透视投影OPENGL和计算着色器

时间:2016-04-24 06:30:44

标签: c++ opengl glsl glm-math compute-shader

我正在关注此网站,以使用计算着色器学习光线跟踪:https://github.com/LWJGL/lwjgl3-wiki/wiki/2.6.1.-Ray-tracing-with-OpenGL-Compute-Shaders-%28Part-I%29

我的问题,该教程详细介绍了获取透视投影的过程。我认为我正确地遵循了他的步骤,但我得到了错误的结果,我相信我在矩阵计算中犯了一个错误。

透视投影的代码 -

//Getting the perspective projection using glm::perspective
glm::mat4 projection = glm::perspective(60.0f, 1024.0f/768.0f, 1.0f, 2.0f);

//My Camera Position
glm::vec3 camPos=glm::vec3(3.0, 2.0, 7.0);

//My View matrix using glm::lookAt
glm::mat4 view = glm::lookAt(camPos, glm::vec3(0.0, 0.5, 0.0),glm::vec3(0.0, 1.0, 0.0));

//Calculating inverse of the view*projection
glm::mat4 inv = glm::inverse(view*projection);

//Calculating the rays from camera position to the corners of the frustum as detailed in the site.
glm::vec4 ray00=glm::vec4(-1, -1, 0, 1) * inv;
ray00 /= ray00.w;
ray00 -= glm::vec4(camPos,1.0);

glm::vec4 ray10 = glm::vec4(+1, -1, 0, 1) * inv;
ray10 /= ray10.w;
ray10 -= glm::vec4(camPos,1.0);

glm::vec4 ray01=glm::vec4(-1, 1, 0, 1) * inv;
ray01 /= ray01.w;
ray01 -= glm::vec4(camPos,1.0);

glm::vec4 ray11 = glm::vec4(+1, +1, 0, 1) * inv;
ray11 /= ray11.w;
ray11 -= glm::vec4(camPos,1.0);

上述转变的结果:

[![在此处输入图像说明] [1]] [1]

作为附加信息,我使用

调用我的计算着色器
//Dispatch Shaders. 
glDispatchCompute ((GLuint)1024.0/16, (GLuint)768.0f/8 , 1);

我也使用

将值传递给着色器
//Querying the location for ray00 and assigning the value. Similarly for the rest
GLuint ray00Id = glGetUniformLocation(computeS, "ray00");
glUniform3f(ray00Id, ray00.x, ray00.y, ray00.z);

GLuint ray01Id = glGetUniformLocation(computeS, "ray01");
glUniform3f(ray01Id, ray01.x, ray01.y, ray01.z);

GLuint ray10Id = glGetUniformLocation(computeS, "ray10");
glUniform3f(ray10Id, ray10.x, ray10.y, ray10.z);

GLuint ray11Id = glGetUniformLocation(computeS, "ray11");
glUniform3f(ray11Id, ray11.x, ray11.y, ray11.z);

GLuint camId = glGetUniformLocation(computeS, "eye");
glUniform3f(camId, camPos.x, camPos.y, camPos.z);

根据derhass建议更新了答案。

我的图片现在看起来像: Latest Image

1 个答案:

答案 0 :(得分:1)

glm库使用标准的OpenGL矩阵约定,这意味着使用乘法顺序Matrix * Vector创建矩阵。所以下面的代码是错误的:

//Calculating inverse of the view*projection
glm::mat4 inv = glm::inverse(view*projection);

视图矩阵的组成(从世界空间转换到眼睛空间)和投影矩阵(从眼睛空间转换到剪辑空间)是projection * view,而不是view * projection,因为你把它放在会在视图之前应用投影。)