How to extend the Cube OpenGL ES 2.0 example for 3D meshes?

[shome]

I have gone through the code of Cube OpenGL ES 2.0 example which displays a cube that can be manipulated using the mouse.

How to extend the above example for a 3D mesh. . I need to compile for wasm.

In my little understanding, I need to modify the function named void GeometryEngine::initCubeGeometry (refer geometryengine.cpp)

Here the cube is represented as triangle strip primitives . How to do the same for a 3D mesh.? Is there any API for the same?
Or do we need to use other libraries such as assimp or Qt3D? I need to compile for wasm

The code for that function is as below:

void GeometryEngine::initCubeGeometry()
{
    // For cube we would need only 8 vertices but we have to
    // duplicate vertex for each face because texture coordinate
    // is different.
    VertexData vertices[] = {
        // Vertex data for face 0
        {QVector3D(-1.0f, -1.0f,  1.0f), QVector2D(0.0f, 0.0f)},  // v0
        {QVector3D( 1.0f, -1.0f,  1.0f), QVector2D(0.33f, 0.0f)}, // v1
        {QVector3D(-1.0f,  1.0f,  1.0f), QVector2D(0.0f, 0.5f)},  // v2
        {QVector3D( 1.0f,  1.0f,  1.0f), QVector2D(0.33f, 0.5f)}, // v3

        // Vertex data for face 1
        {QVector3D( 1.0f, -1.0f,  1.0f), QVector2D( 0.0f, 0.5f)}, // v4
        {QVector3D( 1.0f, -1.0f, -1.0f), QVector2D(0.33f, 0.5f)}, // v5
        {QVector3D( 1.0f,  1.0f,  1.0f), QVector2D(0.0f, 1.0f)},  // v6
        {QVector3D( 1.0f,  1.0f, -1.0f), QVector2D(0.33f, 1.0f)}, // v7

        // Vertex data for face 2
        {QVector3D( 1.0f, -1.0f, -1.0f), QVector2D(0.66f, 0.5f)}, // v8
        {QVector3D(-1.0f, -1.0f, -1.0f), QVector2D(1.0f, 0.5f)},  // v9
        {QVector3D( 1.0f,  1.0f, -1.0f), QVector2D(0.66f, 1.0f)}, // v10
        {QVector3D(-1.0f,  1.0f, -1.0f), QVector2D(1.0f, 1.0f)},  // v11

        // Vertex data for face 3
        {QVector3D(-1.0f, -1.0f, -1.0f), QVector2D(0.66f, 0.0f)}, // v12
        {QVector3D(-1.0f, -1.0f,  1.0f), QVector2D(1.0f, 0.0f)},  // v13
        {QVector3D(-1.0f,  1.0f, -1.0f), QVector2D(0.66f, 0.5f)}, // v14
        {QVector3D(-1.0f,  1.0f,  1.0f), QVector2D(1.0f, 0.5f)},  // v15

        // Vertex data for face 4
        {QVector3D(-1.0f, -1.0f, -1.0f), QVector2D(0.33f, 0.0f)}, // v16
        {QVector3D( 1.0f, -1.0f, -1.0f), QVector2D(0.66f, 0.0f)}, // v17
        {QVector3D(-1.0f, -1.0f,  1.0f), QVector2D(0.33f, 0.5f)}, // v18
        {QVector3D( 1.0f, -1.0f,  1.0f), QVector2D(0.66f, 0.5f)}, // v19

        // Vertex data for face 5
        {QVector3D(-1.0f,  1.0f,  1.0f), QVector2D(0.33f, 0.5f)}, // v20
        {QVector3D( 1.0f,  1.0f,  1.0f), QVector2D(0.66f, 0.5f)}, // v21
        {QVector3D(-1.0f,  1.0f, -1.0f), QVector2D(0.33f, 1.0f)}, // v22
        {QVector3D( 1.0f,  1.0f, -1.0f), QVector2D(0.66f, 1.0f)}  // v23
    };

    // Indices for drawing cube faces using triangle strips.
    // Triangle strips can be connected by duplicating indices
    // between the strips. If connecting strips have opposite
    // vertex order then last index of the first strip and first
    // index of the second strip needs to be duplicated. If
    // connecting strips have same vertex order then only last
    // index of the first strip needs to be duplicated.
    GLushort indices[] = {
         0,  1,  2,  3,  3,     // Face 0 - triangle strip ( v0,  v1,  v2,  v3)
         4,  4,  5,  6,  7,  7, // Face 1 - triangle strip ( v4,  v5,  v6,  v7)
         8,  8,  9, 10, 11, 11, // Face 2 - triangle strip ( v8,  v9, v10, v11)
        12, 12, 13, 14, 15, 15, // Face 3 - triangle strip (v12, v13, v14, v15)
        16, 16, 17, 18, 19, 19, // Face 4 - triangle strip (v16, v17, v18, v19)
        20, 20, 21, 22, 23      // Face 5 - triangle strip (v20, v21, v22, v23)
    };

//! [1]
    // Transfer vertex data to VBO 0
    arrayBuf.bind();
    arrayBuf.allocate(vertices, 24 * sizeof(VertexData));

    // Transfer index data to VBO 1
    indexBuf.bind();
    indexBuf.allocate(indices, 34 * sizeof(GLushort));
//! [1]
}

[8Observer8]

@shome said in How to extend the Cube OpenGL ES 2.0 example for 3D meshes?:

Or do we need to use other libraries such as assimp or Qt3D?

It depends. Requirements vary and tastes differ. You can also use Qt Quick 3D: https://doc.qt.io/qt-6/qtquick3d-index.html I didn't try to build Qt Quick 3D or Qt 3D to WASM. If you research what is difference between Qt 3D and Qt Quick 3D you can post your results in the topic: Qt 3D vs Qt Quick 3D

This is my example of loading an object from COLLADA: https://github.com/8Observer8/load-3d-models-dae-openglwindow-opengles2-qt6-cpp I parse XML using QDomDocument. You can run it on Web: https://6602ec1abe3c4d529f4f5c7a--chic-pasca-73d8ca.netlify.app/

How it look when you set samples to 4:

Like this:

surfaceFormat.setSamples(4);

[8Observer8]

I changed QOpenGLWindow with QOpenGLWidget in the example above. See how it looks: https://clinquant-pastelito-7b506b.netlify.app/

I had the same issue for QOpenGLWindow: https://bugreports.qt.io/browse/QTBUG-122145

But it was solved by setting Depth buffer size to 24:

    QSurfaceFormat surfaceFormat;
    surfaceFormat.setDepthBufferSize(24);
    surfaceFormat.setSamples(4);
    setFormat(surfaceFormat);

But I don't know how to solve the same issue for QOpenGLWidget. Maybe I should create another bug report to QOpenGLWidget. It is why I like QOpenGLWindow.

[8Observer8]

@8Observer8 said in How to extend the Cube OpenGL ES 2.0 example for 3D meshes?:

Maybe I should create another bug report to QOpenGLWidget.

I have created it: https://bugreports.qt.io/browse/QTBUG-123787

[8Observer8]

@8Observer8 said in How to extend the Cube OpenGL ES 2.0 example for 3D meshes?:

I have created it: https://bugreports.qt.io/browse/QTBUG-123787

This problem with the depth buffer on QOpenGLWidget was solved. You should call the next commands in paintGL() (not in 'initializeGL()' like you can make on QOpenGLWindow):

    glEnable(GL_DEPTH_TEST);
    glEnable(GL_CULL_FACE);

But I found another problem with QOpenGLWidget. The setSamples() method doesn't work on QOpenGLWidget for WebAssembly:

    QSurfaceFormat surfaceFormat;
    // surfaceFormat.setDepthBufferSize(24); // Doesn't work on QOpenGLWidget for WebAssembly
    surfaceFormat.setSamples(4); // Doesn't work on QOpenGLWidget for WebAssembly
    setFormat(surfaceFormat);

Compare the same examples on QOpenGLWidget and QOpenGLWindow. This is QOpenGLWidget: https://6604427c5c350c16aa4551c3--clinquant-pastelito-7b506b.netlify.app/ GitHub: https://github.com/8Observer8/load-3d-models-dae-openglwidget-opengles2-qt6-cpp

This is QOpenGLWindow: https://chic-pasca-73d8ca.netlify.app/ GitHub: https://github.com/8Observer8/load-3d-models-dae-openglwidget-opengles2-qt6-cpp

[8Observer8]

I created this bug report about the problem above with anti-aliasing on QOpenGLWidget: The setSamples() method doesn't work on QOpenGLWidget for WebAssembly

@Even-Oscar-Andersen answered in the bug report comments:

reproduced on 6.6.2,

This might be fixed in the not yet released 6.8.x (you will have to build yourself from dev)

I think that rebuilding Qt dev from source may be useful for those who want to use QOpenGLWidget with Qt GUI on WASM with anti-aliasing. Maybe some people don’t need anti-aliasing. I don't need Qt GUI for my current tasks but I need anti-aliasing. I prefer to use QOpenGLWindow because it allows to make smooth animation for WASM without QTimer (animation with timer isn't smooth) using the setSwapInterval(1) method and the frameSwapped signal:

OpenGLWindow.cpp

OpenGLWindow::OpenGLWindow()
{
    setTitle("OpenGL ES 2.0, Qt6, C++");
    resize(m_initialWindowWidth, m_initialWindowHeight);

    QSurfaceFormat surfaceFormat;
    surfaceFormat.setDepthBufferSize(24);
    surfaceFormat.setSamples(4);
    surfaceFormat.setSwapInterval(1);
    connect(this, SIGNAL(frameSwapped()), this, SLOT(update()));
    setFormat(surfaceFormat);
}

OpenGLWindow.h

#include <QtCore/QElapsedTimer>

class OpenGLWindow : public QOpenGLWindow, private QOpenGLFunctions
{
private:
    QElapsedTimer m_elapsedTimer;
}

void OpenGLWindow::paintGL()
{
    float dt = m_elapsedTimer.elapsed() / 1000.f;
    m_elapsedTimer.restart();
    qDebug() << dt;
}