Triangle Normal Surface Detection - c++

So I currently have a triangle mesh (made with bezier curves) that can be changed dynamically. The problem I am facing is trying to figure out which triangles to actually render based on where the camera is at. The camera always looks towards the origin (0,0,0) so I am finding each triangle's normal and taking it's dotproduct with my camera vector. Then, based on the result, determining if the triangle should be "visible" or not.
The following is the code I am using for the calculations:
void bezier_plane()
{
for (int i = 0; i < 20; i++) {
for (int j = 0; j < 20; j++) {
grid[i][j].x = 0;
grid[i][j].y = 0;
grid[i][j].z = 0;
}
}
//Creates the grid using bezier calculation
CalcBezier();
for (int i = 0; i < 19; i++) {
for (int j = 0; j < 19; j++) {
Vector p1, p2, p3, normal;
p1.x = grid[i+1][j+1].x - grid[i][j].x; p1.y = grid[i+1][j+1].y - grid[i][j].y; p1.z = grid[i+1][j+1].z - grid[i][j].z;
p2.x = grid[i+1][j].x - grid[i][j].x; p1.y = grid[i+1][j].y - grid[i][j].y; p1.z = grid[i+1][j].z - grid[i][j].z;
normal = CalcNormal(p2, p1);
double first = dotproduct(normal, Camera);
p3.x = grid[i][j+1].x - grid[i][j].x; p3.y = grid[i][j+1].y - grid[i][j].y; p3.z = grid[i][j+1].z - grid[i][j].z;
normal = CalcNormal(p1, p3);
double second = dotproduct(normal, Camera);
if (first < 0 && second < 0) {
glPolygonMode(GL_FRONT_AND_BACK, GL_LINE);
glColor3f(0, 1, 0);
glBegin(GL_TRIANGLE_STRIP);
glVertex3f(grid[i][j].x, grid[i][j].y, grid[i][j].z);
glVertex3f(grid[i][j+1].x, grid[i][j+1].y, grid[i][j+1].z);
glVertex3f(grid[i+1][j].x, grid[i+1][j].y, grid[i+1][j].z);
glVertex3f(grid[i+1][j+1].x, grid[i+1][j+1].y, grid[i+1][j+1].z);
glEnd();
} else if (first < 0 && second > 0) {
glPolygonMode(GL_FRONT_AND_BACK, GL_LINE);
glColor3f(0, 1, 0);
glBegin(GL_TRIANGLE_STRIP);
glVertex3f(grid[i][j].x, grid[i][j].y, grid[i][j].z);
glVertex3f(grid[i+1][j].x, grid[i+1][j].y, grid[i+1][j].z);
glVertex3f(grid[i+1][j+1].x, grid[i+1][j+1].y, grid[i+1][j+1].z);
glEnd();
} else if (first > 0 && second < 0) {
glPolygonMode(GL_FRONT_AND_BACK, GL_LINE);
glColor3f(0, 1, 0);
glBegin(GL_TRIANGLE_STRIP);
glVertex3f(grid[i][j].x, grid[i][j].y, grid[i][j].z);
glVertex3f(grid[i][j+1].x, grid[i][j+1].y, grid[i][j+1].z);
glVertex3f(grid[i+1][j+1].x, grid[i+1][j+1].y, grid[i+1][j+1].z);
glEnd();
}
}
}
}
Here is CalcNormal:
Vector CalcNormal(Vector p1, Vector p2)
{
Vector normal;
normal.x = (p1.y * p2.z) - (p1.z * p2.y);
normal.y = (p1.z * p2.x) - (p1.x * p2.z);
normal.z = (p1.x * p2. y) - (p1.y * p2.x);
return normal;
}
double dotproduct(Vector normal, Vector Camera)
{
return (normal.x * Camera.x + normal.y * Camera.y + normal.z + Camera.z);
}
Right now, my code gives this result. The part circled in red should NOT be displayed (I believe, the triangles in back).

Your approach of testing the normals will still have visual artifacts, because triangles facing the camera could also be obscured. Imagine if that bulge were at the corner closest to the camera.
You will also have triangles that are partially visible and partially obscured.
A solution that would work on the pixel level would be:
glEnable(GL_DEPTH_TEST)​
Draw the surface first with solid triangles instead of wire frame
Clear the frame buffer, but not the depth buffer
Now draw your entire scene. The depth buffer will prevent obscured pixels from being drawn

"Normal is a global variable" - could it be that that is already your problem? This looks like the worst application of global data I can think of! Instead, calling this thing crossproduct and returning a vector sounds like a good idea, no? Also, the dotproduct should take two vectors as parameter.
That said, your approach is sound. If you always have the same direction for the corners of triangles, the cross product of two sides will give you the normal. Further, if the angle between the normal and the view is less than 90 degrees, it looks away from the view and should be made invisible. Therefore the problem must be in your implementation, and using global state that could be stored in CPU registers anyway is the first thing you should fix.
Edit: You could use operator overloading to the reader's advantage here:
class Vector
{
Vector(){}
Vector(scalar x0, scalar y0, scalar z0): x(x0), y(y0), z(z0){}
float x, y, z;
};
Vector operator-(Vector const& v1, Vector const& v2)
{
return Vector(v1.x - v2.x, v1.y - v2.y, v1.z - v2.z);
}
Then, start the loop body like this:
Vector const point1 = grid[i, j];
Vector const point2 = grid[i + 1, j];
Vector const point3 = grid[i, j + 1];
Vector const point4 = grid[i + 1, j + 1];
These will easily be optimized out by the compiler, while they ease debugging and improve readability. Also note that they are constant, which makes the compiler verify that you don't change them accidentally. Then, you compute the two normals of the two triangles:
Vector const norm1 = crossproduct(point2 - point1, point3 - point1);
Vector const norm2 = crossproduct(point4 - point2, point4 - point3);
Then, you can check the dotproduct for visibility:
bool const visible1 = dotproduct(norm1, Camera) > 0;
bool const visible2 = dotproduct(norm2, Camera) > 0;
Lastly, you could overload glVertex3f() to take a Vector, but I'd stay away from overloading other libraries' functions.

Related

Terrain Collision issues

I'm trying to implement terrain collision for my height map terrain, and I'm following this. The tutorial is for java but I'm using C++, though the principles are the same so it shouldn't be a problem.
To start off we need a function to get the height of the terrain based on the camera's position. WorldX and WorldZ is the camera's position (x, z) and heights is an 2D-array containing all the heights of the vertices.
float HeightMap::getHeightOfTerrain(float worldX, float worldZ, float heights[][256])
{
//Image is (256 x 256)
float gridLength = 256;
float terrainLength = 256;
float terrainX = worldX;
float terrainZ = worldZ;
float gridSquareLength = terrainLength / ((float)gridLength - 1);
int gridX = (int)std::floor(terrainX / gridSquareLength);
int gridZ = (int)std::floor(terrainZ / gridSquareLength);
//Check if position is on the terrain
if (gridX >= gridLength - 1 || gridZ >= gridLength - 1 || gridX < 0 || gridZ < 0)
{
return 0;
}
//Find out where the player is on the grid square
float xCoord = std::fmod(terrainX, gridSquareLength) / gridSquareLength;
float zCoord = std::fmod(terrainZ, gridSquareLength) / gridSquareLength;
float answer = 0.0;
//Top triangle of a square else the bottom
if (xCoord <= (1 - zCoord))
{
answer = barryCentric(glm::vec3(0, heights[gridX][gridZ], 0),
glm::vec3(1, heights[gridX + 1][gridZ], 0), glm::vec3(0, heights[gridX][gridZ + 1], 1),
glm::vec2(xCoord, zCoord));
}
else
{
answer = barryCentric(glm::vec3(1, heights[gridX + 1][gridZ], 0),
glm::vec3(1, heights[gridX + 1][gridZ + 1], 1), glm::vec3(0, heights[gridX][gridZ + 1], 1),
glm::vec2(xCoord, zCoord));
}
return answer;
}
To find the height of the triangle the camera is currently standing on we use the baryCentric interpolation function.
float HeightMap::barryCentric(glm::vec3 p1, glm::vec3 p2, glm::vec3 p3, glm::vec2 pos)
{
float det = (p2.z - p3.z) * (p1.x - p3.x) + (p3.x - p2.x) * (p1.z - p3.z);
float l1 = ((p2.z - p3.z) * (pos.x - p3.x) + (p3.x - p2.x) * (pos.y - p3.z)) / det;
float l2 = ((p3.z - p1.z) * (pos.x - p3.x) + (p1.x - p3.x) * (pos.y - p3.z)) / det;
float l3 = 1.0f - l1 - l2;
return l1 * p1.y + l2 * p2.y + l3 * p3.y;
}
Then we just have to use the height we have calculated to check for
collision during the game
float terrainHeight = heightMap.getHeightOfTerrain(camera.Position.x, camera.Position.z, heights);
if (camera.Position.y < terrainHeight)
{
camera.Position.y = terrainHeight;
};
Now according to the tutorial this should work perfectly fine, but the height is rather off and at some places it doesn't even work. I figured it might have something to do with the translation and scaling part of the terrain
glm::mat4 model;
model = glm::translate(model, glm::vec3(0.0f, -0.3f, -15.0f));
model = glm::scale(model, glm::vec3(0.1f, 0.1f, 0.1f));
and that I should multiply the values of the heights array by 0.1, as the scaling does that part for the terrain on the GPU side, but that didn't do the trick.
Note
In the tutorial the first lines in the getHeightOfTerrain function says
float terrainX = worldX - x;
float terrainZ = worldZ - z;
where x and z is the world position of the terrain. This is done to get the player position relative to the terrain's position. I tried with the values from the translation part, but it doensn't work either. I changed these lines because it doesn't seem necessary.
float terrainX = worldX - x;
float terrainZ = worldZ - z;
Those lines are, in fact, very necessary, unless your terrain is always at the origin.
Your code resource (tutorial) assumes that you haven't scaled or rotated the terrain in any way. The x and z variables are the XZ position of the terrain which take care of cases where the terrain is translated.
Ideally, you should transform the world position vector from world space to object space (using the inverse of the model matrix you use for the terrain), something like
vec3 localPosition = inverse(model) * vec4(worldPosition, 1)
And then use localPosition.x and localPosition.z instead of terrainX and terrainZ.

calculating vertex normals in opengl with c++

could anyone please help me calculating vertex normals in OpenGL?
I am loading an obj file and adding Gouraud shading by calculating vertex normals without using glNormal3f or glLight functions..
I have declared functions like operators, crossproduct, innerproduct,and etc..
I have understood that in order to get vertex normals, I first need to calculate surface normal aka normal vector with crossproduct.. and also
since I am loading an obj file.. and I am placing the three points of Faces of the obj file in id1,id2,id3 something like that
I would be grateful if anyone can help me writing codes or give me a guideline how to start the codes. please ...
thanks..
its to draw
FACE cur_face = cube.face[i];
glColor3f(cube.vertex_color[cur_face.id1].x,cube.vertex_color[cur_face.id1].y,cube.vertex_color[cur_face.id1].z);
glVertex3f(cube.vertex[cur_face.id1].x,cube.vertex[cur_face.id1].y,cube.vertex[cur_face.id1].z);
glColor3f(cube.vertex_color[cur_face.id2].x,cube.vertex_color[cur_face.id2].y,cube.vertex_color[cur_face.id2].z);
glVertex3f(cube.vertex[cur_face.id2].x,cube.vertex[cur_face.id2].y,cube.vertex[cur_face.id2].z);
glColor3f(cube.vertex_color[cur_face.id3].x,cube.vertex_color[cur_face.id3].y,cube.vertex_color[cur_face.id3].z);
glVertex3f(cube.vertex[cur_face.id3].x,cube.vertex[cur_face.id3].y,cube.vertex[cur_face.id3].z);
}
This is the equation for color calculation
VECTOR kd;
VECTOR ks;
kd=VECTOR(0.8, 0.8, 0.8);
ks=VECTOR(1.0, 0.0, 0.0);
double inner = kd.InnerProduct(ks);
int i, j;
for(i=0;i<cube.vertex.size();i++)
{
VECTOR n = cube.vertex_normal[i];
VECTOR l = VECTOR(100,100,0) - cube.vertex[i];
VECTOR v = VECTOR(0,0,1) - cube.vertex[i];
float xl = n.InnerProduct(l)/n.Magnitude();
VECTOR x = (n * (1.0/ n.Magnitude())) * xl;
VECTOR r = x - (l-x);
VECTOR color = kd * (n.InnerProduct(l)) + ks * pow((v.InnerProduct(r)),10);
cube.vertex_color[i] = color;
*This answer is for triangular mesh and can be extended to poly mesh as well.
tempVertices stores list of all vertices.
vertexIndices stores details of faces(triangles) of the mesh in a vector (in a flat manner).
std::vector<glm::vec3> v_normal;
// initialize vertex normals to 0
for (int i = 0; i != tempVertices.size(); i++)
{
v_normal.push_back(glm::vec3(0.0f, 0.0f, 0.0f));
}
// For each face calculate normals and append to the corresponding vertices of the face
for (unsigned int i = 0; i < vertexIndices.size(); i += 3)
{
//vi v(i+1) v(i+2) are the three faces of a triangle
glm::vec3 A = tempVertices[vertexIndices[i] - 1];
glm::vec3 B = tempVertices[vertexIndices[i + 1] - 1];
glm::vec3 C = tempVertices[vertexIndices[i + 2] - 1];
glm::vec3 AB = B - A;
glm::vec3 AC = C - A;
glm::vec3 ABxAC = glm::cross(AB, AC);
v_normal[vertexIndices[i] - 1] += ABxAC;
v_normal[vertexIndices[i + 1] - 1] += ABxAC;
v_normal[vertexIndices[i + 2] - 1] += ABxAC;
}
Now normalize each v_normal and use.
Note that the number of vertex normals is equal to the number of vertices of the mesh.
This code works fine on my machine
glm::vec3 computeFaceNormal(glm::vec3 p1, glm::vec3 p2, glm::vec3 p3) {
// Uses p2 as a new origin for p1,p3
auto a = p3 - p2;
auto b = p1 - p2;
// Compute the cross product a X b to get the face normal
return glm::normalize(glm::cross(a, b));
}
void Mesh::calculateNormals() {
this->normals = std::vector<glm::vec3>(this->vertices.size());
// For each face calculate normals and append it
// to the corresponding vertices of the face
for (unsigned int i = 0; i < this->indices.size(); i += 3) {
glm::vec3 A = this->vertices[this->indices[i]];
glm::vec3 B = this->vertices[this->indices[i + 1LL]];
glm::vec3 C = this->vertices[this->indices[i + 2LL]];
glm::vec3 normal = computeFaceNormal(A, B, C);
this->normals[this->indices[i]] += normal;
this->normals[this->indices[i + 1LL]] += normal;
this->normals[this->indices[i + 2LL]] += normal;
}
// Normalize each normal
for (unsigned int i = 0; i < this->normals.size(); i++)
this->normals[i] = glm::normalize(this->normals[i]);
}
It seems all you need to implement is the function to get the average vector from N vectors. This is one of the ways to do it:
struct Vector3f {
float x, y, z;
};
typedef struct Vector3f Vector3f;
Vector3f averageVector(Vector3f *vectors, int count) {
Vector3f toReturn;
toReturn.x = .0f;
toReturn.y = .0f;
toReturn.z = .0f;
// sum all the vectors
for(int i=0; i<count; i++) {
Vector3f toAdd = vectors[i];
toReturn.x += toAdd.x;
toReturn.y += toAdd.y;
toReturn.z += toAdd.z;
}
// divide with number of vectors
// TODO: check (count == 0)
float scale = 1.0f/count;
toReturn.x *= scale;
toReturn.y *= scale;
toReturn.z *= scale;
return toReturn;
}
I am sure you can port that to your C++ class. The result should then be normalized unless the length iz zero.
Find all surface normals for every vertex you have. Then use the averageVector and normalize the result to get the smooth normals you are looking for.
Still as already mentioned you should know that this is not appropriate for edged parts of the shape. In those cases you should use the surface vectors directly. You would probably be able to solve most of such cases by simply ignoring a surface normal(s) that are too different from the others. Extremely edgy shapes like cube for instance will be impossible with this procedure. What you would get for instance is:
{
1.0f, .0f, .0f,
.0f, 1.0f, .0f,
.0f, .0f, 1.0f
}
With the normalized average of {.58f, .58f, .58f}. The result would pretty much be an extremely low resolution sphere rather then a cube.

C++ Raytracer with opengl display skew in specific resolution

I have a ray tracer (from www.scratchapixel.com) that I use to write a image to memory that I then display at once using Opengl (glut). I use the width and height and divide the screen to get a Opengl point for every pixels. It kinda works.
My problem is that my width has to be between 500 and 799. It cannot be <= 499 of >= 800, witch doesn't make sense to me. The image becomes skew. I have tried it on 2 computers with the same result.
799x480
800x480
Here's the full code:
#define _USE_MATH_DEFINES
#include <cstdlib>
#include <cstdio>
#include <cmath>
#include <fstream>
#include <vector>
#include <iostream>
#include <cassert>
// OpenGl
#include "GL/glut.h"
GLuint width = 799, height = 480;
GLdouble width_step = 2.0f / width, height_step = 2.0f / height;
const int MAX_RAY_DEPTH = 3;
const double INFINITY = HUGE_VAL;
template<typename T>
class Vec3
{
public:
T x, y, z;
// Vector constructors.
Vec3() : x(T(0)), y(T(0)), z(T(0)) {}
Vec3(T xx) : x(xx), y(xx), z(xx) {}
Vec3(T xx, T yy, T zz) : x(xx), y(yy), z(zz) {}
// Vector normalisation.
Vec3& normalize()
{
T nor = x * x + y * y + z * z;
if (nor > 1) {
T invNor = 1 / sqrt(nor);
x *= invNor, y *= invNor, z *= invNor;
}
return *this;
}
// Vector operators.
Vec3<T> operator * (const T &f) const { return Vec3<T>(x * f, y * f, z * f); }
Vec3<T> operator * (const Vec3<T> &v) const { return Vec3<T>(x * v.x, y * v.y, z * v.z); }
T dot(const Vec3<T> &v) const { return x * v.x + y * v.y + z * v.z; }
Vec3<T> operator - (const Vec3<T> &v) const { return Vec3<T>(x - v.x, y - v.y, z - v.z); }
Vec3<T> operator + (const Vec3<T> &v) const { return Vec3<T>(x + v.x, y + v.y, z + v.z); }
Vec3<T>& operator += (const Vec3<T> &v) { x += v.x, y += v.y, z += v.z; return *this; }
Vec3<T>& operator *= (const Vec3<T> &v) { x *= v.x, y *= v.y, z *= v.z; return *this; }
Vec3<T> operator - () const { return Vec3<T>(-x, -y, -z); }
};
template<typename T>
class Sphere
{
public:
// Sphere variables.
Vec3<T> center; /// position of the sphere
T radius, radius2; /// sphere radius and radius^2
Vec3<T> surfaceColor, emissionColor; /// surface color and emission (light)
T transparency, reflection; /// surface transparency and reflectivity
// Sphere constructor.
// position(c), radius(r), surface color(sc), reflectivity(refl), transparency(transp), emission color(ec)
Sphere(const Vec3<T> &c, const T &r, const Vec3<T> &sc,
const T &refl = 0, const T &transp = 0, const Vec3<T> &ec = 0) :
center(c), radius(r), surfaceColor(sc), reflection(refl),
transparency(transp), emissionColor(ec), radius2(r * r)
{}
// compute a ray-sphere intersection using the geometric solution
bool intersect(const Vec3<T> &rayorig, const Vec3<T> &raydir, T *t0 = NULL, T *t1 = NULL) const
{
// we start with a vector (l) from the ray origin (rayorig) to the center of the curent sphere.
Vec3<T> l = center - rayorig;
// tca is a vector length in the direction of the normalise raydir.
// its length is streched using dot until it forms a perfect right angle triangle with the l vector.
T tca = l.dot(raydir);
// if tca is < 0, the raydir is going in the opposite direction. No need to go further. Return false.
if (tca < 0) return false;
// if we keep on into the code, it's because the raydir may still hit the sphere.
// l.dot(l) gives us the l vector length to the power of 2. Then we use Pythagoras' theorem.
// remove the length tca to the power of two (tca * tca) and we get a distance from the center of the sphere to the power of 2 (d2).
T d2 = l.dot(l) - (tca * tca);
// if this distance to the center (d2) is greater than the radius to the power of 2 (radius2), the raydir direction is missing the sphere.
// No need to go further. Return false.
if (d2 > radius2) return false;
// Pythagoras' theorem again: radius2 is the hypotenuse and d2 is one of the side. Substraction gives the third side to the power of 2.
// Using sqrt, we obtain the length thc. thc is how deep tca goes into the sphere.
T thc = sqrt(radius2 - d2);
if (t0 != NULL && t1 != NULL) {
// remove thc to tca and you get the length from the ray origin to the surface hit point of the sphere.
*t0 = tca - thc;
// add thc to tca and you get the length from the ray origin to the surface hit point of the back side of the sphere.
*t1 = tca + thc;
}
// There is a intersection with a sphere, t0 and t1 have surface distances values. Return true.
return true;
}
};
std::vector<Sphere<double> *> spheres;
// function to mix 2 T varables.
template<typename T>
T mix(const T &a, const T &b, const T &mix)
{
return b * mix + a * (T(1) - mix);
}
// This is the main trace function. It takes a ray as argument (defined by its origin
// and direction). We test if this ray intersects any of the geometry in the scene.
// If the ray intersects an object, we compute the intersection point, the normal
// at the intersection point, and shade this point using this information.
// Shading depends on the surface property (is it transparent, reflective, diffuse).
// The function returns a color for the ray. If the ray intersects an object, it
// returns the color of the object at the intersection point, otherwise it returns
// the background color.
template<typename T>
Vec3<T> trace(const Vec3<T> &rayorig, const Vec3<T> &raydir,
const std::vector<Sphere<T> *> &spheres, const int &depth)
{
T tnear = INFINITY;
const Sphere<T> *sphere = NULL;
// Try to find intersection of this raydir with the spheres in the scene
for (unsigned i = 0; i < spheres.size(); ++i) {
T t0 = INFINITY, t1 = INFINITY;
if (spheres[i]->intersect(rayorig, raydir, &t0, &t1)) {
// is the rayorig inside the sphere (t0 < 0)? If so, use the second hit (t0 = t1)
if (t0 < 0) t0 = t1;
// tnear is the last sphere intersection (or infinity). Is t0 in front of tnear?
if (t0 < tnear) {
// if so, update tnear to this closer t0 and update the closest sphere
tnear = t0;
sphere = spheres[i];
}
}
}
// At this moment in the program, we have the closest sphere (sphere) and the closest hit position (tnear)
// For this pixel, if there's no intersection with a sphere, return a Vec3 with the background color.
if (!sphere) return Vec3<T>(.5); // Grey background color.
// if we keep on with the code, it is because we had an intersection with at least one sphere.
Vec3<T> surfaceColor = 0; // initialisation of the color of the ray/surface of the object intersected by the ray.
Vec3<T> phit = rayorig + (raydir * tnear); // point of intersection.
Vec3<T> nhit = phit - sphere->center; // normal at the intersection point.
// if the normal and the view direction are not opposite to each other,
// reverse the normal direction.
if (raydir.dot(nhit) > 0) nhit = -nhit;
nhit.normalize(); // normalize normal direction
// The angle between raydir and the normal at point hit (not used).
//T s_angle = acos(raydir.dot(nhit)) / ( sqrt(raydir.dot(raydir)) * sqrt(nhit.dot(nhit)));
//T s_incidence = sin(s_angle);
T bias = 1e-5; // add some bias to the point from which we will be tracing
// Do we have transparency or reflection?
if ((sphere->transparency > 0 || sphere->reflection > 0) && depth < MAX_RAY_DEPTH) {
T IdotN = raydir.dot(nhit); // raydir.normal
// I and N are not pointing in the same direction, so take the invert.
T facingratio = std::max(T(0), -IdotN);
// change the mix value between reflection and refraction to tweak the effect (fresnel effect)
T fresneleffect = mix<T>(pow(1 - facingratio, 3), 1, 0.1);
// compute reflection direction (not need to normalize because all vectors
// are already normalized)
Vec3<T> refldir = raydir - nhit * 2 * raydir.dot(nhit);
Vec3<T> reflection = trace(phit + (nhit * bias), refldir, spheres, depth + 1);
Vec3<T> refraction = 0;
// if the sphere is also transparent compute refraction ray (transmission)
if (sphere->transparency) {
T ior = 1.2, eta = 1 / ior;
T k = 1 - eta * eta * (1 - IdotN * IdotN);
Vec3<T> refrdir = raydir * eta - nhit * (eta * IdotN + sqrt(k));
refraction = trace(phit - nhit * bias, refrdir, spheres, depth + 1);
}
// the result is a mix of reflection and refraction (if the sphere is transparent)
surfaceColor = (reflection * fresneleffect + refraction * (1 - fresneleffect) * sphere->transparency) * sphere->surfaceColor;
}
else {
// it's a diffuse object, no need to raytrace any further
// Look at all sphere to find lights
double shadow = 1.0;
for (unsigned i = 0; i < spheres.size(); ++i) {
if (spheres[i]->emissionColor.x > 0) {
// this is a light
Vec3<T> transmission = 1.0;
Vec3<T> lightDirection = spheres[i]->center - phit;
lightDirection.normalize();
T light_angle = (acos(raydir.dot(lightDirection)) / ( sqrt(raydir.dot(raydir)) * sqrt(lightDirection.dot(lightDirection))));
T light_incidence = sin(light_angle);
for (unsigned j = 0; j < spheres.size(); ++j) {
if (i != j) {
T t0, t1;
// Does the ray from point hit to the light intersect an object?
// If so, calculate the shadow.
if (spheres[j]->intersect(phit + (nhit * bias), lightDirection, &t0, &t1)) {
shadow = std::max(0.0, shadow - (1.0 - spheres[j]->transparency));
transmission = transmission * spheres[j]->surfaceColor * shadow;
//break;
}
}
}
// For each light found, we add light transmission to the pixel.
surfaceColor += sphere->surfaceColor * transmission *
std::max(T(0), nhit.dot(lightDirection)) * spheres[i]->emissionColor;
}
}
}
return surfaceColor + sphere->emissionColor;
}
// Main rendering function. We compute a camera ray for each pixel of the image,
// trace it and return a color. If the ray hits a sphere, we return the color of the
// sphere at the intersection point, else we return the background color.
Vec3<double> *image = new Vec3<double>[width * height];
static Vec3<double> cam_pos = Vec3<double>(0);
template<typename T>
void render(const std::vector<Sphere<T> *> &spheres)
{
Vec3<T> *pixel = image;
T invWidth = 1 / T(width), invHeight = 1 / T(height);
T fov = 30, aspectratio = T(width) / T(height);
T angle = tan(M_PI * 0.5 * fov / T(180));
// Trace rays
for (GLuint y = 0; y < height; ++y) {
for (GLuint x = 0; x < width; ++x, ++pixel) {
T xx = (2 * ((x + 0.5) * invWidth) - 1) * angle * aspectratio;
T yy = (1 - 2 * ((y + 0.5) * invHeight)) * angle;
Vec3<T> raydir(xx, yy, -1);
raydir.normalize();
*pixel = trace(cam_pos, raydir, spheres, 0);
}
}
}
//********************************** OPEN GL ***********************************************
void init(void)
{
/* Select clearing (background) color */
glClearColor(0.0, 0.0, 0.0, 0.0);
glShadeModel(GL_FLAT);
/* Initialize viewing values */
//glMatrixMode(GL_PROJECTION);
gluOrtho2D(0,width,0,height);
}
void advanceDisplay(void)
{
cam_pos.z = cam_pos.z - 2;
glutPostRedisplay();
}
void backDisplay(void)
{
cam_pos.z = cam_pos.z + 2;
glutPostRedisplay();
}
void resetDisplay(void)
{
Vec3<double> new_cam_pos;
new_cam_pos = cam_pos;
cam_pos = new_cam_pos;
glutPostRedisplay();
}
void reshape(int w, int h)
{
glLoadIdentity();
gluOrtho2D(0,width,0,height);
glLoadIdentity();
}
void mouse(int button, int state, int x, int y)
{
switch (button)
{
case GLUT_LEFT_BUTTON:
if(state == GLUT_DOWN)
{
glutIdleFunc(advanceDisplay);
}
break;
case GLUT_MIDDLE_BUTTON:
if(state == GLUT_DOWN)
{
glutIdleFunc(resetDisplay);
}
break;
case GLUT_RIGHT_BUTTON:
if(state == GLUT_DOWN)
{
glutIdleFunc(backDisplay);
}
break;
}
}
void display(void)
{
int i;
float x, y;
/* clear all pixels */
glClear(GL_COLOR_BUFFER_BIT);
glPushMatrix();
render<double>(spheres); // Creates the image and put it to memory in image[].
i=0;
glBegin(GL_POINTS);
for(y=1.0f;y>-1.0;y=y-height_step)
{
for(x=1.0f;x>-1.0;x=x-width_step)
{
glColor3f((std::min(double(1), image[i].x)),
(std::min(double(1), image[i].y)),
(std::min(double(1), image[i].z)));
glVertex2f(x, y);
if(i < width*height)
{
i = i + 1;
}
}
}
glEnd();
glPopMatrix();
glutSwapBuffers();
}
int main(int argc, char **argv)
{
// position, radius, surface color, reflectivity, transparency, emission color
spheres.push_back(new Sphere<double>(Vec3<double>(0, -10004, -20), 10000, Vec3<double>(0.2), 0.0, 0.0));
spheres.push_back(new Sphere<double>(Vec3<double>(3, 0, -15), 2, Vec3<double>(1.00, 0.1, 0.1), 0.65, 0.95));
spheres.push_back(new Sphere<double>(Vec3<double>(1, -1, -18), 1, Vec3<double>(1.0, 1.0, 1.0), 0.9, 0.9));
spheres.push_back(new Sphere<double>(Vec3<double>(-2, 2, -15), 2, Vec3<double>(0.1, 0.1, 1.0), 0.05, 0.5));
spheres.push_back(new Sphere<double>(Vec3<double>(-4, 3, -18), 1, Vec3<double>(0.1, 1.0, 0.1), 0.3, 0.7));
spheres.push_back(new Sphere<double>(Vec3<double>(-4, 0, -25), 1, Vec3<double>(1.00, 0.1, 0.1), 0.65, 0.95));
spheres.push_back(new Sphere<double>(Vec3<double>(-1, 1, -25), 2, Vec3<double>(1.0, 1.0, 1.0), 0.0, 0.0));
spheres.push_back(new Sphere<double>(Vec3<double>(2, 2, -25), 1, Vec3<double>(0.1, 0.1, 1.0), 0.05, 0.5));
spheres.push_back(new Sphere<double>(Vec3<double>(5, 3, -25), 2, Vec3<double>(0.1, 1.0, 0.1), 0.3, 0.7));
// light
spheres.push_back(new Sphere<double>(Vec3<double>(-10, 20, 0), 3, Vec3<double>(0), 0, 0, Vec3<double>(3)));
spheres.push_back(new Sphere<double>(Vec3<double>(0, 10, 0), 3, Vec3<double>(0), 0, 0, Vec3<double>(1)));
glutInit(&argc, argv);
glutInitDisplayMode(GLUT_DOUBLE | GLUT_RGB);
glutInitWindowSize(width, height);
glutInitWindowPosition(10,10);
glutCreateWindow(argv[0]);
init();
glutDisplayFunc(display);
glutReshapeFunc(reshape);
glutMouseFunc(mouse);
glutMainLoop();
delete [] image;
while (!spheres.empty()) {
Sphere<double> *sph = spheres.back();
spheres.pop_back();
delete sph;
}
return 0;
}
This is where the image is written to memory:
Vec3<double> *image = new Vec3<double>[width * height];
static Vec3<double> cam_pos = Vec3<double>(0);
template<typename T>
void render(const std::vector<Sphere<T> *> &spheres)
{
Vec3<T> *pixel = image;
T invWidth = 1 / T(width), invHeight = 1 / T(height);
T fov = 30, aspectratio = T(width) / T(height);
T angle = tan(M_PI * 0.5 * fov / T(180));
// Trace rays
for (GLuint y = 0; y < height; ++y) {
for (GLuint x = 0; x < width; ++x, ++pixel) {
T xx = (2 * ((x + 0.5) * invWidth) - 1) * angle * aspectratio;
T yy = (1 - 2 * ((y + 0.5) * invHeight)) * angle;
Vec3<T> raydir(xx, yy, -1);
raydir.normalize();
*pixel = trace(cam_pos, raydir, spheres, 0);
}
}
}
This is where I read it back and write it to each point of Opengl:
void display(void)
{
int i;
float x, y;
/* clear all pixels */
glClear(GL_COLOR_BUFFER_BIT);
glPushMatrix();
render<double>(spheres); // Creates the image and put it to memory in image[].
i=0;
glBegin(GL_POINTS);
for(y=1.0f;y>-1.0;y=y-height_step)
{
for(x=1.0f;x>-1.0;x=x-width_step)
{
glColor3f((std::min(double(1), image[i].x)),
(std::min(double(1), image[i].y)),
(std::min(double(1), image[i].z)));
glVertex2f(x, y);
if(i < width*height)
{
i = i + 1;
}
}
}
glEnd();
glPopMatrix();
glutSwapBuffers();
}
I have no idea what is causing this. Is it a bad design? An Opengl display mode? I don't know.
Is it a bad design?
Yes! Upload your rendered scene to a texture and then render a quad with it:
// g++ -O3 main.cpp -lglut -lGL -lGLU
#include <cstdlib>
#include <cstdio>
#include <cmath>
#include <fstream>
#include <vector>
#include <iostream>
#include <cassert>
// OpenGl
#include "GL/glut.h"
GLuint width = 800, height = 480;
GLdouble width_step = 2.0f / width;
GLdouble height_step = 2.0f / height;
const int MAX_RAY_DEPTH = 3;
template<typename T>
class Vec3
{
public:
T x, y, z;
// Vector constructors.
Vec3() : x(T(0)), y(T(0)), z(T(0)) {}
Vec3(T xx) : x(xx), y(xx), z(xx) {}
Vec3(T xx, T yy, T zz) : x(xx), y(yy), z(zz) {}
// Vector normalisation.
Vec3& normalize()
{
T nor = x * x + y * y + z * z;
if (nor > 1) {
T invNor = 1 / sqrt(nor);
x *= invNor, y *= invNor, z *= invNor;
}
return *this;
}
// Vector operators.
Vec3<T> operator * (const T &f) const { return Vec3<T>(x * f, y * f, z * f); }
Vec3<T> operator * (const Vec3<T> &v) const { return Vec3<T>(x * v.x, y * v.y, z * v.z); }
T dot(const Vec3<T> &v) const { return x * v.x + y * v.y + z * v.z; }
Vec3<T> operator - (const Vec3<T> &v) const { return Vec3<T>(x - v.x, y - v.y, z - v.z); }
Vec3<T> operator + (const Vec3<T> &v) const { return Vec3<T>(x + v.x, y + v.y, z + v.z); }
Vec3<T>& operator += (const Vec3<T> &v) { x += v.x, y += v.y, z += v.z; return *this; }
Vec3<T>& operator *= (const Vec3<T> &v) { x *= v.x, y *= v.y, z *= v.z; return *this; }
Vec3<T> operator - () const { return Vec3<T>(-x, -y, -z); }
};
template<typename T>
class Sphere
{
public:
// Sphere variables.
Vec3<T> center; /// position of the sphere
T radius, radius2; /// sphere radius and radius^2
Vec3<T> surfaceColor, emissionColor; /// surface color and emission (light)
T transparency, reflection; /// surface transparency and reflectivity
// Sphere constructor.
// position(c), radius(r), surface color(sc), reflectivity(refl), transparency(transp), emission color(ec)
Sphere(const Vec3<T> &c, const T &r, const Vec3<T> &sc,
const T &refl = 0, const T &transp = 0, const Vec3<T> &ec = 0) :
center(c), radius(r), surfaceColor(sc), reflection(refl),
transparency(transp), emissionColor(ec), radius2(r * r)
{}
// compute a ray-sphere intersection using the geometric solution
bool intersect(const Vec3<T> &rayorig, const Vec3<T> &raydir, T *t0 = NULL, T *t1 = NULL) const
{
// we start with a vector (l) from the ray origin (rayorig) to the center of the curent sphere.
Vec3<T> l = center - rayorig;
// tca is a vector length in the direction of the normalise raydir.
// its length is streched using dot until it forms a perfect right angle triangle with the l vector.
T tca = l.dot(raydir);
// if tca is < 0, the raydir is going in the opposite direction. No need to go further. Return false.
if (tca < 0) return false;
// if we keep on into the code, it's because the raydir may still hit the sphere.
// l.dot(l) gives us the l vector length to the power of 2. Then we use Pythagoras' theorem.
// remove the length tca to the power of two (tca * tca) and we get a distance from the center of the sphere to the power of 2 (d2).
T d2 = l.dot(l) - (tca * tca);
// if this distance to the center (d2) is greater than the radius to the power of 2 (radius2), the raydir direction is missing the sphere.
// No need to go further. Return false.
if (d2 > radius2) return false;
// Pythagoras' theorem again: radius2 is the hypotenuse and d2 is one of the side. Substraction gives the third side to the power of 2.
// Using sqrt, we obtain the length thc. thc is how deep tca goes into the sphere.
T thc = sqrt(radius2 - d2);
if (t0 != NULL && t1 != NULL) {
// remove thc to tca and you get the length from the ray origin to the surface hit point of the sphere.
*t0 = tca - thc;
// add thc to tca and you get the length from the ray origin to the surface hit point of the back side of the sphere.
*t1 = tca + thc;
}
// There is a intersection with a sphere, t0 and t1 have surface distances values. Return true.
return true;
}
};
std::vector<Sphere<double> *> spheres;
// function to mix 2 T varables.
template<typename T>
T mix(const T &a, const T &b, const T &mix)
{
return b * mix + a * (T(1) - mix);
}
// This is the main trace function. It takes a ray as argument (defined by its origin
// and direction). We test if this ray intersects any of the geometry in the scene.
// If the ray intersects an object, we compute the intersection point, the normal
// at the intersection point, and shade this point using this information.
// Shading depends on the surface property (is it transparent, reflective, diffuse).
// The function returns a color for the ray. If the ray intersects an object, it
// returns the color of the object at the intersection point, otherwise it returns
// the background color.
template<typename T>
Vec3<T> trace(const Vec3<T> &rayorig, const Vec3<T> &raydir,
const std::vector<Sphere<T> *> &spheres, const int &depth)
{
T tnear = INFINITY;
const Sphere<T> *sphere = NULL;
// Try to find intersection of this raydir with the spheres in the scene
for (unsigned i = 0; i < spheres.size(); ++i) {
T t0 = INFINITY, t1 = INFINITY;
if (spheres[i]->intersect(rayorig, raydir, &t0, &t1)) {
// is the rayorig inside the sphere (t0 < 0)? If so, use the second hit (t0 = t1)
if (t0 < 0) t0 = t1;
// tnear is the last sphere intersection (or infinity). Is t0 in front of tnear?
if (t0 < tnear) {
// if so, update tnear to this closer t0 and update the closest sphere
tnear = t0;
sphere = spheres[i];
}
}
}
// At this moment in the program, we have the closest sphere (sphere) and the closest hit position (tnear)
// For this pixel, if there's no intersection with a sphere, return a Vec3 with the background color.
if (!sphere) return Vec3<T>(.5); // Grey background color.
// if we keep on with the code, it is because we had an intersection with at least one sphere.
Vec3<T> surfaceColor = 0; // initialisation of the color of the ray/surface of the object intersected by the ray.
Vec3<T> phit = rayorig + (raydir * tnear); // point of intersection.
Vec3<T> nhit = phit - sphere->center; // normal at the intersection point.
// if the normal and the view direction are not opposite to each other,
// reverse the normal direction.
if (raydir.dot(nhit) > 0) nhit = -nhit;
nhit.normalize(); // normalize normal direction
// The angle between raydir and the normal at point hit (not used).
//T s_angle = acos(raydir.dot(nhit)) / ( sqrt(raydir.dot(raydir)) * sqrt(nhit.dot(nhit)));
//T s_incidence = sin(s_angle);
T bias = 1e-5; // add some bias to the point from which we will be tracing
// Do we have transparency or reflection?
if ((sphere->transparency > 0 || sphere->reflection > 0) && depth < MAX_RAY_DEPTH) {
T IdotN = raydir.dot(nhit); // raydir.normal
// I and N are not pointing in the same direction, so take the invert.
T facingratio = std::max(T(0), -IdotN);
// change the mix value between reflection and refraction to tweak the effect (fresnel effect)
T fresneleffect = mix<T>(pow(1 - facingratio, 3), 1, 0.1);
// compute reflection direction (not need to normalize because all vectors
// are already normalized)
Vec3<T> refldir = raydir - nhit * 2 * raydir.dot(nhit);
Vec3<T> reflection = trace(phit + (nhit * bias), refldir, spheres, depth + 1);
Vec3<T> refraction = 0;
// if the sphere is also transparent compute refraction ray (transmission)
if (sphere->transparency) {
T ior = 1.2, eta = 1 / ior;
T k = 1 - eta * eta * (1 - IdotN * IdotN);
Vec3<T> refrdir = raydir * eta - nhit * (eta * IdotN + sqrt(k));
refraction = trace(phit - nhit * bias, refrdir, spheres, depth + 1);
}
// the result is a mix of reflection and refraction (if the sphere is transparent)
surfaceColor = (reflection * fresneleffect + refraction * (1 - fresneleffect) * sphere->transparency) * sphere->surfaceColor;
}
else {
// it's a diffuse object, no need to raytrace any further
// Look at all sphere to find lights
double shadow = 1.0;
for (unsigned i = 0; i < spheres.size(); ++i) {
if (spheres[i]->emissionColor.x > 0) {
// this is a light
Vec3<T> transmission = 1.0;
Vec3<T> lightDirection = spheres[i]->center - phit;
lightDirection.normalize();
T light_angle = (acos(raydir.dot(lightDirection)) / ( sqrt(raydir.dot(raydir)) * sqrt(lightDirection.dot(lightDirection))));
T light_incidence = sin(light_angle);
for (unsigned j = 0; j < spheres.size(); ++j) {
if (i != j) {
T t0, t1;
// Does the ray from point hit to the light intersect an object?
// If so, calculate the shadow.
if (spheres[j]->intersect(phit + (nhit * bias), lightDirection, &t0, &t1)) {
shadow = std::max(0.0, shadow - (1.0 - spheres[j]->transparency));
transmission = transmission * spheres[j]->surfaceColor * shadow;
//break;
}
}
}
// For each light found, we add light transmission to the pixel.
surfaceColor += sphere->surfaceColor * transmission *
std::max(T(0), nhit.dot(lightDirection)) * spheres[i]->emissionColor;
}
}
}
return surfaceColor + sphere->emissionColor;
}
// Main rendering function. We compute a camera ray for each pixel of the image,
// trace it and return a color. If the ray hits a sphere, we return the color of the
// sphere at the intersection point, else we return the background color.
Vec3<double> *image = new Vec3<double>[width * height];
static Vec3<double> cam_pos = Vec3<double>(0);
template<typename T>
void render(const std::vector<Sphere<T> *> &spheres)
{
Vec3<T> *pixel = image;
T invWidth = 1 / T(width), invHeight = 1 / T(height);
T fov = 30, aspectratio = T(width) / T(height);
T angle = tan(M_PI * 0.5 * fov / T(180));
// Trace rays
for (GLuint y = 0; y < height; ++y) {
for (GLuint x = 0; x < width; ++x, ++pixel) {
T xx = (2 * ((x + 0.5) * invWidth) - 1) * angle * aspectratio;
T yy = (1 - 2 * ((y + 0.5) * invHeight)) * angle;
Vec3<T> raydir(xx, yy, -1);
raydir.normalize();
*pixel = trace(cam_pos, raydir, spheres, 0);
}
}
}
//********************************** OPEN GL ***********************************************
void advanceDisplay(void)
{
cam_pos.z = cam_pos.z - 2;
glutPostRedisplay();
}
void backDisplay(void)
{
cam_pos.z = cam_pos.z + 2;
glutPostRedisplay();
}
void resetDisplay(void)
{
Vec3<double> new_cam_pos;
new_cam_pos = cam_pos;
cam_pos = new_cam_pos;
glutPostRedisplay();
}
void mouse(int button, int state, int x, int y)
{
switch (button)
{
case GLUT_LEFT_BUTTON:
if(state == GLUT_DOWN)
{
glutIdleFunc(advanceDisplay);
}
break;
case GLUT_MIDDLE_BUTTON:
if(state == GLUT_DOWN)
{
glutIdleFunc(resetDisplay);
}
break;
case GLUT_RIGHT_BUTTON:
if(state == GLUT_DOWN)
{
glutIdleFunc(backDisplay);
}
break;
}
}
GLuint tex = 0;
void display(void)
{
int i;
float x, y;
render<double>(spheres); // Creates the image and put it to memory in image[].
std::vector< unsigned char > buf;
buf.reserve( width * height * 3 );
for( size_t y = 0; y < height; ++y )
{
for( size_t x = 0; x < width; ++x )
{
// flip vertically (height-y) because the OpenGL texture origin is in the lower-left corner
// flip horizontally (width-x) because...the original code did so
size_t i = (height-y) * width + (width-x);
buf.push_back( (unsigned char)( std::min(double(1), image[i].x) * 255.0 ) );
buf.push_back( (unsigned char)( std::min(double(1), image[i].y) * 255.0 ) );
buf.push_back( (unsigned char)( std::min(double(1), image[i].z) * 255.0 ) );
}
}
/* clear all pixels */
glClearColor(0.0, 0.0, 0.0, 0.0);
glClear(GL_COLOR_BUFFER_BIT);
glMatrixMode( GL_PROJECTION );
glLoadIdentity();
glMatrixMode( GL_MODELVIEW );
glLoadIdentity();
glEnable( GL_TEXTURE_2D );
glBindTexture( GL_TEXTURE_2D, tex );
glTexSubImage2D
(
GL_TEXTURE_2D, 0,
0, 0,
width, height,
GL_RGB,
GL_UNSIGNED_BYTE,
&buf[0]
);
glBegin( GL_QUADS );
glTexCoord2i( 0, 0 );
glVertex2i( -1, -1 );
glTexCoord2i( 1, 0 );
glVertex2i( 1, -1 );
glTexCoord2i( 1, 1 );
glVertex2i( 1, 1 );
glTexCoord2i( 0, 1 );
glVertex2i( -1, 1 );
glEnd();
glutSwapBuffers();
}
int main(int argc, char **argv)
{
// position, radius, surface color, reflectivity, transparency, emission color
spheres.push_back(new Sphere<double>(Vec3<double>(0, -10004, -20), 10000, Vec3<double>(0.2), 0.0, 0.0));
spheres.push_back(new Sphere<double>(Vec3<double>(3, 0, -15), 2, Vec3<double>(1.00, 0.1, 0.1), 0.65, 0.95));
spheres.push_back(new Sphere<double>(Vec3<double>(1, -1, -18), 1, Vec3<double>(1.0, 1.0, 1.0), 0.9, 0.9));
spheres.push_back(new Sphere<double>(Vec3<double>(-2, 2, -15), 2, Vec3<double>(0.1, 0.1, 1.0), 0.05, 0.5));
spheres.push_back(new Sphere<double>(Vec3<double>(-4, 3, -18), 1, Vec3<double>(0.1, 1.0, 0.1), 0.3, 0.7));
spheres.push_back(new Sphere<double>(Vec3<double>(-4, 0, -25), 1, Vec3<double>(1.00, 0.1, 0.1), 0.65, 0.95));
spheres.push_back(new Sphere<double>(Vec3<double>(-1, 1, -25), 2, Vec3<double>(1.0, 1.0, 1.0), 0.0, 0.0));
spheres.push_back(new Sphere<double>(Vec3<double>(2, 2, -25), 1, Vec3<double>(0.1, 0.1, 1.0), 0.05, 0.5));
spheres.push_back(new Sphere<double>(Vec3<double>(5, 3, -25), 2, Vec3<double>(0.1, 1.0, 0.1), 0.3, 0.7));
// light
spheres.push_back(new Sphere<double>(Vec3<double>(-10, 20, 0), 3, Vec3<double>(0), 0, 0, Vec3<double>(3)));
spheres.push_back(new Sphere<double>(Vec3<double>(0, 10, 0), 3, Vec3<double>(0), 0, 0, Vec3<double>(1)));
glutInit(&argc, argv);
glutInitDisplayMode(GLUT_DOUBLE | GLUT_RGB);
glutInitWindowSize(width, height);
glutInitWindowPosition(10,10);
glutCreateWindow(argv[0]);
glutDisplayFunc(display);
glutMouseFunc(mouse);
glGenTextures( 1, &tex );
glBindTexture( GL_TEXTURE_2D, tex );
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE );
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE );
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR );
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR );
glPixelStorei( GL_UNPACK_ALIGNMENT, 1 );
glTexImage2D( GL_TEXTURE_2D, 0, 3, width, height, 0, GL_RGB, GL_UNSIGNED_BYTE, NULL );
glutMainLoop();
delete [] image;
while (!spheres.empty()) {
Sphere<double> *sph = spheres.back();
spheres.pop_back();
delete sph;
}
return 0;
}
How to load and display images is also explained on www.scratchapixel.com. Strange you didn't see this lesson:
http://www.scratchapixel.com/lessons/3d-basic-lessons/lesson-5-colors-and-digital-images/source-code/
It's all in there and they explain you how to display images using GL textures indeed.

Find mouse world-coordinates (3D) on a quadtree heightmaped terrain

I'm trying to find the mouse position in world coordinates but am having trouble finding the right code. At the moment I use this to determine the ray:
float pointX, pointY;
D3DXMATRIX projectionMatrix, viewMatrix, inverseViewMatrix, worldMatrix, translateMatrix, inverseWorldMatrix;
D3DXVECTOR3 direction, origin, rayOrigin, rayDirection;
bool intersect, result;
// Move the mouse cursor coordinates into the -1 to +1 range.
pointX = ((2.0f * (float)mouseX) / (float)m_screenWidth) - 1.0f;
pointY = (((2.0f * (float)mouseY) / (float)m_screenHeight) - 1.0f) * -1.0f;
// Adjust the points using the projection matrix to account for the aspect ratio of the viewport.
m_Direct3D->GetProjectionMatrix(projectionMatrix);
pointX = pointX / projectionMatrix._11;
pointY = pointY / projectionMatrix._22;
// Get the inverse of the view matrix.
m_Camera->GetViewMatrix(viewMatrix);
D3DXMatrixInverse(&inverseViewMatrix, NULL, &viewMatrix);
// Calculate the direction of the picking ray in view space.
direction.x = (pointX * inverseViewMatrix._11) + (pointY * inverseViewMatrix._21) + inverseViewMatrix._31;
direction.y = (pointX * inverseViewMatrix._12) + (pointY * inverseViewMatrix._22) + inverseViewMatrix._32;
direction.z = (pointX * inverseViewMatrix._13) + (pointY * inverseViewMatrix._23) + inverseViewMatrix._33;
// Get the origin of the picking ray which is the position of the camera.
origin = m_Camera->GetPosition();
This gives me the origin and direction of the ray.
But...
I use a custom mesh (not the one from directX) with a heightmap, separated into quadtrees and I don't know if my logic is correct, I tried using the frustum to determine which nodes in the quadtree are visible and so do the checking intersection of triangles only on those nodes, here is this code:
Note* m_mousepos is a vector.
bool QuadTreeClass::getTriangleRay(NodeType* node, FrustumClass* frustum, ID3D10Device* device, D3DXVECTOR3 vPickRayDir, D3DXVECTOR3 vPickRayOrig){
bool result;
int count, i, j, indexCount;
unsigned int stride, offset;
float fBary1, fBary2;
float fDist;
D3DXVECTOR3 v0, v1, v2;
float p1, p2, p3;
// Check to see if the node can be viewed.
result = frustum->CheckCube(node->positionX, 0.0f, node->positionZ, (node->width / 2.0f));
if(!result)
{
return false;
}
// If it can be seen then check all four child nodes to see if they can also be seen.
count = 0;
for(i=0; i<4; i++)
{
if(node->nodes[i] != 0)
{
count++;
getTriangleRay(node->nodes[i], frustum, device, vPickRayOrig, vPickRayDir);
}
}
// If there were any children nodes then dont continue
if(count != 0)
{
return false;
}
// Now intersect each triangle in this node
j = 0;
for(i=0; i<node->triangleCount; i++){
j = i * 3;
v0 = D3DXVECTOR3( node->vertexArray[j].x, node->vertexArray[j].y, node->vertexArray[j].z);
j++;
v1 = D3DXVECTOR3( node->vertexArray[j].x, node->vertexArray[j].y, node->vertexArray[j].z);
j++;
v2 = D3DXVECTOR3( node->vertexArray[j].x, node->vertexArray[j].y, node->vertexArray[j].z);
result = IntersectTriangle( vPickRayOrig, vPickRayDir, v0, v1, v2, &fDist, &fBary1, &fBary2);
if(result == true){
// intersection = true, so get a aproximate center of the triangle on the world
p1 = (v0.x + v0.x + v0.x)/3;
p2 = (v0.y + v1.y + v2.y)/3;
p3 = (v0.z + v1.z + v2.z)/3;
m_mousepos = D3DXVECTOR3(p1, p2, p3);
return true;
}
}
}
bool QuadTreeClass::IntersectTriangle( const D3DXVECTOR3& orig, const D3DXVECTOR3& dir,D3DXVECTOR3& v0, D3DXVECTOR3& v1, D3DXVECTOR3& v2, FLOAT* t, FLOAT* u, FLOAT* v ){
// Find vectors for two edges sharing vert0
D3DXVECTOR3 edge1 = v1 - v0;
D3DXVECTOR3 edge2 = v2 - v0;
// Begin calculating determinant - also used to calculate U parameter
D3DXVECTOR3 pvec;
D3DXVec3Cross( &pvec, &dir, &edge2 );
// If determinant is near zero, ray lies in plane of triangle
FLOAT det = D3DXVec3Dot( &edge1, &pvec );
D3DXVECTOR3 tvec;
if( det > 0 )
{
tvec = orig - v0;
}
else
{
tvec = v0 - orig;
det = -det;
}
if( det < 0.0001f )
return FALSE;
// Calculate U parameter and test bounds
*u = D3DXVec3Dot( &tvec, &pvec );
if( *u < 0.0f || *u > det )
return FALSE;
// Prepare to test V parameter
D3DXVECTOR3 qvec;
D3DXVec3Cross( &qvec, &tvec, &edge1 );
// Calculate V parameter and test bounds
*v = D3DXVec3Dot( &dir, &qvec );
if( *v < 0.0f || *u + *v > det )
return FALSE;
// Calculate t, scale parameters, ray intersects triangle
*t = D3DXVec3Dot( &edge2, &qvec );
FLOAT fInvDet = 1.0f / det;
*t *= fInvDet;
*u *= fInvDet;
*v *= fInvDet;
return TRUE;
}
Please is this code right? If it is then my problem must be related to the quadtree.
Thanks!
Iterating over all visible triangle to find the intersection is very expensive. Additional the cost will rise if your heightmap gets finer.
For my heightmap I use a different approach:
I do a step-by-step search regarding the height on the clickray starting at the origin. At every step the current position is moved along the ray and tested against the height of the heightmap (therefore you need a heightfunction). If the current position is below the heightmap, the last intervall is searched again by an additional iteration to find a finer position. This works as long as your heightmap hasn't a too high frequency in the heightvalues regarding to the stepsize (otherwise you could jump over a peak).

Object picking with ray casting

I've been having a problem with inaccuracies in my ray casting algorithm for detecting mouse hits within a box. I'm completely at a loss as to how to fix this properly and it's been bugging me for weeks.
The problem is easiest described with a picture (box centered around [0, 0, -30]):
The black lines represent the actual hitbox which is drawn and the green box represents what actually appears to get hit. Notice how it's offset (which seems to get larger if the box is further from the origin) and is slightly smaller than the drawn hitbox.
Here's some relevant code,
ray-box cast:
double BBox::checkFaceIntersection(Vector3 points[4], Vector3 normal, Ray3 ray) {
double rayDotNorm = ray.direction.dot(normal);
if(rayDotNorm == 0) return -1;
Vector3 intersect = points[0] - ray.origin;
double t = intersect.dot(normal) / rayDotNorm;
if(t < 0) return -1;
// Check if first point is from under or below polygon
bool positive = false;
double firstPtDot = ray.direction.dot( (ray.origin - points[0]).cross(ray.origin - points[1]) );
if(firstPtDot > 0) positive = true;
else if(firstPtDot < 0) positive = false;
else return -1;
// Check all signs are the same
for(int i = 1; i < 4; i++) {
int nextPoint = (i+1) % 4;
double rayDotPt = ray.direction.dot( (ray.origin - points[i]).cross(ray.origin - points[nextPoint]) );
if(positive && rayDotPt < 0) {
return -1;
}
else if(!positive && rayDotPt > 0) {
return -1;
}
}
return t;
}
mouse to ray:
GLint viewport[4];
GLdouble modelMatrix[16];
GLdouble projectionMatrix[16];
glGetIntegerv(GL_VIEWPORT, viewport);
glGetDoublev(GL_MODELVIEW_MATRIX, modelMatrix);
glGetDoublev(GL_PROJECTION_MATRIX, projectionMatrix);
GLfloat winY = GLfloat(viewport[3] - mouse_y);
Ray3 ray;
double x, y, z;
gluUnProject( (double) mouse_x, winY, 0.0f, // Near
modelMatrix, projectionMatrix, viewport,
&x, &y, &z );
ray.origin = Vector3(x, y, z);
gluUnProject( (double) mouse_x, winY, 1.0f, // Far
modelMatrix, projectionMatrix, viewport,
&x, &y, &z );
ray.direction = Vector3(x, y, z);
if(bbox.checkBoxIntersection(ray) != -1) {
std::cout << "Hit!" << std::endl;
}
I've tried drawing the actual ray as a line and it seems to intersect the drawn box correctly.
I had the offset problem partially fixed by minusing all the points and the ray origin/direction by the boxes position, but I have no idea why that worked and the size of the hitbox still remained inaccurate.
Any ideas/alternative approaches? I have other code to supply if it's needed.
You're assuming a wrong direction. Correct would be:
ray.direction = Vector3(far.x - near.x, far.y - near.y, far.z - near.z);
Without subtracting near and far intersection points, your direction will be off.