I have a question regarding WPF.
I have a Canvas that serves as a visual editor! I have a few 'nodes' positioned in the Canvas using 'X' and 'Y' properties (Canvas.Left and Canvas.Top). Now, I need this Canvas to let the user Zoom (in & out) and Pan around, as he want's to.
I implemented kind of a hack to emulate that behavior. This is the Code that let's the user 'pan' around in the Canvas:
///In file MainWindow.xaml.cs
private void ZoomPanCanvas_MouseMove(object sender, MouseEventArgs e) {
if (IsMouseDown) {
///Change the Cursor to Scroll
if (mNetworkUI.Cursor != Cursors.ScrollAll)
mNetworkUI.Cursor = Cursors.ScrollAll;
var currPosition = e.GetPosition(mNetworkUI);
var diff = currPosition - MouseLastPosition;
var p = new Point(diff.X, diff.Y);
mNetworkUI.ViewModel.Network.SetTransformOffset(p);
MouseLastPosition = currPosition;
}
}
///In file NetworkViewModel.cs
public void SetTransformOffset(Point newOffset) {
for (int i = 0; i < Nodes.Count; i++) {
Nodes[i].X += newOffset.X;
Nodes[i].Y += newOffset.Y;
}
}
Where the 'Nodes' are my editor-nodes displayed in the Canvas. The Zooming (with respect to the mouse position works as follows:
///File MainWindow.xaml.cs
private void ZoomPanCanvas_MouseWheel(object sender, MouseWheelEventArgs e) {
///Determine the Scaling Factor and Scale the Rule-Editor
var factor = (e.Delta > 0) ? (1.1) : (1 / 1.1);
currrentScale = factor * currrentScale;
ScaleNetwork();
///Translate the Nodes to the desired Positions
var pos = e.GetPosition(mNetworkUI);
var transform = new ScaleTransform(factor, factor, pos.X, pos.Y);
var offSet = new Point(transform.Value.OffsetX, transform.Value.OffsetY);
mNetworkUI.ViewModel.Network.SetTransformOffset(offSet);
}
///Also in MainWindow.xaml.cs
private void ScaleNetwork() {
mNetworkUI.RenderTransform = new ScaleTransform(currrentScale, currrentScale);
mNetworkUI.Width = ZoomPanCanvas.ActualWidth / currrentScale;
mNetworkUI.Height = ZoomPanCanvas.ActualHeight / currrentScale;
}
So, in the 'panning' I calculate the difference to the last mouse position and use that vector to manipulate the nodes, not the Canvas itself. When I zoom, I determine the new zoom, set a new RenderTransform, resize the Canvas to again fill the provided space and again re-position the nodes in the Canvas.
It works very well for now. I can 'pan & zoom' around how I want, but I realized, that with many nodes present in my 'network' (connected nodes), things get quite slow.
One reason is, that on every movement of a node some events are raised resulting in a noticable delay when panning.
How is such a thing (without fixed Canvas-size and Scrollbars) possible in a performant manner? Is there a control out there that I can use? Is this possible with the Extended WPF toolkit's ZoomBox control?
Thank you!