One approach is to overload the MouseWheel event handler. The current coordinates of the mouse are available here, too.
- Use the mouse screen coordinates to acquire (to "pick") the world
coordinate corresponding to the primitive under the mouse.
- Adjust the
Camera.Position and Camera.ZoomFactor to 'move' the camera closer to the point under the mouse and to achieve the required 'directional zoom' effect.
Here is a complete example from the ILNumerics website:
using System;
using System.Windows.Forms;
using ILNumerics;
using ILNumerics.Drawing;
using ILNumerics.Drawing.Plotting;
using static ILNumerics.Globals;
using static ILNumerics.ILMath;
namespace ILNumerics.Examples.DirectionalZoom {
public partial class Form1 : Form {
public Form1() {
InitializeComponent();
}
private void panel2_Load(object sender, EventArgs e) {
Array<float> X = 0, Y = 0, Z = CreateData(X, Y);
var surface = new Surface(Z, X, Y, colormap: Colormaps.Winter);
surface.UseLighting = true;
surface.Wireframe.Visible = false;
panel2.Scene.Camera.Add(surface);
// setup mouse handlers
panel2.Scene.Camera.Projection = Projection.Orthographic;
panel2.Scene.Camera.MouseDoubleClick += Camera_MouseDoubleClick;
panel2.Scene.Camera.MouseWheel += Camera_MouseWheel;
// initial zoom all
ShowAll(panel2.Scene.Camera);
}
private void Camera_MouseWheel(object sender, Drawing.MouseEventArgs e) {
// Update: added comments.
// the next conditionals help to sort out some calls not needed. Helpful for performance.
if (!e.DirectionUp) return;
if (!(e.Target is Triangles)) return;
// make sure to start with the SceneSyncRoot - the copy of the scene which receives
// user interaction and is eventually used for rendering. See: https://ilnumerics.net/scene-management.html
var cam = panel2.SceneSyncRoot.First<Camera>();
if (Equals(cam, null)) return; // TODO: error handling. (Should not happen in regular setup, though.)
// in case the user has configured limited interaction
if (!cam.AllowZoom) return;
if (!cam.AllowPan) return; // this kind of directional zoom "comprises" a pan operation, to some extent.
// find mouse coordinates. Works only if mouse is over a Triangles shape (surfaces, but not wireframes):
using (var pick = panel2.PickPrimitiveAt(e.Target as Drawable, e.Location)) {
if (pick.NextVertex.IsEmpty) return;
// acquire the target vertex coordinates (world coordinates) of the mouse
Array<float> vert = pick.VerticesWorld[pick.NextVertex[0], r(0, 2), 0];
// and transform them into a Vector3 for easier computations
var vertVec = new Vector3(vert.GetValue(0), vert.GetValue(1), vert.GetValue(2));
// perform zoom: we move the camera closer to the target
float scale = Math.Sign(e.Delta) * (e.ShiftPressed ? 0.01f : 0.2f); // adjust for faster / slower zoom
var offs = (cam.Position - vertVec) * scale; // direction on the line cam.Position -> target vertex
cam.Position += offs; // move the camera on that line
cam.LookAt += offs; // keep the camera orientation
cam.ZoomFactor *= (1 + scale);
// TODO: consider adding: the lookat point now moved away from the center / the surface due to our zoom.
// In order for better rotations it makes sense to place the lookat point back to the surface,
// by adjusting cam.LookAt appropriately. Otherwise, one could use cam.RotationCenter.
e.Cancel = true; // don't execute common mouse wheel handlers
e.Refresh = true; // immediate redraw at the end of event handling
}
}
private void Camera_MouseDoubleClick(object sender, Drawing.MouseEventArgs e) {
var cam = panel2.Scene.Camera;
ShowAll(cam);
e.Cancel = true;
e.Refresh = true;
}
// Some sample data. Replace this with your own data!
private static RetArray<float> CreateData(OutArray<float> Xout, OutArray<float> Yout) {
using (Scope.Enter()) {
Array<float> x_ = linspace<float>(0, 20, 100);
Array<float> y_ = linspace<float>(0, 18, 80);
Array<float> Y = 1, X = meshgrid(x_, y_, Y);
Array<float> Z = abs(sin(sin(X) + cos(Y))) + .01f * abs(sin(X * Y));
if (!isnull(Xout)) {
Xout.a = X;
}
if (!isnull(Yout)) {
Yout.a = Y;
}
return -Z;
}
}
// See: https://ilnumerics.net/examples.php?exid=7b0b4173d8f0125186aaa19ee8e09d2d
public static double ShowAll(Camera cam) {
// Update: adjusts the camera Position too.
// this example works only with orthographic projection. You will need to take the view frustum
// into account, if you want to make this method work with perspective projection also. however,
// the general functioning would be similar....
if (cam.Projection != Projection.Orthographic) {
throw new NotImplementedException();
}
// get the overall extend of the cameras scene content
var limits = cam.GetLimits();
// take the maximum of width/ height
var maxExt = limits.HeightF > limits.WidthF ? limits.HeightF : limits.WidthF;
// make sure the camera looks at the unrotated bounding box
cam.Reset();
// center the camera view
cam.LookAt = limits.CenterF;
cam.Position = cam.LookAt + Vector3.UnitZ * 10;
// apply the zoom factor: the zoom factor will scale the 'left', 'top', 'bottom', 'right' limits
// of the view. In order to fit exactly, we must take the "radius"
cam.ZoomFactor = maxExt * .50;
return cam.ZoomFactor;
}
}
}
Note, that the new handler performs the directional zoom only when the mouse is located over an object hold by this Camera! If, instead, the mouse is placed on the background of the scene or over some other Camera / plot cube object no effect will be visible and the common zoom feature is performed (zooming in/out to the look-at point).
