OpenGL 缺陷 - 不需要的插值

Posted

技术标签:

【中文标题】OpenGL 缺陷 - 不需要的插值【英文标题】:OpenGL defect - undesirable interpolation 【发布时间】:2017-05-28 13:10:30 【问题描述】:

我正在尝试使用 Bèzier Surfaces 渲染犹他州茶壶,并且成功了。然而,我有一些完全不受欢迎的缺陷。如您所见here,茶壶看起来不错,但存在一些不必要的“插值”。您认为这可能是什么原因造成的?顺便说一句,当我将对象写入 .ply 文件时,我可以设法使用 meshlab 打开它,没有问题,也没有缺陷。最后,我使用 Xcode 作为我的开发环境。

非常感谢您提供的任何帮助。

编辑:添加了部分代码。首先从this旧论文中检索到控制点和使用的补丁。

struct position3D  GLfloat x, y, z; ;
struct position3D teapot_cp_vertices[] = 
    // 1
     1.4   ,   0.0   ,  2.4 ,
     1.4   ,  -0.784 ,  2.4 ,
     0.784 ,  -1.4   ,  2.4 ,
     0.0   ,  -1.4   ,  2.4 ,
                 .
                 .
                 .


#define TEAPOT_NB_PATCHES 32
#define ORDER 3
unsigned short teapot_patches[][ORDER + 1][ORDER + 1] =

    // rim
      1,   2,   3,   4 , 5,   6,   7,   8 , 9,  10,  11,  12 , 13,  14,  15,  16,  ,
      4,  17,  18,  19 , 8,  20,  21,  22 , 12,  23,  24,  25 , 16,  26,  27,  28,  ,
      19,  29,  30,  31 , 22,  32,  33,  34 , 25,  35,  36,  37 , 28,  38,  39,  40,  ,
      31,  41,  42,   1 , 34,  43,  44,   5 , 37,  45,  46,   9 , 40,  47,  48,  13,  ,
       ...
       ...
       ...
    

所有的点和补丁都可以在提到的论文中找到。

用于渲染茶壶的顶点和三角形的计算方法是:

int factorial(int n)

    assert(n >= 0);
    return (n == 1 || n == 0) ? 1 : factorial(n - 1) * n;

float binomial_coefficient(int i, int n) 
    assert(i >= 0); assert(n >= 0);
    return 1.0f * factorial(n) / (factorial(i) * factorial(n-i));

float bernstein_polynomial(int i, int n, float u) 
    return binomial_coefficient(i, n) * powf(u, i) * powf(1-u, n-i);

void build_control_points_k(int p, struct position3D control_points_k[][ORDER+1]) 
    for (int i = 0; i <= ORDER; i++) 
        for (int j = 0; j <= ORDER; j++) 
            control_points_k[i][j] = teapot_cp_vertices[teapot_patches[p][i][j] - 1];

        
    


Vertex compute_position(struct position3D control_points_k[][ORDER+1], float u, float v) 
    Vertex result = *new Vertex();
    for (int i = 0; i <= ORDER; i++) 
        float poly_i = bernstein_polynomial(i, ORDER, u);
        for (int j = 0; j <= ORDER; j++) 
            float poly_j = bernstein_polynomial(j, ORDER, v);
            result.x += poly_i * poly_j * control_points_k[i][j].x;
            result.y += poly_i * poly_j * control_points_k[i][j].y;
            result.z += poly_i * poly_j * control_points_k[i][j].z;
            result.r = 0; //default colour 
            result.g = 0;
            result.b = 0;
        
    
    return result;

#define RESU 10 //resolution in u axis
#define RESV 10 //resolution in v axis

void Object3D::build_teapot() 
    vlist = new Vertex[TEAPOT_NB_PATCHES * RESU*RESV]; //vertex list
    tlist = new Triangle[TEAPOT_NB_PATCHES * (RESU-1)*(RESV-1) * 2]; //triangle list

    //calculate vertices
    for (int p = 0; p < TEAPOT_NB_PATCHES; p++) 
        struct position3D control_points_k[ORDER+1][ORDER+1];
        build_control_points_k(p, control_points_k);
        for (int ru = 0; ru <= RESU-1; ru++) 
            float u = 1.0 * ru / (RESU-1);
            for (int rv = 0; rv <= RESV-1; rv++) 
                float v = 1.0 * rv / (RESV-1);
                vlist[p*RESU*RESV + ru*RESV + rv] = compute_position(control_points_k, u, v);
                vlist[p*RESU*RESV + ru*RESV + rv].r = 1.0 * p / TEAPOT_NB_PATCHES;
                vlist[p*RESU*RESV + ru*RESV + rv].g = 1.0 * p / TEAPOT_NB_PATCHES;
                vlist[p*RESU*RESV + ru*RESV + rv].b = 0.7;
            
        
    
//calculate triangle vertex orders or namely triangles
int n = 0;
Triangle tmpTrg = *new Triangle();
tmpTrg.nverts = 3;
for (int p = 0; p < TEAPOT_NB_PATCHES; p++) 
    for (int ru = 0; ru < RESU-1; ru++)
        for (int rv = 0; rv < RESV-1; rv++) 
            // ABCD is a square
            // triangle in the order ABC is the first one
            tmpTrg.verts = new int[tmpTrg.nverts];
            tmpTrg.verts[0] = p*RESU*RESV +  ru   *RESV +  rv   ;
            tmpTrg.verts[1] = p*RESU*RESV +  ru   *RESV + (rv+1);
            tmpTrg.verts[2] = p*RESU*RESV + (ru+1)*RESV + (rv+1);
            tlist[n] = tmpTrg;
            n++;
            // triangle in the order CDA is the second one
            tmpTrg.verts = new int[tmpTrg.nverts];
            tmpTrg.verts[0] = p*RESU*RESV + (ru+1)*RESV + (rv+1);
            tmpTrg.verts[1] = p*RESU*RESV + (ru+1)*RESV +  rv   ;
            tmpTrg.verts[2] = p*RESU*RESV +  ru   *RESV +  rv   ;

            tlist[n] = tmpTrg;
            n++;
        
    

这里是GL初始化函数:

void init(int w, int h)

    // Init GLFW
    glfwInit();
    glfwWindowHint(GLFW_CONTEXT_VERSION_MAJOR, 3);
    glfwWindowHint(GLFW_CONTEXT_VERSION_MINOR, 3);
    glfwWindowHint(GLFW_OPENGL_PROFILE, GLFW_OPENGL_CORE_PROFILE);
    // macOSX requirement :
    glfwWindowHint(GLFW_OPENGL_FORWARD_COMPAT, GL_TRUE);
    glfwWindowHint(GLFW_RESIZABLE, GL_FALSE);
    glfwWindowHint(GLFW_SAMPLES, 4);

    window = glfwCreateWindow(w, h, "OpenGLTeapot", nullptr, nullptr); // Windowed
    glfwMakeContextCurrent(window);

    glfwSetKeyCallback(window, key_callback);

    // Initialize GLEW to setup the OpenGL Function pointers
    glewExperimental = GL_TRUE;
    glewInit();

    // Define the viewport dimensions
    glViewport(0, 0, w, h);

    // Enable depth test
    glEnable(GL_DEPTH_TEST);

编辑 2: 我尝试使用 Visual Studio15 在 Windows 计算机上运行代码,并且没有这样的缺陷。有没有人知道什么可能导致这样一个愚蠢的问题?

编辑 3:这是对象创建代码:

void Object3D::CreateObject()


    int attributeCount = 6;
    vertexCount        = TEAPOT_NB_PATCHES * RESU*RESV;
    triangleCount      = TEAPOT_NB_PATCHES * (RESU-1)*(RESV-1) * 2;
    build_teapot();

    //Bind the vertex and index buffers
    glGenVertexArrays(1, &VAO);
    glGenBuffers(1, &VBO);
    glGenBuffers(1, &EBO);
    // Bind our Vertex Array Object first, then bind and set our buffers and pointers.
    glBindVertexArray(VAO);
    //Convert our vertex list into a continuous array, copy the vertices into the vertex buffer.
    float* vertexData = new float[vertexCount * attributeCount];
    for (int i = 0; i < vertexCount; i++)
        memcpy(&vertexData[i*attributeCount], 
            vlist[i].getAsArray(), sizeof(float)*attributeCount);
    glBindBuffer(GL_ARRAY_BUFFER, VBO);
    glBufferData(GL_ARRAY_BUFFER, sizeof(float)*attributeCount*vertexCount, vertexData, GL_STATIC_DRAW);
    //Copy the index data found in the list of triangles into the element array buffer (index array)
    //We are using a triangles, so we need triangleCount * 3 indices.
    int* indexData = new int[triangleCount * 3];
    for (int i = 0; i < triangleCount; i++)
        memcpy(&indexData[i * 3], tlist[i].verts, sizeof(int) * 3);
    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, EBO);
    glBufferData(GL_ELEMENT_ARRAY_BUFFER, sizeof(int)*3*triangleCount, indexData, GL_STATIC_DRAW);
    // Position attribute
    glVertexAttribPointer(0, 3, GL_FLOAT, GL_FALSE, attributeCount * sizeof(GLfloat), (GLvoid*)0);
    glEnableVertexAttribArray(0);
    // Color attribute
    glVertexAttribPointer(1, 3, GL_FLOAT, GL_FALSE, attributeCount * sizeof(GLfloat), (GLvoid*)(3 * sizeof(GLfloat)));
    glEnableVertexAttribArray(1);
    // Unbind VAO
    glBindVertexArray(0);

    // Delete temporary buffers
    delete[] vertexData;
    delete[] indexData;

以及主要功能:

int main()

    int screenWidth = 800;
    int screenHeight = 600;
    init(screenWidth, screenHeight);
    std::vector<Object3D*> listOfObjects;


    Object3D* pObj = new Object3D();


    pObj->CreateObject();
    listOfObjects.push_back(pObj);

    //Create the shaders. 
    Shader shader(VertexShaderPath, FragmentShaderPath);

    while (!glfwWindowShouldClose(window))
    
        glfwPollEvents();
        // Clear the colorbuffer
        glClearColor(0.0f, 0.3f, 0.3f, 1.0f);
        glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);

        // Use the shader
        shader.Use();
        // Transformations
        // Create camera transformation
        glm::mat4 view;
        glm::vec3 cameraPos = glm::vec3(0.0f, 150.0f, 100.0f);
        glm::vec3 cameraTarget = glm::vec3(0.0, 80.0f, 20.0f);
        glm::vec3 cameraUp = glm::vec3(0.0f, 1.0f, 0.0f);
        view = glm::lookAt(cameraPos, cameraTarget, cameraUp);
        // Create projection transformation
        glm::mat4 projection;
        projection = glm::perspective<float>(90.0, (float)screenWidth / (float)screenHeight, 0.1f, 1000.0f);
        // Get the uniform locations
        GLint modelLoc = glGetUniformLocation(shader.Program, "model");
        GLint viewLoc = glGetUniformLocation(shader.Program, "view");
        GLint projLoc = glGetUniformLocation(shader.Program, "projection");
        // Pass the view and projection matrices to the shaders
        glUniformMatrix4fv(viewLoc, 1, GL_FALSE, glm::value_ptr(view));
        glUniformMatrix4fv(projLoc, 1, GL_FALSE, glm::value_ptr(projection));
        // Put the bottom of the object on XZ plane and scale it up
        pObj->modelMatrix = glm::scale(glm::mat4(1.0f), glm::vec3(30.0f));
        pObj->modelMatrix = glm::rotate(pObj->modelMatrix, -90.0f, glm::vec3(1.0f, 0.0f,0.0f));
        for (auto pObj : listOfObjects)
        
            glBindVertexArray(pObj->VAO);

            glUniformMatrix4fv(modelLoc, 1, GL_FALSE, glm::value_ptr(pObj->modelMatrix));
            glDrawElements(GL_TRIANGLES, (pObj->vertexCount) * 6, GL_UNSIGNED_INT, 0);
            //glDrawElements(GL_TRIANGLES, 50, GL_UNSIGNED_INT, 0);
            glBindVertexArray(0);
        
        // Swap the buffers

        glfwSwapBuffers(window);

    
    for (auto pObj : listOfObjects)
    
        glDeleteVertexArrays(1, &pObj->VAO);
        glDeleteBuffers(1, &pObj->VBO);
        glDeleteBuffers(1, &pObj->EBO);
        delete pObj;
    
    glfwTerminate();
    return 0;

【问题讨论】:

欢迎来到 SO。请阅读此how-to-ask 并按照那里的指南使用附加信息(例如描述您的编程问题的代码和错误消息)来完善您的问题。 谢谢。问题是,没有错误消息。这就是为什么我没有添加任何源代码。我曾想过分享整个代码,但由于这很容易作为家庭作业提供,我不想成为任何抄袭尝试的一部分。 这很可能是零事故。没有看到任何代码,没有人可以帮助你。也没有人想看到你所有的代码。您应该改为将问题限制在Minimal, Complete and Verifiable example。 虽然我不认为是零意外除法的原因,但我还是添加了一些可能包含这样问题的部分。我找了一个这样的案例,没有找到。 对我来说没有什么是不正确的,而且您正在编写良好的 ply 文件并在另一台机器上运行良好这一事实表明问题出在 GL 渲染代码中。你能展示你的网格渲染代码吗? 【参考方案1】:

问题似乎与 glDrawElements 函数的偏移有关。当我将该行更改为:

glDrawElements(GL_TRIANGLES, (listOfObjects[i]->vertexCount) * 6, GL_UNSIGNED_INT, (void*)(sizeof(Vertex)));

它可以正确渲染。

【讨论】:

以上是关于OpenGL 缺陷 - 不需要的插值的主要内容,如果未能解决你的问题,请参考以下文章

OpenGL广告牌插值问题

使用余弦函数的OpenGL颜色插值?

渲染具有透明度的纹理时 OpenGL 不需要的像素

Opengl_07_插值

openGL 纹理05

opengl纹理映射总结