Develop Reference

Issues with calculating normals of a triangle

I'm rendering some basic triangles, and I'm generating the normals in code. This is my "calculate normals" method. It gets the face normal of a triangle from its three vertices:
public static Vector3f calculateNormal(Vector3f v0, Vector3f v1, Vector3f v2) {
Vector3f u = Vector3f.sub(v2, v0, null);
Vector3f w = Vector3f.sub(v1, v0, null);
Vector3f n = new Vector3f();
Vector3f.cross(u, w, n);
n.normalise(n);
if (Float.isNaN(n.x) || Float.isNaN(n.y) || Float.isNaN(n.z)) {
System.out.println("It's NaN!");
return new Vector3f(0,1,0);
}
return n;
}
Except it outputs "NaN" for about half the triangles. I tried switching the order of the vertices, and that doesn't do anything.

Your calculateNormal function looks okay to me :)
But, the triangles with NaN normals, probably have co-linear vertices (i.e., they are degenerated triangles). Try checking if n.length is zero (or almost zero) to detect this edge case.

Without knowing which math library you are using, I cannot answer specifically. However, to find the cross product with 3 points, you need to find the vector from one point to the other two.
Vector3f toPoint2 = point2 - point1;
Vector3f toPoint3 = point3 - point1;
Once you have the two vectors, you need to find their cross product, which is orthogonal to both toPoint2 and toPoint3. In order to find the cross product, find the determinate of the following matrix:
[i j k ]
[toPoint2.x toPoint2.y toPoint2.z]
[toPoint3.x toPoint3.y toPoint3.z]
The cross product is:
Vector3f vec = new Vector3f(toPoint2.y * toPoint3.z - toPoint2.z * toPoint3.y,
toPoint2.z * toPoint3.x - toPoint2.x * toPoint3.z,
toPoint2.x * toPoint3.y - toPoint2.y * toPoint3.x)
This cross product will result in the normal vector of point1 with respect to point2 and point3.

Ray-tracing triangles

I am writing a raytracer in java, and I was able to get tracing of spheres working, but I believe I have something wrong with how I am tracing triangles.
Here is the basic algorithm, as I understand it:
First determine if the ray even intersects the plane that the triangle is on.
Clip all points so they are on the same plane as the triangle (so to the xy plane, as an example).
Determine if the potential intersection point falls inside or out of the triangle, based on the number of polygon edges you cross when sending out a ray in an arbitrary direction along the new plane.
Now, here is my implementation of that (specifically the first point):
public Vector getIntersectionVector(Ray ray)
{
Vector planeIntersectionVector = getPlaneIntersectionVector(ray, getPlaneNormal());
if (planeIntersectionVector != null)
{
if (isIntersectionVectorInsideTriangle(planeIntersectionVector))
{
return planeIntersectionVector;
}
else
{
return null;
}
}
else
{
return null;
}
}
Where getPlaceIntersectionVector() is:
private Vector getPlaneIntersectionVector(Ray ray, Vector planeNormal)
{
double vd = planeNormal.dotProduct(ray.getDirection());
//(p_n \dot r_d) == 0, parallel. (p_n \dot r_d) > 0 Plane normal pointing away from ray.
if (vd >= 0)
{
return null;
}
double distance = planeNormal.distance(0d, 0d, 0d);
double vo = -(planeNormal.dotProduct(ray.getOrigin()) + distance);
double intersectionDistance = vo / vd;
//intersectionDistance <= 0 means the "intersection" is behind the ray, so not a real intersection
return (intersectionDistance <= 0) ? null : ray.getLocation(intersectionDistance);
}
Which basically tries to mimic this:
And this:
Where t is the distance along the ray that the point hits, ro is the origin of the ray, rd is the direction of the ray, pn refers to the plane normal of the triangle/plane, and d is the distance from the plane that the triangle is on to the origin (0,0,0)
Am I doing that wrong? When I send out the ray from the first pixel in the image (0,0), I am seeing that the intersectionDistance (or t) is almost 1100, which intuitively seems wrong to me. I would think that the intersection point would be much closer.
Here is the relevant data:
Ray origin (0,0,1), Ray Direction is roughly (0.000917, -0.4689, -0.8833).
Triangle has vertices as (-0.2, 0.1, 0.1), (-0.2, -0.5, 0.2), (-0.2, 0.1, -0.3), which makes the plane normal (-1, 0, 0).
According to my code, the Ray intersects the plane 1090 distance away, which as I mentioned before, seems wrong to me. The scene is only -1.0 to 1.0 in every direction, which means the intersection is very very far in the distance.
Am I doing the plane intersection wrong?
Please let me know where to clarify points, and if you need any more information.

The problem is this line:
double distance = planeNormal.distance(0d, 0d, 0d);
A plane is defined by a normal and a distance from the plane to the origin. The distance of a vector from the origin is the length of a vector, so what you are calculating there is just the length of the plane normal (which will always be 1.0 if it has been normalized).
The distance of the plane from the origin is an extra piece of information that needs to be passed into your function, you can't just calculate it from the normal because it's independent of it (you can can have lots of planes with normals pointing in the same direction at different distances from the origin).
So define your function something like this:
private Vector getPlaneIntersectionVector(Ray ray, Vector planeNormal, double planeDistance)
and call it something like this:
Vector planeIntersectionVector = getPlaneIntersectionVector(ray, getPlaneNormal(), getPlaneDistance());
Then you can calculate vo like this:
double vo = -(planeNormal.dotProduct(ray.getOrigin()) + planeDistance);

Slightly different approach:
Let's triangle vertices are V0, V1, V2
Edge vectors are
A = V1-V0
B = V2 - V0
Ray has parametric equation (as you wrote)
P = R0 + t * Rd
From the other side, intersection point has parametric coordinates u, v in the triangle plane
P = V0 + u * A + v * B
So you can write system of three linear equation for x, y, z coordinates and solve it for t, u, v. If determinant ius non-zero (ray is not parallel to the plane), and t >=0, and u, v, u+v lie in range 0..1, then P is inside triangle.
R0.X + t * Rd.X = V0.X + u * A.X + v * B.X
R0.Y + t * Rd.Y = V0.Y + u * A.Y + v * B.Y
R0.Z + t * Rd.Z = V0.Z + u * A.Z + v * B.Z

Issues with Raytracing triangles (orientation and coloring)

EDIT: I found out that all the pixels were upside down because of the difference between screen and world coordinates, so that is no longer a problem.
EDIT: After following a suggestion from #TheVee (using absolute values), my image got much better, but I'm still seeing issues with color.
I having a little trouble with ray-tracing triangles. This is a follow-up to my previous question about the same topic. The answers to that question made me realize that I needed to take a different approach. The new approach I took worked much better, but I'm seeing a couple of issues with my raytracer now:
There is one triangle that never renders in color (it is always black, even though it's color is supposed to be yellow).
Here is what I am expecting to see:
But here is what I am actually seeing:
Addressing debugging the first problem, even if I remove all other objects (including the blue triangle), the yellow triangle is always rendered black, so I don't believe that it is an issues with my shadow rays that I am sending out. I suspect that it has to do with the angle that the triangle/plane is at relative to the camera.
Here is my process for ray-tracing triangles which is based off of the process in this website.
Determine if the ray intersects the plane.
If it does, determine if the ray intersects inside of the triangle (using parametric coordinates).
Here is the code for determining if the ray hits the plane:
private Vector getPlaneIntersectionVector(Ray ray)
{
double epsilon = 0.00000001;
Vector w0 = ray.getOrigin().subtract(getB());
double numerator = -(getPlaneNormal().dotProduct(w0));
double denominator = getPlaneNormal().dotProduct(ray.getDirection());
//ray is parallel to triangle plane
if (Math.abs(denominator) < epsilon)
{
//ray lies in triangle plane
if (numerator == 0)
{
return null;
}
//ray is disjoint from plane
else
{
return null;
}
}
double intersectionDistance = numerator / denominator;
//intersectionDistance < 0 means the "intersection" is behind the ray (pointing away from plane), so not a real intersection
return (intersectionDistance >= 0) ? ray.getLocationWithMagnitude(intersectionDistance) : null;
}
And once I have determined that the ray intersects the plane, here is the code to determine if the ray is inside the triangle:
private boolean isIntersectionVectorInsideTriangle(Vector planeIntersectionVector)
{
//Get edges of triangle
Vector u = getU();
Vector v = getV();
//Pre-compute unique five dot-products
double uu = u.dotProduct(u);
double uv = u.dotProduct(v);
double vv = v.dotProduct(v);
Vector w = planeIntersectionVector.subtract(getB());
double wu = w.dotProduct(u);
double wv = w.dotProduct(v);
double denominator = (uv * uv) - (uu * vv);
//get and test parametric coordinates
double s = ((uv * wv) - (vv * wu)) / denominator;
if (s < 0 || s > 1)
{
return false;
}
double t = ((uv * wu) - (uu * wv)) / denominator;
if (t < 0 || (s + t) > 1)
{
return false;
}
return true;
}
Is think that I am having some issue with my coloring. I think that it has to do with the normals of the various triangles. Here is the equation I am considering when I am building my lighting model for spheres and triangles:
Now, here is the code that does this:
public Color calculateIlluminationModel(Vector normal, boolean isInShadow, Scene scene, Ray ray, Vector intersectionPoint)
{
//c = cr * ca + cr * cl * max(0, n \dot l)) + cl * cp * max(0, e \dot r)^p
Vector lightSourceColor = getColorVector(scene.getLightColor()); //cl
Vector diffuseReflectanceColor = getColorVector(getMaterialColor()); //cr
Vector ambientColor = getColorVector(scene.getAmbientLightColor()); //ca
Vector specularHighlightColor = getColorVector(getSpecularHighlight()); //cp
Vector directionToLight = scene.getDirectionToLight().normalize(); //l
double angleBetweenLightAndNormal = directionToLight.dotProduct(normal);
Vector reflectionVector = normal.multiply(2).multiply(angleBetweenLightAndNormal).subtract(directionToLight).normalize(); //r
double visibilityTerm = isInShadow ? 0 : 1;
Vector ambientTerm = diffuseReflectanceColor.multiply(ambientColor);
double lambertianComponent = Math.max(0, angleBetweenLightAndNormal);
Vector diffuseTerm = diffuseReflectanceColor.multiply(lightSourceColor).multiply(lambertianComponent).multiply(visibilityTerm);
double angleBetweenEyeAndReflection = scene.getLookFrom().dotProduct(reflectionVector);
angleBetweenEyeAndReflection = Math.max(0, angleBetweenEyeAndReflection);
double phongComponent = Math.pow(angleBetweenEyeAndReflection, getPhongConstant());
Vector phongTerm = lightSourceColor.multiply(specularHighlightColor).multiply(phongComponent).multiply(visibilityTerm);
return getVectorColor(ambientTerm.add(diffuseTerm).add(phongTerm));
}
I am seeing that the dot product between the normal and the light source is -1 for the yellow triangle, and about -.707 for the blue triangle, so I'm not sure if the normal being the wrong way is the problem. Regardless, when I added made sure the angle between the light and the normal was positive (Math.abs(directionToLight.dotProduct(normal));), it caused the opposite problem:
I suspect that it will be a small typo/bug, but I need another pair of eyes to spot what I couldn't.
Note: My triangles have vertices(a,b,c), and the edges (u,v) are computed using a-b and c-b respectively (also, those are used for calculating the plane/triangle normal). A Vector is made up of an (x,y,z) point, and a Ray is made up of a origin Vector and a normalized direction Vector.
Here is how I am calculating normals for all triangles:
private Vector getPlaneNormal()
{
Vector v1 = getU();
Vector v2 = getV();
return v1.crossProduct(v2).normalize();
}
Please let me know if I left out anything that you think is important for solving these issues.
EDIT: After help from #TheVee, this is what I have at then end:
There are still problems with z-buffering, And with phong highlights with the triangles, but the problem I was trying to solve here was fixed.

It is an usual problem in ray tracing of scenes including planar objects that we hit them from a wrong side. The formulas containing the dot product are presented with an inherent assumption that light is incident at the object from a direction to which the outer-facing normal is pointing. This can be true only for half the possible orientations of your triangle and you've been in bad luck to orient it with its normal facing away from the light.
Technically speaking, in a physical world your triangle would not have zero volume. It's composed of some layer of material which is just thin. On either side it has a proper normal that points outside. Assigning a single normal is a simplification that's fair to take because the two only differ in sign.
However, if we made a simplification we need to account for it. Having what technically is an inwards facing normal in our formulas gives negative dot products, which case they are not made for. It's like light was coming from the inside of the object or that it hit a surface could not possibly be in its way. That's why they give an erroneous result. The negative value will subtract light from other sources, and depending on the magnitude and implementation may result in darkening, full black, or numerical underflow.
But because we know the correct normal is either what we're using or its negative, we can simply fix the cases at once by taking a preventive absolute value where a positive dot product is implicitly assumed (in your code, that's angleBetweenLightAndNormal). Some libraries like OpenGL do that for you, and on top use the additional information (the sign) to choose between two different materials (front and back) you may provide if desired. Alternatively, they can be set to not draw the back faces for solid object at all because they will be overdrawn by front faces in solid objects anyway (known as face culling), saving about half of the numerical work.

Rotation won't work in Java physics engine

I am making a java rigid body physics engine, and it has gone great so far, until I tried to implement rotation. I don't know where the problem is coming from. I have methods calculating the moment of inertia of convex polygons and circles using formulas from these websites:
http://lab.polygonal.de/?p=57
http://en.wikipedia.org/wiki/List_of_moments_of_inertia
This is the code for the polygon moment of inertia:
public float momentOfInertia() {
Vector C = centerOfMass().subtract(position); //center of mass
Line[] sides = sides(); //sides of the polygon
float moi = 0; //moment of inertia
for(int i = 0; i < sides.length; i++) {
Line l = sides[i]; //current side of polygon being looped through
Vector p1 = C; //points 1, 2, and 3 are the points of the triangle
Vector p2 = l.point1;
Vector p3 = l.point2;
Vector Cp = p1.add(p2).add(p3).divide(3); //center of mass of the triangle, or C'
float d = new Line(C, Cp).length(); //distance between center of mass
Vector bv = p2.subtract(p1); //vector for side b of triangle
float b = bv.magnitude(); //scalar for length of side b
Vector u = bv.divide(b); //unit vector for side b
Vector cv = p3.subtract(p1); //vector for side c of triangle, only used to calculate variables a and h
float a = cv.dot(u); //length of a in triangle
Vector av = u.multiply(a); //vector for a in triangle
Vector hv = cv.subtract(av); //vector for height of triangle, or h in diagram
float h = hv.magnitude(); //length of height of triangle, or h in diagram
float I = ((b*b*b*h)-(b*b*h*a)+(b*h*a*a)+(b*h*h*h))/36; //calculate moment of inertia of individual triangle
float M = (b*h)/2; //mass or area of triangle
moi += I+M*d*d; //equation in sigma series of website
}
return moi;
}
And this is for the circle:
public float momentOfInertia() {
return (float) Math.pow(radius, 2)*area()/2;
}
I know for a fact that the area functions work fine, I have checked them. I just don't know how to check if the moment of inertia equations are wrong.
For collision detection, I used the separating axis theorem to check for any combination of two polygons and circles, where it can find out whether they are colliding, the normal velocity of the collision, and the contact point of the collision. These methods all work beautifully.
I might also like to say how positions are organized. Every body has a position and a shape, either a polygon or a circle. Each shape has a position, and polygons have individual vertices. So if I want to find the absolute position of a vertex of a polygon-shaped body, I need to add the positions of the body, the polygon, and the vertex itself. The center of mass equation is in absolute position according to the shape, with no account for the body. The center of mass and moment of inertia methods are in the Shape class.
For every body, the constants are being updated according to the force and torque in the body's update method where dt is delta time. I also rotate the polygon based on the difference in rotation, because the vertices are ever changing.
public void update(float dt) {
if(mass != 0) {
momentum = momentum.add(force.multiply(dt));
velocity = momentum.divide(mass);
position = position.add(velocity.multiply(dt));
angularMomentum += torque*dt;
angularVelocity = angularMomentum/momentOfInertia;
angle += angularVelocity*dt;
shape.rotate(angularVelocity*dt);
}
}
Finally, I also have a CollisionResolver class which fixes the collision of two colliding bodies, involving applying the normal force and friction. Here is the class's only method which does all of this:
public static void resolveCollision(Body a, Body b, float dt) {
//calculate normal vector
Vector norm = CollisionDetector.normal(a, b);
Vector normb = norm.multiply(-1);
//undo overlap between bodies
float ratio1 = a.mass/(a.mass+b.mass);
float ratio2 = b.mass/(b.mass+a.mass);
a.position = a.position.add(norm.multiply(ratio1));
b.position = b.position.add(normb.multiply(ratio2));
//calculate contact point of collision and other values needed for rotation
Vector cp = CollisionDetector.contactPoint(a, b, norm);
Vector c = a.shape.centerOfMass().add(a.position);
Vector cb = b.shape.centerOfMass().add(b.position);
Vector d = cp.subtract(c);
Vector db = cp.subtract(cb);
//create the normal force vector from the velocity
Vector u = norm.unit();
Vector ub = u.multiply(-1);
Vector F = new Vector(0, 0);
boolean doA = a.mass != 0;
if(doA) {
F = a.force;
}else {
F = b.force;
}
Vector n = new Vector(0, 0);
Vector nb = new Vector(0, 0);
if(doA) {
Vector Fyp = u.multiply(F.dot(u));
n = Fyp.multiply(-1);
nb = Fyp;
}else{
Vector Fypb = ub.multiply(F.dot(ub));
n = Fypb;
nb = Fypb.multiply(-1);
}
//calculate normal force for body A
float r = a.restitution;
Vector v1 = a.velocity;
Vector vy1p = u.multiply(u.dot(v1));
Vector vx1p = v1.subtract(vy1p);
Vector vy2p = vy1p.multiply(-r);
Vector v2 = vy2p.add(vx1p);
//calculate normal force for body B
float rb = b.restitution;
Vector v1b = b.velocity;
Vector vy1pb = ub.multiply(ub.dot(v1b));
Vector vx1pb = v1b.subtract(vy1pb);
Vector vy2pb = vy1pb.multiply(-rb);
Vector v2b = vy2pb.add(vx1pb);
//calculate friction for body A
float mk = (a.friction+b.friction)/2;
Vector v = a.velocity;
Vector vyp = u.multiply(v.dot(u));
Vector vxp = v.subtract(vyp);
float fk = -n.multiply(mk).magnitude();
Vector fkv = vxp.unit().multiply(fk); //friction force
Vector vr = vxp.subtract(d.multiply(a.angularVelocity));
Vector fkvr = vr.unit().multiply(fk); //friction torque - indicated by r for rotation
//calculate friction for body B
Vector vb = b.velocity;
Vector vypb = ub.multiply(vb.dot(ub));
Vector vxpb = vb.subtract(vypb);
float fkb = -nb.multiply(mk).magnitude();
Vector fkvb = vxpb.unit().multiply(fkb); //friction force
Vector vrb = vxpb.subtract(db.multiply(b.angularVelocity));
Vector fkvrb = vrb.unit().multiply(fkb); //friction torque - indicated by r for rotation
//move bodies based on calculations
a.momentum = v2.multiply(a.mass).add(fkv.multiply(dt));
if(a.mass != 0) {
a.velocity = a.momentum.divide(a.mass);
a.position = a.position.add(a.velocity.multiply(dt));
}
b.momentum = v2b.multiply(b.mass).add(fkvb.multiply(dt));
if(b.mass != 0) {
b.velocity = b.momentum.divide(b.mass);
b.position = b.position.add(b.velocity.multiply(dt));
}
//apply torque to bodies
float t = (d.cross(fkvr)+d.cross(n));
float tb = (db.cross(fkvrb)+db.cross(nb));
if(a.mass != 0) {
a.angularMomentum = t*dt;
a.angularVelocity = a.angularMomentum/a.momentOfInertia;
a.angle += a.angularVelocity*dt;
a.shape.rotate(a.angularVelocity*dt);
}
if(b.mass != 0) {
b.angularMomentum = tb*dt;
b.angularVelocity = b.angularMomentum/b.momentOfInertia;
b.angle += b.angularVelocity*dt;
b.shape.rotate(b.angularVelocity*dt);
}
}
As for the actual problem, both the circles and polygons rotate very slowly and often in wrong directions. I know I am throwing a lot out there, but this problem has been bugging me for a while, and I would appreciate any help I can get.
Thanks.

This answer addresses the "I just don't know how to check if the moment of inertia equations are wrong." part of the question.
There are several possible approaches, some of which you may have already tried, and they can be used in combination:
Unit testing
Take your moment of inertia code and apply it to problems with known solutions from a tutorial or textbook.
Dimensional analysis
I would recommend this anyway for any scientific or engineering program. You may have deleted comments for compactness of posted code, but they are important. Annotate each variable that represents a physical quantity with its units. Check that every expression you evaluate has the right units, based on its inputs, for its result variable. For example, in the classic equation F=ma in SI units: F is in Newtons, equivalent to kg.m/(s^2), m is in kg, a is in m/(s^2), so it all balances. Be careful with transitions between physics world coordinates and screen coordinates.
Program simplification
Try working first with only one instance of one very simple shape for which you can do all the calculations by hand. Since some of your problems do not relate to rotation, a circle may be a good first choice because of its symmetry. Debug that, comparing intermediate results to equivalent results from paper-and-pencil (and calculator). Gradually add more instances of the same shape, then debug a single instance of the next shape...
Deliberate error
Given that you suspect your inertia calculations, try setting arbitrary values slightly different from your calculations, and see what differences they make in the display. Are the effects similar to the problems you are seeing? If so, keep it as a hypothesis.
As a more general note, programs that do iterative simulation can be very vulnerable to accumulated floating point error. Unless you have a real need to save space, and have done enough analysis of the numerical stability of your code to be sure float is OK, I strongly recommend using double instead. This is probably not your current problem, but is something that could become an issue later.

Rotate object to face point

I would like to rotate an object to face a point which I'm have a bit of of trouble with.
So I'm starting with an object that has a base at zero and is aligned on the y axis.
I would like to rotate it so that the top of the object is facing the destination
My process so far is to:
Given axis A
find the distance between my position and my look position: D
create a direction vector: V = D.normalize()
find the right vector: R = A cross D
find the up vector: U = D cross R
find the angle between up and direction: ANGLE = acos((U dot D) / (U.length * D.length))
rotate by angle scaled by direction on each axis
here is the code representation of that. I'm not sure what exactly is wrong with this I've worked it out on paper and to my knowledge this approach should work but the results are completely incorrect when drawn. If anyone sees any flaws and could point me in the right direction it would be great.
Vector3 distance = new Vector3(from.x, from.y, from.z).sub(to.x, to.y, to.z);
final Vector3 axis = new Vector3(0, 1, 0);
final Vector3 direction = distance.clone().normalize();
final Vector3 right = (axis.clone().cross(direction));
final Vector3 up = (distance.clone().cross(right));
float angle = (float) Math.acos((up.dot(direction)/ (up.length() * direction.length())));
bondObject.rotateLocal(angle, direction.x , direction.y, direction.z);

The basic idea here is as follows.
Determine which way the object is facing: directionA
Determine which way the object should be facing: directionB
Determine the angle between those directions: rotationAngle
Determine the rotation axis: rotationAxis
Here's the modified code.
Vector3 distance = new Vector3(from.x, from.y, from.z).sub(to.x, to.y, to.z);
if (distance.length() < DISTANCE_EPSILON)
{
//exit - don't do any rotation
//distance is too small for rotation to be numerically stable
}
//Don't actually need to call normalize for directionA - just doing it to indicate
//that this vector must be normalized.
final Vector3 directionA = new Vector3(0, 1, 0).normalize();
final Vector3 directionB = distance.clone().normalize();
float rotationAngle = (float)Math.acos(directionA.dot(directionB));
if (Math.abs(rotationAngle) < ANGLE_EPSILON)
{
//exit - don't do any rotation
//angle is too small for rotation to be numerically stable
}
final Vector3 rotationAxis = directionA.clone().cross(directionB).normalize();
//rotate object about rotationAxis by rotationAngle