纯opengl变焦相机到鼠标pos

Question

Im trying to edit my mouse zoom function to zoom towards the mouse pos in 3d world, I can get the 3d cords of the mouse if i need that just not sure how to do it.

At the moment my zoom just zooms towards the screen center.

void CCamera::OnMouseWheel(float zDelta)
{
    Position -= Reference;

    if(zDelta < 0 && length(Position) < 500.0f)
    {
        Position += Position * 0.1f;
    }

    if(zDelta > 0 && length(Position) > 0.05f)
    {
        Position -= Position * 0.1f;
    }

    Position += Reference;

    CalculateViewMatrix();
}

void CCamera::CalculateViewMatrix()
{
    ViewMatrix = mat4x4(X.x, Y.x, Z.x, 0.0f, X.y, Y.y, Z.y, 0.0f, X.z, Y.z, Z.z, 0.0f, -dot(X, Position), -dot(Y, Position), -dot(Z, Position), 1.0f);
    ViewMatrixInverse = inverse(ViewMatrix);
    ViewProjectionMatrix = ProjectionMatrix * ViewMatrix;
    ViewProjectionMatrixInverse = ViewMatrixInverse * ProjectionMatrixInverse;
}

Answer 1

You may need to do some editing on your code to make sure everything is compatible such as editing the viewgraph. But here is a code snippet from my project that does exactly what you are saying. I think that the code is pretty self-explanatory,

void modelDefinition::onMouseWheel(wxMouseEvent &event)
{
    if(event.GetWheelRotation() != 0)
    {
        /* This section of the code was adapted from Agro2D */

        _cameraX += (((2.0 / this->GetSize().GetWidth()) * (event.GetX() - this->GetSize().GetWidth() / 2.0)) / _zoomFactor) * (this->GetSize().GetWidth() / this->GetSize().GetHeight());
        _cameraY += (-(2.0 / this->GetSize().GetHeight()) * (event.GetY() - this->GetSize().GetHeight() / 2.0)) / _zoomFactor;

        if(!_preferences.getMouseZoomReverseState())
        {
            if(event.GetWheelRotation() > 0)
                _zoomFactor *= pow(1.2, -(event.GetWheelDelta()) / 150.0);
            else
                _zoomFactor *= pow(1.2, (event.GetWheelDelta()) / 150.0);
        }
        else
        {
            if(event.GetWheelRotation() < 0)
                _zoomFactor *= pow(1.2, -(event.GetWheelDelta()) / 150.0);
            else
                _zoomFactor *= pow(1.2, (event.GetWheelDelta()) / 150.0);
        }

        /* This will recalculate the new position of the mouse. Assuming that the mouse does not move at all during the process
         * This also enables the feature where the zoom will zoom in/out at the position of the mouse */

        _cameraX -= (((2.0 / this->GetSize().GetWidth()) * (event.GetX() - this->GetSize().GetWidth() / 2.0)) / _zoomFactor) * (this->GetSize().GetWidth() / this->GetSize().GetHeight());
        _cameraY -= (-(2.0 / this->GetSize().GetHeight()) * (event.GetY() - this->GetSize().GetHeight() / 2.0)) / _zoomFactor;
    }

    this->Refresh();// This will force the canvas to experience a redraw event
}

I should mention that the function this->Refresh will just cause my screen to update. Yours might be different.

Also, _cameraX and _cameraY stores the offset of the openGL canvas.

Here is the constructor for my canvas:

modelDefinition::modelDefinition(wxWindow *par, const wxPoint &point, const wxSize &size, problemDefinition &definition) : wxGLCanvas(par, wxID_ANY, NULL, point, size, wxBORDER_DOUBLE | wxBORDER_RAISED)
{
    _geometryContext = new wxGLContext(this);
    wxGLCanvas::SetCurrent(*_geometryContext);

    _localDefinition = &definition;

    _editor.setZoomFactorAddress(_zoomFactor);

    glViewport(0, 0, (double)this->GetSize().GetWidth(), (double)this->GetSize().GetHeight());

    glClearColor(1.0f, 1.0f, 1.0f, 1.0f);
    glMatrixMode(GL_MODELVIEW);// The matrix mode specifies which matrix stack is the target for matrix operations
    glLoadIdentity();// Initial value
    glTranslated((float)this->GetSize().GetWidth() / 2.0f, (float)this->GetSize().GetHeight() / 2.0f, 0.0f);// This will move the camera to the center of the screen

    glMatrixMode(GL_PROJECTION);
    glLoadIdentity();

    glOrtho(-1.0, 1.0, -1.0, 1.0, -1.0, 1.0);
    GLenum error = glGetError();
    if(error != GL_NO_ERROR)
    {
    //  wxMessageBox("Error - " + gluErrorString(error));
        return;
    }

    glMatrixMode(GL_MODELVIEW);
}

And here is the code that handles updating the position of my view:

void modelDefinition::updateProjection()
{
    // First, load the projection matrix and reset the view to a default view
    glMatrixMode(GL_PROJECTION);
    glLoadIdentity();

    //Reset to modelview matrix
    glMatrixMode(GL_MODELVIEW);
    glLoadIdentity();

    /* This section will handle the translation (panning) and scaled (zooming). 
     * Needs to be called each time a draw occurs in order to update the placement of all the components */
    if(_zoomFactor < 1e-9)
        _zoomFactor = 1e-9;

    if(_zoomFactor > 1e6)
        _zoomFactor = 1e6;

    glScaled(_zoomFactor / (this->GetSize().GetWidth() / this->GetSize().GetHeight()), _zoomFactor, 1.0);

    glTranslated(-_cameraX, -_cameraY, 0.0);
}

Answer 2

Let's suppose you have the 3D coordinates of the mouse cursor. Undoing the viewport transformation you get these coordinates in NDC space, and also by matrix inversion you can get View space or Model space coordinates. Now, you have a vector from camera to these mouse coordinates (in whatever space you like). The last step is moving the camera following that vector by an amount depending on mouse wheel-rotation.

If you don't have those 3D mouse coordinates... you have a big problem, due to it's not possible to get 3D information from a 2D data. You lack the "depth in the screen", Z value in NDC space.