Sign in to follow this  

OpenGL Raytracing via compute shader

Recommended Posts

StanLee    157

I am trying to do some raytracing on the GPU via the compute shader in OpenGL and I came across a very strange behaviour.

For every pixel in the screen I launch a compute shader invocation and this is how the compute shader looks like:



#version 430

struct Camera{
    vec4    pos, dir, up, xAxis ;
    float   focalLength;
    float   pW, pH;

struct Sphere{
    vec4    position;
    float   radius;

struct Ray{
    vec3    origin;
    vec3    dir;

uniform Camera      camera;
uniform uint        width;
uniform uint        height;

uniform image2D outputTexture;

float hitSphere(Ray r, Sphere s){
    float s_ov = dot(r.origin, r.dir);
    float s_mv = dot(, r.dir);
    float s_mm = dot(,;
    float s_mo = dot(, r.origin);
    float s_oo = dot(r.origin, r.origin);
    float d = s_ov*s_ov-2.0f*s_ov*s_mv+s_mv*s_mv-s_mm+2.0f*s_mo*s_oo+s.radius*s.radius;
    if(d < 0){
        return -1.0f;
    } else if(d == 0){
        return (s_mv-s_ov);
    } else {
        float t1 = 0, t2 = 0;
        t1 = s_mv-s_ov;
        t2 = (t1-sqrt(d));
        t1 = (t1+sqrt(d));
        return t1>t2? t2 : t1 ;

Ray initRay(uint x, uint y, Camera cam){
    Ray ray;
    ray.origin =;
    ray.dir = * cam.focalLength + vec3(1, 0, 0)*( float(x-(width/2)))*cam.pW
                              + * (float(y-(height/2))*cam.pH);
    ray.dir = normalize(ray.dir);
    return ray;

layout (local_size_x = 16, local_size_y = 16, local_size_z = 1) in;
void main(){
    uint x = gl_GlobalInvocationID.x;
    uint y = gl_GlobalInvocationID.y;
    if(x < 1024 && y < 768){
        float t = 0.0f;

        Ray r = initRay(x, y, camera);
        Sphere sp ={vec4(0.0f, 0.0f, 20.0f, 0.0f), 2.0f};

        t = hitSphere(r, sp);
        if(t <= -0.001f){
            imageStore(outputTexture, ivec2(x, y), vec4(0.0, 0.0, 1.0, 1.0));
        } else {
            imageStore(outputTexture, ivec2(x, y), vec4(0.0, 1.0, 0.0, 1.0));

Rendering on the GPU yields the following broken image:


Rendering on the CPU with the same algorithm yields this image:


I can't figure out the problem since I just copied and pasted the "hitSphere()" and "initRay()" functions into my compute shader. First I thought I haven't dispatched enough work groups, but then the background wouldn't be blue, so this can't be the case. This is how I dispatch my compute shader:

#define WORK_GROUP_SIZE 16
//width = 1024, height = 768
void OpenGLRaytracer::renderScene(int width, int height){

    glDispatchCompute(width/WORK_GROUP_SIZE, height/WORK_GROUP_SIZE,1);


Then I changed the position of the sphere in x direction to the right:


In y direction to the top:


And in both directions (right and top):


When I change the position far enough in both directions to the left and to the bottom, then the sphere actually disappears. It seems that all the calculations on the GPU only work in one quarter of the image (top-right) and happen to yield false results in the other three quarters.


I am totally clueless at the moment and don't even know how to start fixing this.

Edited by StanLee

Share this post

Link to post
Share on other sites
powly k    657

Not sure how you derived that sphere intersection formula, but it looks quite different to what I've used - are your implementations exactly the same on CPU and GPU?

Share this post

Link to post
Share on other sites
Vilem Otte    2938

Assuming your intersection code is right (it should be), the bug is elsewhere:


Ray initRay(uint x, uint y, Camera cam){
    Ray ray;
    ray.origin =;
    ray.dir = * cam.focalLength + vec3(1, 0, 0)*( float(x-(width/2)))*cam.pW
                              + * (float(y-(height/2))*cam.pH);
    ray.dir = normalize(ray.dir);
    return ray;

look at line:


ray.dir = * cam.focalLength + vec3(1, 0, 0)*( float(x-(width/2)))*cam.pW
                              + * (float(y-(height/2))*cam.pH);

Basically you x and are unsigned integers, your width and height are also unsigned integers - so F.E. x - (width/2) when width/2 is larger than x is a problem - it's like subtracting 1 - 3 in unsigned ints, it doesn't give you back correct answer (thats -2) ... so either use ints instead of uints, or convert all uints to float before computing anything (hope I didn't forget bracket somewhere):


ray.dir = * cam.focalLength + vec3(1, 0, 0)*( float(x) - float(width/2))*cam.pW
                              + * ((float(y)-float(height/2))*cam.pH);

Share this post

Link to post
Share on other sites
StanLee    157

Wow, first I felt baffled and now I feel stupid. :D
This gave me hours of desperation. I assumed everything from wrong initialization of my image up to driver bugs but didn't see this mistake.
Thank you very much! :)

Share this post

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this  

  • Similar Content

    • By povilaslt2
      Hello. I'm Programmer who is in search of 2D game project who preferably uses OpenGL and C++. You can see my projects in GitHub. Project genre doesn't matter (except MMO's :D).
    • By ZeldaFan555
      Hello, My name is Matt. I am a programmer. I mostly use Java, but can use C++ and various other languages. I'm looking for someone to partner up with for random projects, preferably using OpenGL, though I'd be open to just about anything. If you're interested you can contact me on Skype or on here, thank you!
      Skype: Mangodoor408
    • By tyhender
      Hello, my name is Mark. I'm hobby programmer. 
      So recently,I thought that it's good idea to find people to create a full 3D engine. I'm looking for people experienced in scripting 3D shaders and implementing physics into engine(game)(we are going to use the React physics engine). 
      And,ye,no money =D I'm just looking for hobbyists that will be proud of their work. If engine(or game) will have financial succes,well,then maybe =D
      Sorry for late replies.
      I mostly give more information when people PM me,but this post is REALLY short,even for me =D
      So here's few more points:
      Engine will use openGL and SDL for graphics. It will use React3D physics library for physics simulation. Engine(most probably,atleast for the first part) won't have graphical fron-end,it will be a framework . I think final engine should be enough to set up an FPS in a couple of minutes. A bit about my self:
      I've been programming for 7 years total. I learned very slowly it as "secondary interesting thing" for like 3 years, but then began to script more seriously.  My primary language is C++,which we are going to use for the engine. Yes,I did 3D graphics with physics simulation before. No, my portfolio isn't very impressive. I'm working on that No,I wasn't employed officially. If anybody need to know more PM me. 
    • By Zaphyk
      I am developing my engine using the OpenGL 3.3 compatibility profile. It runs as expected on my NVIDIA card and on my Intel Card however when I tried it on an AMD setup it ran 3 times worse than on the other setups. Could this be a AMD driver thing or is this probably a problem with my OGL code? Could a different code standard create such bad performance?
    • By Kjell Andersson
      I'm trying to get some legacy OpenGL code to run with a shader pipeline,
      The legacy code uses glVertexPointer(), glColorPointer(), glNormalPointer() and glTexCoordPointer() to supply the vertex information.
      I know that it should be using setVertexAttribPointer() etc to clearly define the layout but that is not an option right now since the legacy code can't be modified to that extent.
      I've got a version 330 vertex shader to somewhat work:
      #version 330 uniform mat4 osg_ModelViewProjectionMatrix; uniform mat4 osg_ModelViewMatrix; layout(location = 0) in vec4 Vertex; layout(location = 2) in vec4 Normal; // Velocity layout(location = 3) in vec3 TexCoord; // TODO: is this the right layout location? out VertexData { vec4 color; vec3 velocity; float size; } VertexOut; void main(void) { vec4 p0 = Vertex; vec4 p1 = Vertex + vec4(Normal.x, Normal.y, Normal.z, 0.0f); vec3 velocity = (osg_ModelViewProjectionMatrix * p1 - osg_ModelViewProjectionMatrix * p0).xyz; VertexOut.velocity = velocity; VertexOut.size = TexCoord.y; gl_Position = osg_ModelViewMatrix * Vertex; } What works is the Vertex and Normal information that the legacy C++ OpenGL code seem to provide in layout location 0 and 2. This is fine.
      What I'm not getting to work is the TexCoord information that is supplied by a glTexCoordPointer() call in C++.
      What layout location is the old standard pipeline using for glTexCoordPointer()? Or is this undefined?
      Side note: I'm trying to get an OpenSceneGraph 3.4.0 particle system to use custom vertex, geometry and fragment shaders for rendering the particles.
  • Popular Now