy.at<float>(x, y) = v;
}
}
// Recreate output image if it has wrong size or type.
if(face.cols != width || face.rows != height ||
face.type() != in.type()) {
face = Mat(width, height, in.type());
}
// Do actual resampling using OpenCV's remap
remap(in, face, mapx, mapy,
CV_INTER_LINEAR, BORDER_CONSTANT, Scalar(0, 0, 0));
}
Given the following input:
The following faces are generated:
Image courtesy of Optonaut.
UPDATE 2: It looks like someone else had already built a far superior web application than my own. Their conversion runs client-side, so there aren't any uploads and downloads to worry about.
I suppose if you hate JavaScript for some reason, or are trying to do this on your mobile, then my web application below is okay.
UPDATE: I've published a simple web application where you can upload a panorama and have it return the six skybox images in a ZIP file.
The source is a cleaned up reimplementation of what's below, and is available on GitHub.
The application is presently running on a single free-tier Heroku dyno, but please don't attempt to use it as an API. If you want automation, make your own deployment; single click Deploy to Heroku available.
Original: Here's a (naively) modified version of Salix Alba's absolutely fantastic answer that converts one face at a time, spits out six different images and preserves the original image's file type.
Aside from the fact most use cases probably expect six separate images, the main advantage of converting one face at a time is that it makes working with huge images a lot less memory intensive.
#!/usr/bin/env python
import sys
from PIL import Image
from math import pi, sin, cos, tan, atan2, hypot, floor
from numpy import clip
# get x,y,z coords from out image pixels coords
# i,j are pixel coords
# faceIdx is face number
# faceSize is edge length
def outImgToXYZ(i, j, faceIdx, faceSize):
a = 2.0 * float(i) / faceSize
b = 2.0 * float(j) / faceSize
if faceIdx == 0: # back
(x,y,z) = (-1.0, 1.0 - a, 1.0 - b)
elif faceIdx == 1: # left
(x,y,z) = (a - 1.0, -1.0, 1.0 - b)
elif faceIdx == 2: # front
(x,y,z) = (1.0, a - 1.0, 1.0 - b)
elif faceIdx == 3: # right
(x,y,z) = (1.0 - a, 1.0, 1.0 - b)
elif faceIdx == 4: # top
(x,y,z) = (b - 1.0, a - 1.0, 1.0)
elif faceIdx == 5: # bottom
(x,y,z) = (1.0 - b, a - 1.0, -1.0)
return (x, y, z)
# convert using an inverse transformation
def convertFace(imgIn, imgOut, faceIdx):
inSize = imgIn.size
outSize = imgOut.size
inPix = imgIn.load()
outPix = imgOut.load()
faceSize = outSize[0]
for xOut in xrange(faceSize):
for yOut in xrange(faceSize):
(x,y,z) = outImgToXYZ(xOut, yOut, faceIdx, faceSize)
theta = atan2(y,x) # range -pi to pi
r = hypot(x,y)
phi = atan2(z,r) # range -pi/2 to pi/2
# source img coords
uf = 0.5 * inSize[0] * (theta + pi) / pi
vf = 0.5 * inSize[0] * (pi/2 - phi) / pi
# Use bilinear interpolation between the four surrounding pixels
ui = floor(uf) # coord of pixel to bottom left
vi = floor(vf)
u2 = ui+1 # coords of pixel to top right
v2 = vi+1
mu = uf-ui # fraction of way across pixel
nu = vf-vi
# Pixel values of four corners
A = inPix[ui % inSize[0], clip(vi, 0, inSize[1]-1)]
B = inPix[u2 % inSize[0], clip(vi, 0, inSize[1]-1)]
C = inPix[ui % inSize[0], clip(v2, 0, inSize[1]-1)]
D = inPix[u2 % inSize[0], clip(v2, 0, inSize[1]-1)]
# interpolate
(r,g,b) = (
A[0]*(1-mu)*(1-nu) + B[0]*(mu)*(1-nu) + C[0]*(1-mu)*nu+D[0]*mu*nu,
A[1]*(1-mu)*(1-nu) + B[1]*(mu)*(1-nu) + C[1]*(1-mu)*nu+D[1]*mu*nu,
A[2]*(1-mu)*(1-nu) + B[2]*(mu)*(1-nu) + C[2]*(1-mu)*nu+D[2]*mu*nu )
outPix[xOut, yOut] = (int(round(r)), int(round(g)), int(round(b)))
imgIn = Image.open(sys.argv[1])
inSize = imgIn.size
faceSize = inSize[0] / 4
components = sys.argv[1].rsplit('.', 2)
FACE_NAMES = {
0: 'back',
1: 'left',
2: 'front',
3: 'right',
4: 'top',
5: 'bottom'
}
for face in xrange(6):
imgOut = Image.new("RGB", (faceSize, faceSize), "black")
convertFace(imgIn, imgOut, face)
imgOut.save(components[0] + "_" + FACE_NAMES[face] + "." + components[1])
I wrote a script to cut the generated cubemap into individual files (posx.png, negx.png, posy.png, negy.png, posz.png and negz.png). It will also pack the 6 files into a .zip file.
The source is here: https://github.com/dankex/compv/blob/master/3d-graphics/skybox/cubemap-cut.py
You can modify the array to set the image files:
name_map = [ \
["", "", "posy", ""],
["negz", "negx", "posz", "posx"],
["", "", "negy", ""]]
The converted files are:
First: unless you really have to convert the images yourself (i.e., because of some specific software requirement), don't.
The reason is that, even though there is a very simple mapping between equirectangular projection and cubic projection, the mapping between the areas is not simple: when you establish a correspondence between a specific point of your destination image and a point in the source with an elementary computation, as soon as you convert both points to pixels by rounding you are doing a very raw approximation that doesn't consider the size of the pixels, and the quality of the image is bound to be low.
Second: even if you need to do the conversion at runtime, are you sure that you need to do the conversion at all? Unless there is some very stringent performance problem, if you just need a skybox, create a very big sphere, stitch the equirectangular texure on it, and off you go. Three.js provides the sphere already, as far as I remember ;-)
Third: NASA provides a tool to convert between all conceivable projections (I just found out, tested it, and works like a charm). You can find it here:
G.Projector — Global Map Projector
And I find it reasonable to think that the guys know what they are doing ;-)
It turns out that the "guys" know what they do up to some point: the generated cubemap has an hideous border which makes the conversion not that easy...
I found the definitive tool for equirectangular to cubemap conversion, and it's called erect2cubic.
It's a small utility that generates a script to be fed to hugin, in this way:
erect2cubic --erect=input.png --ptofile=cube.pto
nona -o cube_prefix cube.pto
(information siphoned from Vinay's Hacks page)
And it will generate all six cubemap faces. I'm using it for my project and it works like a charm!
The only downside of this approach is that the script erect2cubit is not in the standard Ubuntu distribution (which is what I'm using) and I had to resort to the instructions at a blog describing how to install and use erect2cubic to find out how to install it.
It is totally worth it!
cmft Studio supports conversion/filtering of various HDR/LDR projections to cubemaps.
https://github.com/dariomanesku/cmftStudio
Here's a JavaScript version of Benjamin Dobell's code. The convertFace needs to be passed two ìmageData objects and a face ID (0-6).
The provided code can safely be used in a web worker, since it has no dependencies.
// convert using an inverse transformation
function convertFace(imgIn, imgOut, faceIdx) {
var inPix = shimImgData(imgIn),
outPix = shimImgData(imgOut),
faceSize = imgOut.width,
pi = Math.PI,
pi_2 = pi/2;
for(var xOut=0; xOut<faceSize; xOut++) {
for(var yOut=0; yOut<faceSize; yOut++) {
var xyz = outImgToXYZ(xOut, yOut, faceIdx, faceSize);
var theta = Math.atan2(xyz.y, xyz.x); // range -pi to pi
var r = Math.hypot(xyz.x, xyz.y);
var phi = Math.atan2(xyz.z, r); // range -pi/2 to pi/2
// source image coordinates
var uf = 0.5 * imgIn.width * (theta + pi) / pi;
var vf = 0.5 * imgIn.width * (pi_2 - phi) / pi;
// Use bilinear interpolation between the four surrounding pixels
var ui = Math.floor(uf); // coordinate of pixel to bottom left
var vi = Math.floor(vf);
var u2 = ui + 1; // coordinates of pixel to top right
var v2 = vi + 1;
var mu = uf - ui; // fraction of way across pixel
var nu = vf - vi;
// Pixel values of four corners
var A = inPix.getPx(ui % imgIn.width, clip(vi, 0, imgIn.height-1));
var B = inPix.getPx(u2 % imgIn.width, clip(vi, 0, imgIn.height-1));
var C = inPix.getPx(ui % imgIn.width, clip(v2, 0, imgIn.height-1));
var D = inPix.getPx(u2 % imgIn.width, clip(v2, 0, imgIn.height-1));
// interpolate
var rgb = {
r:A[0]*(1-mu)*(1-nu) + B[0]*(mu)*(1-nu) + C[0]*(1-mu)*nu + D[0]*mu*nu,
g:A[1]*(1-mu)*(1-nu) + B[1]*(mu)*(1-nu) + C[1]*(1-mu)*nu + D[1]*mu*nu,
b:A[2]*(1-mu)*(1-nu) + B[2]*(mu)*(1-nu) + C[2]*(1-mu)*nu + D[2]*mu*nu
};
rgb.r = Math.round(rgb.r);
rgb.g = Math.round(rgb.g);
rgb.b = Math.round(rgb.b);
outPix.setPx(xOut, yOut, rgb);
} // for(var yOut=0; yOut<faceSize; yOut++) {...}
} // for(var xOut=0;xOut<faceSize;xOut++) {...}
} // function convertFace(imgIn, imgOut, faceIdx) {...}
// get x, y, z coordinates from out image pixels coordinates
// i,j are pixel coordinates
// faceIdx is face number
// faceSize is edge length
function outImgToXYZ(i, j, faceIdx, faceSize) {
var a = 2 * i / faceSize,
b = 2 * j / faceSize;
switch(faceIdx) {
case 0: // back
return({x:-1, y:1-a, z:1-b});
case 1: // left
return({x:a-1, y:-1, z:1-b});
case 2: // front
return({x: 1, y:a-1, z:1-b});
case 3: // right
return({x:1-a, y:1, z:1-b});
case 4: // top
return({x:b-1, y:a-1, z:1});
case 5: // bottom
return({x:1-b, y:a-1, z:-1});
}
} // function outImgToXYZ(i, j, faceIdx, faceSize) {...}
function clip(val, min, max) {
return(val<min ? min : (val>max ? max : val));
}
function shimImgData(imgData) {
var w = imgData.width*4,
d = imgData.data;
return({
getPx:function(x, y) {
x = x*4 + y*w;
return([d[x], d[x+1], d[x+2]]);
},
setPx:function(x, y, rgb) {
x = x*4 + y*w;
d[x] = rgb.r;
d[x+1] = rgb.g;
d[x+2] = rgb.b;
d[x+3] = 255; // alpha
}
});
} // function shimImgData(imgData) {...}
I created a solution for this problem using OpenGL and made a command-line tool around it. It works both on images and videos, and it is the fastest tool that I found out there.
Convert360 - Project on GitHub.
OpenGL Shader - The fragment shader used for the re-projection.
The usage is as simple as:
pip install convert360
convert360 -i ~/Pictures/Barcelona/sagrada-familia.jpg -o example.png -s 300 300
To get something like this:
There are various representations of environment maps. Here is a nice overview.
Overview - Panoramic Images
If you use Photosphere (or any panorama app for that matter), you most likely already have the horizontal latitude / longitude representation.
You can then simply draw a textured three.js SphereGeometry. Here is a tutorial on how to render earth.
Tutorial - How to Make the Earth in WebGL?
Best of luck :).
A very simple C++ application which converts an equirectangular panorama to cube map based on the answer by Salix Alba:
Photo Panorama Converter
Perhaps I am missing something here. But it seems that most if not all the presented transformation code may be somewhat incorrect. They take a spherical panorama (equirectangular --- 360 deg horizontally and 180 deg vertically) and seem to convert to the cube faces using a Cartesian <-> cylindrical transformation. Should they not be using a Cartesian <-> spherical transformation?
See Spherical Coordinates.
I suppose that as long as they reverse the calculation to go from the cube faces to the panorama, then it should work out. But the images of the cube faces may be slightly different when using the spherical transformation.
If I start with this equirectangular (spherical panorama):
Then if I use a cylindrical transformation (which I am not 100% sure is correct at this time), I get this result:
But if I use a spherical transformation, I get this result:
They are not the same. But my spherical transformation result seems to match the result of Danke Xie, but his link does not show the kind of transformation he is using, as best I can read it.
So am I misunderstanding the code being used by many of the contributors to this topic?
kubi can convert from an equirectangular image to cube faces. I have written it to be fast and flexible. It provides options to choose the output layout (six separate images is the default) and decide on the resampling method.

Categories