Silverlight 5: Terrible Performance Issues with DropShadowEffect - silverlight

Environment Info:
Windows 7 64bit SP1
6GB Memory
Intel(R) Core(TM) i5-2400S CPU # 2.50GHz (4 cores)
Silverlight 5
I have very terrible performance issues with DropShadowEffect in Silerlight.
I created a minimized demo for it:
http://www.peterlee.com.cn/public/ShadowEffectTestDemo/ShadowEffectTestTestPage.html
Click the button Zoom to try to zoom the canvas (with a Star shape with DropShadowEffect). You will find that when the Zoom = 64, your CPU and memory are very crazy.
But, if you remove the DropShadowEffect by clicking the Remove Effect button, and then zoom it, everything is okay.
However, if we use a TextBlock with DropShadowEffect, everything is fine. You can try it by clicking the "Use TextBlock` button.
I don't know what Silverlight is dealing with the TextBlock and my customized Star shape for the DropShadowEffect.
Please help me out. Thanks.
UPDATE:
According to the replies:
I also tried it in Release mode (now the sample demo link was built in Release mode);
I also added GPA acceleration suggested by ChrisF:
<object data="data:application/x-silverlight-2," type="application/x-silverlight-2" width="100%" height="100%">
<param name="EnableGPUAcceleration" value="true" />
<param name="source" value="ShadowEffectTest.xap"/>
<param name="onError" value="onSilverlightError" />
<param name="background" value="white" />
<param name="minRuntimeVersion" value="5.0.61118.0" />
<param name="autoUpgrade" value="true" />
<a href="http://go.microsoft.com/fwlink/?LinkID=149156&v=5.0.61118.0" style="text-decoration:none">
<img src="http://go.microsoft.com/fwlink/?LinkId=161376" alt="Get Microsoft Silverlight" style="border-style:none"/>
</a>
</object>
Here is the UPDATED code
MainPage.xmal:
<UserControl x:Class="ShadowEffectTest.MainPage"
xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
xmlns:d="http://schemas.microsoft.com/expression/blend/2008"
xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"
mc:Ignorable="d"
d:DesignHeight="300" d:DesignWidth="400">
<Grid x:Name="LayoutRoot" Background="White">
<Grid.RowDefinitions>
<RowDefinition Height="Auto" />
<RowDefinition Height="Auto" />
</Grid.RowDefinitions>
<Grid.ColumnDefinitions>
<ColumnDefinition Width="Auto" />
<ColumnDefinition Width="Auto" />
</Grid.ColumnDefinitions>
<Button x:Name="btnZoom" Width="75" Height="23" Content="Zoom"
Grid.Column="0" Grid.Row="0"
Click="btnZoom_Click"/>
<Button x:Name="btnRemoveEffect" Width="100" Height="23" Content="Remove Effect"
Grid.Row="0" Grid.Column="1"
Click="btnRemoveEffect_Click"/>
<Button x:Name="btnUseTextBlock" Width="120" Height="23" Content="Use TextBlock"
Grid.Row="1" Grid.Column="0"
Click="btnUseTextBlock_Click" />
<Canvas x:Name="mainCanvas" Width="200" Height="150" Background="LightBlue"
Grid.Column="1" Grid.Row="1" />
</Grid>
</UserControl>
MainPage.xaml.cs:
using System;
using System.Collections.Generic;
using System.Linq;
using System.Net;
using System.Windows;
using System.Windows.Controls;
using System.Windows.Documents;
using System.Windows.Input;
using System.Windows.Media;
using System.Windows.Media.Animation;
using System.Windows.Shapes;
using System.Diagnostics;
using System.Windows.Media.Effects;
namespace ShadowEffectTest
{
public partial class MainPage : UserControl
{
public MainPage()
{
InitializeComponent();
// These are the points for a Star shape
List<Point> points = new List<Point>()
{
new Point(0, 36.3634469292486),
new Point(-12.3606797688474, 43.2188191057189),
new Point(-24.7213595376948, 50.0741912821893),
new Point(-37.0820393065422, 56.9295634586596),
new Point(-34.4313595448175, 42.9654855127096),
new Point(-31.7806797830927, 29.0014075667595),
new Point(-29.130000021368, 15.0373296208095),
new Point(-39.4200000080385, 5.31010468057062),
new Point(-49.709999994709, -4.41712025966827),
new Point(-59.9999999813794, -14.1443451999072),
new Point(-46.0011100186002, -15.919247828416),
new Point(-32.002220055821, -17.694150456925),
new Point(-18.0033300930418, -19.4690530854338),
new Point(-12.0022200620278, -32.3361808961241),
new Point(-6.00111003101392, -45.2033087068145),
new Point(0, -58.0704365175048),
new Point(6.00111003101392, -45.2033087068145),
new Point(12.0022200620278, -32.3361808961241),
new Point(18.0033300930418, -19.4690530854338),
new Point(32.002220055821, -17.694150456925),
new Point(46.0011100186002, -15.919247828416),
new Point(59.9999999813794, -14.1443451999072),
new Point(49.709999994709, -4.41712025966827),
new Point(39.4200000080385, 5.31010468057062),
new Point(29.130000021368, 15.0373296208095),
new Point(31.7806797830927, 29.0014075667595),
new Point(34.4313595448175, 42.9654855127096),
new Point(37.0820393065422, 56.9295634586596),
new Point(24.7213595376948, 50.0741912821893),
new Point(12.3606797688474, 43.2188191057189),
new Point(0, 36.3634469292486)
};
uie = RenderBezier(points);
// This the evil for the performance issues (crazy memory and CPU)
uie.Effect = ShadowEffect;
uie.CacheMode = new BitmapCache();
uie.SetValue(Canvas.LeftProperty, 25.0);
uie.SetValue(Canvas.TopProperty, 25.0);
mainCanvas.Children.Add(uie);
}
private UIElement uie = null;
public Path RenderBezier(List<Point> ctrlPoints, List<List<Point>> innersCtrlPoints = null)
{
// Step 0: Merge ctrlPoints lists
List<List<Point>> ctrlPointsLists;
if (innersCtrlPoints == null)
{
ctrlPointsLists = new List<List<Point>>(1);
ctrlPointsLists.Add(ctrlPoints);
}
else
{
ctrlPointsLists = new List<List<Point>>(1 + innersCtrlPoints.Count);
ctrlPointsLists.Add(ctrlPoints);
foreach (List<Point> list in innersCtrlPoints)
ctrlPointsLists.Add(list);
}
PathGeometry pg = new PathGeometry();
foreach (List<Point> list in ctrlPointsLists)
{
// Step 0: check (Debug.Assert)
Debug.Assert(list.Count % 3 == 1,
"public Path RenderBezier(IList<Point> ctrlPoints): number of control points is not 3n+1.");
int n = (list.Count - 1) / 3; // Number of BezierSegments
Debug.Assert(n > 0,
"public Path RenderBezier(IList<Point> ctrlPoints): at least one Bezier segment required.");
// Step 1: Add BezierSegments to PathFigure
PathFigure pf = new PathFigure();
pf.StartPoint = list[0];
for (int i = 0; i < n; ++i)
pf.Segments.Add(GetBezierSegment(
list[3 * i + 1],
list[3 * i + 2],
list[3 * i + 3]
));
// Step 2: Add PathFigures to PathGeometry
pg.Figures.Add(pf);
}
// Step 3: Add PathGemotry to GeometryGroup
GeometryGroup gg = new GeometryGroup();
gg.Children.Add(pg);
// Step 4: Set GeometryGroup as Path.Data
Path path = new Path();
path.Data = gg;
// Step 5: Set some Path properties
// if (ShowOutline)
{
path.Stroke = new SolidColorBrush(Colors.Black);
path.StrokeThickness = 1.0;
path.StrokeEndLineCap = PenLineCap.Round;
path.StrokeLineJoin = PenLineJoin.Round;
path.StrokeStartLineCap = PenLineCap.Round;
}
// Finally, return it
return path;
}
// This the evil for the performance issues (crazy memory and CPU)
private static DropShadowEffect ShadowEffect
{
get
{
return new DropShadowEffect()
{
Color = Colors.Blue,
BlurRadius = 5,
Direction = 0,
ShadowDepth = 0
};
}
}
private static BezierSegment GetBezierSegment(Point p1, Point p2, Point p3)
{
BezierSegment bs = new BezierSegment();
bs.Point1 = p1;
bs.Point2 = p2;
bs.Point3 = p3;
return bs;
}
public static readonly double[] ZoomingSteps = new double[]
{
1.0,
1.5,
2.0,
3.0,
4.0,
6.0,
8.0,
12.0,
16.0,
24.0,
32.0,
48.0,
64.0,
128.0
};
private int index = 0;
private void btnZoom_Click(object sender, RoutedEventArgs e)
{
if (index >= ZoomingSteps.Length - 1)
return;
ScaleTransform st = new ScaleTransform();
st.ScaleX = st.ScaleY = ZoomingSteps[index++];
btnZoom.Content = ZoomingSteps[index].ToString();
mainCanvas.RenderTransform = st;
}
private void btnRemoveEffect_Click(object sender, RoutedEventArgs e)
{
index = 0;
btnZoom.Content = "Zoom";
uie.Effect = null;
mainCanvas.RenderTransform = new ScaleTransform();
}
// If I use TextBlock as the child UIElement, then everything is okay
// path = new TextBlock() { Text = "Text Block" };
private void btnUseTextBlock_Click(object sender, RoutedEventArgs e)
{
mainCanvas.Children.Remove(uie);
index = 0;
btnZoom.Content = "Zoom";
uie = new TextBlock() { Text = "Text Block" };
mainCanvas.Children.Add(uie);
uie.Effect = ShadowEffect;
uie.CacheMode = new BitmapCache();
mainCanvas.RenderTransform = new ScaleTransform();
}
}
}

I've tried running WinDbg against your application and dumpheap the application. here is the Result. Now dont worry, I dont understand half of it either. But I went a little deeper and there seem to be large amounts of arrays/lists where atleast 80% are null.
You might want to look at WinDbg yourself. Here is a small Tutorial
I'll try to look at it in the morning if I can and I hope I've helped you atleast in the right direction

Here's my trick: never use DropShadowEffect. Create a clone of whatever you want to create a drop shadow for, put it behind the main content, and use a BlurEffect on it. Much better performance, much better quality. I have no idea why.

I think the problem is here:
uie.CacheMode = new BitmapCache();
Typically, this tells silverlight to try and cache the bitmap in memory so that the computer does not need to re-render the element in your application.
Look here for the suggested tips on silverlight performance:
http://msdn.microsoft.com/en-us/library/cc189071(v=vs.95).aspx

The problem has to do with how the rendering pipeline of Silverlight works. The DropShadowEffect is just a pixel shader, and pixel shaders work in Silverlight by creating a buffered copy of the image they will be applied to and then allow you to change pixel properties of the real image using the values of the buffered copy. This buffered copy is the entire image and is not affected by clipping. Because zooming creates a very large pre-clipped image, the pixel shader must make a physical buffered copy of a very large pre-clipped image...this is why the performance is so bad when you zoom.
The only solution I have found for this is to disable the pixel shader effect while you are zooming (or panning the zoomed image), and add it back when the image is zoomed in (or panning is complete). This way you don't incur the performance penalty of the pixel shader being applied for every frame of zooming or panning.

Related

Dynamically modifying a Path, Geometry, Shape in XAML

I am stuck. I want to do some sophisticated animations in XAML, in which the geometry of the image gets modified at runtime. I want to start out simple, and then make something far more interesting, but nothing seems to work.
For now, all I want to do is draw an arc on the screen using a for-loop, and this arc will be modified by a slider. The slider will control the starting point of the arc. So, if I have the slider at Pi/4, it will draw an arc from Pi/4 to 2Pi.
I've tried so many different ways. Right now I've created a class of the type Shape, and tried to modify the DefiningGeometry which is a property of the Shape class. The startRadians gets modified by the slider, so that part works OK, I got the binding to work. But, after startRadians gets changed (it is a DependencyProperty, btw) I want the class to re-calculate the geometry of the circle. (Like a cherry pie that is missing a bigger piece as startRadians gets changed.) The real problem is that DefiningGeometry is a read-only property, so I can't change it on the fly. (Am I right about this?) Even if I could, I don't know the way to write the line of code so that DrawMyArc fires again, and the results get reloaded into DefiningGeometry.
OK, so I need some guidance. Should I change the parent class, so that I have an easily modifiable Path/geometry? I am at a loss here.
Should I use an entirely different approach, like where you dynamically make/delete the geometry using StreamGeometry?
Here's the relevant code:
using System;
using System.Collections.Generic;
using System.Windows;
using System.Windows.Media;
using System.Windows.Shapes;
namespace January14noon
{
public class myCircle : Shape
{
public double startRadians
{
get { return (double)GetValue(startRadiansProperty); }
set { SetValue(startRadiansProperty, value); }
}
protected override Geometry DefiningGeometry
{
get
{
return DrawMyArc(100, 200, true, 40, 40, 360, startRadians, 2 * Math.PI);
}
}
// Using a DependencyProperty as the backing store for startRadians. This enables animation, styling, binding, etc...
public static readonly DependencyProperty startRadiansProperty =
DependencyProperty.Register("startRadians", typeof(double), typeof(myCircle), new PropertyMetadata(Math.PI / 4, new PropertyChangedCallback(startRadians_PropertyChanged)));
private static void startRadians_PropertyChanged(DependencyObject d, DependencyPropertyChangedEventArgs e)
{
//
//
}
public PathGeometry DrawMyArc(double centX, double centY, bool CW, double radiusX, double radiusY, int numberOfSegs, double startRad, double endRad)
{
double[,] rawPoints = new double[numberOfSegs, 2];
List<LineSegment> segments = new List<LineSegment>();
double arcLength;
arcLength = endRad - startRad;
for (int i = 0; i < numberOfSegs; i++)
{
rawPoints[i, 0] = radiusX * Math.Sin(i * (arcLength / numberOfSegs) + startRad) + centX;
rawPoints[i, 1] = radiusY * -Math.Cos(i * (arcLength / numberOfSegs) + startRad) + centY;
segments.Add(new LineSegment(new Point(rawPoints[i, 0], rawPoints[i, 1]), true));
}
LineSegment[] segArray = segments.ToArray();
PathFigure figure = new PathFigure(new Point(centX, centY), segments, false);
PathGeometry myGeometry = new PathGeometry();
myGeometry.Figures.Add(figure);
return myGeometry;
}
}
}
And XAML:
<Window x:Class="January14noon.MainWindow"
xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
xmlns:d="http://schemas.microsoft.com/expression/blend/2008"
xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"
xmlns:local="clr-namespace:January14noon"
mc:Ignorable="d"
Title="MainWindow" Height="350" Width="525">
<Window.Resources>
<local:myCircle x:Key="myCircleDataSource" d:IsDataSource="True"/>
</Window.Resources>
<Grid DataContext="{Binding Source={StaticResource myCircleDataSource}}">
<Slider x:Name="slider" VerticalAlignment="Top" Margin="0,0,163.898,0" Value="{Binding startRadians, Mode=OneWayToSource}"/>
<TextBox x:Name="myTextBox" HorizontalAlignment="Right" Height="23" TextWrapping="Wrap" Text="{Binding startRadians}" VerticalAlignment="Top" Width="147.797"/>
<!--<local:myCircle x:Name="instanceOfCircle" />-->
<local:myCircle Stroke="Black" StrokeThickness="2"/>
</Grid>
</Window>
Any help would be appreciated. Just general approach, something specific even words of encouragement.
TYIA
Any dependency property that affects visual appearance of a control might be registered with appropriate FrameworkPropertyMetadataOptions, e.g. AffectsMeasure, AffectsArrange or AffectsRender.
Note also that class, property and method names in C# are supposed to use Pascal Casing, i.e. start with an uppercase letter
public static readonly DependencyProperty StartRadiansProperty =
DependencyProperty.Register(nameof(StartRadians), typeof(double), typeof(MyCircle),
new FrameworkPropertyMetadata(Math.PI / 4,
FrameworkPropertyMetadataOptions.AffectsMeasure, StartRadiansPropertyChanged));
public double StartRadians
{
get { return (double)GetValue(StartRadiansProperty); }
set { SetValue(StartRadiansProperty, value); }
}
private static void StartRadiansPropertyChanged(
DependencyObject d, DependencyPropertyChangedEventArgs e)
{
...
}
Make sure you bind the visible circle to the data source:
<Window.Resources>
<local:myCircle x:Key="myCircleDataSource" d:IsDataSource="True"/>
</Window.Resources>
<Grid DataContext="{Binding Source={StaticResource myCircleDataSource}}">
<Slider x:Name="slider" VerticalAlignment="Top" Margin="0,0,163.898,0" Value="{Binding startRadians, Mode=OneWayToSource}"/>
<TextBox x:Name="myTextBox" HorizontalAlignment="Right" Height="23" TextWrapping="Wrap" Text="{Binding startRadians}" VerticalAlignment="Top" Width="147.797"/>
<!--<local:myCircle x:Name="instanceOfCircle" />-->
<local:myCircle Stroke="Black" StrokeThickness="2" startRadians="{Binding startRadians}"/>
</Grid>
and invalidate the visual when the property changes:
private static void startRadians_PropertyChanged(DependencyObject d, DependencyPropertyChangedEventArgs e)
{
var circle = (myCircle)d;
circle.InvalidateVisual(); // <-- Invalidate!
}
Invalidating will tell engine to re-render this visual, which will call your DrawMyArc() method

scrollable wrap with overflow

I need to display a block of text in a resizable column. The text should wrap with overflow but, for a given column size, the user should be able to scroll horizontally to view the overflown text.
I do not believe this can be achieved w/ out of the box controls; if I'm wrong about that please tell me. I have attempted to achieve this with a custom control:
public class Sizer : ContentPresenter
{
static Sizer()
{
ContentPresenter.ContentProperty.OverrideMetadata(typeof(Sizer), new FrameworkPropertyMetadata(ContentChanged)); ;
}
public Sizer() : base() {}
protected override Size MeasureOverride(Size constraint)
{
var childWidth = Content==null ? 0.0 : ((FrameworkElement)Content).RenderSize.Width;
var newWidth = Math.Max(RenderSize.Width, childWidth);
return base.MeasureOverride(new Size(newWidth, constraint.Height));
}
private static void ContentChanged(DependencyObject dep, DependencyPropertyChangedEventArgs args)
{
var #this = dep as Sizer;
var newV = args.NewValue as FrameworkElement;
var oldV = args.OldValue as FrameworkElement;
if (oldV != null)
oldV.SizeChanged -= #this.childSizeChanged;
if(newV!=null)
newV.SizeChanged += #this.childSizeChanged;
}
private void childSizeChanged(object sender, SizeChangedEventArgs e)
{
this.InvalidateMeasure();
}
}
...and I can test it in a simple WPF application like so:
<Grid>
<Grid.ColumnDefinitions>
<ColumnDefinition Width="200"/>
<ColumnDefinition Width="*"/>
</Grid.ColumnDefinitions>
<ScrollViewer Grid.Column="0" VerticalScrollBarVisibility="Disabled" HorizontalScrollBarVisibility="Auto" VerticalAlignment="Top">
<local:Sizer>
<local:Sizer.Content>
<TextBlock Background="Coral" Text="aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaa"
TextWrapping="WrapWithOverflow" />
</local:Sizer.Content>
</local:Sizer>
</ScrollViewer>
<GridSplitter Grid.Column="1" Background="Black" VerticalAlignment="Stretch" HorizontalAlignment="Left" Width="5" />
</Grid>
This works, after a fashion. It wraps and displays the text correctly, and the horizontal scrollbar lets me view the overflown text. If I resize the column to the right (larger) the text re-wraps correctly. However, if I resize the column to the left (smaller) the text will not re-wrap. So, for instance, if I resize the column to the right so that all the text is on one line it will remain all on one line regardless of any subsequent re-sizing. This is an unacceptable bug.
I have tinkered w/ this code a great deal although I haven't had what you'd a call a good strategy for finding a solution. I do not know and have not been able to discover any mechanism for forcing a textblock to re-wrap its contents. Any advice?
I was able to get this working (w/ some limitations) by adding the following to the custom control:
//stores first ScrollViewer ancestor
private ScrollViewer parentSV = null;
//stores collection of observed valid widths
private SortedSet<double> wrapPoints = new SortedSet<double>();
protected override void OnVisualParentChanged(DependencyObject oldParent)
{
parentSV = this.FindParent<ScrollViewer>();
base.OnVisualParentChanged(oldParent);
}
... and editing MeasureOverride list this:
protected override Size MeasureOverride(Size constraint)
{
if (parentSV != null)
{
var childWidth = Content == null ? 0.0 : ((FrameworkElement)Content).RenderSize.Width;
var viewportWidth = parentSV.ViewportWidth;
if (childWidth > viewportWidth + 5)
wrapPoints.Add(childWidth);
var pt = wrapPoints.FirstOrDefault(d => d > viewportWidth);
if (pt < childWidth)
childWidth = pt;
return base.MeasureOverride(new Size(childWidth, constraint.Height));
}
else
return base.MeasureOverride(constraint);
}
I do not like this at all. It doesn't work for WrapPanels (although that isn't a required use case for me), it flickers occasionally and I'm concerned that the wrapPoints collection may grow very large. But it does what I need it to do.

How to obtain the scaled size of a WPF Visual element

I am rendering a WPF Visual (UserControl) to a bitmap but the problem is that the rendered image is the size of the UserControl before it is scaled/transformed. So let's say the UserControl was designed at 200x200 pixels. When I render to BMP I'm using the UserControl's ActualWidth and ActualHeightt which report 200 and 200 respectively. Problem is the UserControl is in a Canvas and is auto sized (set to scale/fill with the Window size) to something closer to 1200 x 1200 (it changes)
I've done some reading and searching and so far can't figure out how to determine the effective size, that is the size the control is being painted on screen.
I came across this question which sounded hopeful but the Transform returned does not contain scaling data. Well it does, but they are both 1.
Get element position after transform
Any suggestions on where to look for the render scaling would be great!
[UPDATE] As suggested, I'm including the relevant code:
public static Bitmap PngBitmap(this Visual visual)
{
// Get height and width
int width = (int)(double)visual.GetValue(
FrameworkElement.ActualWidthProperty);
int height = (int)(double)visual.GetValue(
FrameworkElement.ActualHeightProperty);
// Render
RenderTargetBitmap rtb =
new RenderTargetBitmap(
width,
height,
96,
96,
PixelFormats.Default);
rtb.Render(visual);
// Encode
PngBitmapEncoder encoder = new PngBitmapEncoder();
encoder.Frames.Add(BitmapFrame.Create(rtb));
System.IO.MemoryStream stream = new System.IO.MemoryStream();
encoder.Save(stream);
// Create Bitmap
Bitmap bmp = new Bitmap(stream);
stream.Close();
return bmp;
}
public static BitmapSource BitmapSource(this Visual visual)
{
Bitmap bmp = visual.PngBitmap();
IntPtr hBitmap = bmp.GetHbitmap();
BitmapSizeOptions sizeOptions = BitmapSizeOptions.FromEmptyOptions();
return Imaging.CreateBitmapSourceFromHBitmap(
hBitmap,
IntPtr.Zero,
Int32Rect.Empty,
sizeOptions);
}
[Update #2] Added the XAML - The Grid element was removed because it was HUGE and from my reading of the XAML the Canvas containing the keyboard UserControl was NOT part of the Grid element.
<UserControl xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
xmlns:d="http://schemas.microsoft.com/expression/blend/2008"
xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"
xmlns:local="clr-namespace:PMD.HECAT.DashboardModule"
xmlns:PMD_HECAT_DashboardModule_VirtualKeyboard="clr-namespace:PMD.HECAT.DashboardModule.VirtualKeyboard"
xmlns:System_Windows_Controls="clr-namespace:System.Windows.Controls;assembly=System.Windows.Controls.Input.Toolkit"
xmlns:PMD_HECAT_DashboardModule_Input="clr-namespace:PMD.HECAT.DashboardModule.Input"
xmlns:control="clr-namespace:PMD.HECAT.DashboardModule.Controls"
x:Class="PMD.HECAT.DashboardModule.CandidateElectrodeView"
x:Name="UserControl"
mc:Ignorable="d"
d:DesignWidth="1024" d:DesignHeight="768" Width="640" Height="360">
<UserControl.Resources>
<ResourceDictionary>
<ResourceDictionary.MergedDictionaries>
<ResourceDictionary Source="../Themes/DashboardStyles.xaml" />
<ResourceDictionary Source="../Themes/ImageButtons.xaml" />
<ResourceDictionary Source="CandidateViewResources.xaml" />
</ResourceDictionary.MergedDictionaries>
</ResourceDictionary>
</UserControl.Resources>
<UserControl.Triggers>
<EventTrigger RoutedEvent="PMD_HECAT_DashboardModule_VirtualKeyboard:VirtualKeyboardView.KeyboardClose" SourceName="virtualKeyboardView">
<BeginStoryboard Storyboard="{StaticResource OnKeyboardClose1}"/>
</EventTrigger>
</UserControl.Triggers>
<Canvas Width="100" HorizontalAlignment="Left">
<PMD_HECAT_DashboardModule_VirtualKeyboard:VirtualKeyboardView x:Name="virtualKeyboardView" Height="222" Width="550" RenderTransformOrigin="0.5,0.5" Opacity="0" Active="False">
<PMD_HECAT_DashboardModule_VirtualKeyboard:VirtualKeyboardView.RenderTransform>
<TransformGroup>
<ScaleTransform/>
<SkewTransform/>
<RotateTransform/>
<TranslateTransform X="40" Y="400"/>
</TransformGroup>
</PMD_HECAT_DashboardModule_VirtualKeyboard:VirtualKeyboardView.RenderTransform>
</PMD_HECAT_DashboardModule_VirtualKeyboard:VirtualKeyboardView>
<Rectangle Stroke="White" Opacity="0.7" Fill="White" Height="370" Width="654.851" Canvas.Left="687" Canvas.Top="0" />
</Canvas>
</UserControl>
I know lots of time passed since the question was asked but it doesn't hurt to post some more info :)
I use this code to find size (scale) of visuals with applied render transform(s).
GeneralTransform t = transformedVisual.TransformToVisual(parentVisual);
Vector topLeft = (Vector) t.Transform(new Point(0,0));
Vector topRight = (Vector) t.Transform(new Point(originalWidthOfTransformedVisual, 0));
double renderedWidth = (topRight-topLeft).Length;
double widthScale = renderedWidth / originalWidthOfTransformedVisual;
Hi i had a similar problem if the visual wasn't displayed before to force this rendering
uiElement.Measure(new Size(double.PositiveInfinity, double.PositiveInfinity));
uiElement.Arrange(new Rect(new Point(0, 0), uiElement.DesiredSize));
Size _size = uiElement.DesiredSize;
this works at least in my case to fore the UIElement to Render it Size
If there is a RenderTransform you will hardly be able to determine the size unless you you some serious calculations
In most other cases you will be able to get the size of the FrameworkElement by using the ActualHeight and ActualWidth properties. Still, these properties will give you the size of the bounding box which will be sufficient for the scenario you describe.
You say you use the Visual's width and height.
First, there are two problems.
A Visual, http://msdn.microsoft.com/en-us/library/system.windows.media.visual.aspx does not have any Properties for Height and Width.
Now, assuming you have a FrameworkElement, which does have a Height and Width property, are you using the Height and Width Properties, or the ActualHeight and ActualWidth properties?
If you are not, then those are the properties you are looking for.
If you are using them, or you are not using a FrameworkElement, then you should post some code.
If you could use animation to perform the Transform then animation provides Completed event. That should be a good place to extract new transformed size.
Based on #bor's answer, I simplified the code to allow a more common use:
private static Rect GetTransformedVisualBounds(Visual source, Visual target, Rect bounds)
{
return source.TransformToVisual(target).TransformBounds(bounds);
}
source is a Visual which can have transformations applied;
target is the Visual to which the coordinates will be converted to;
bounds is the non-transformed bounds of the source Visual;
The following is an example of use of the above method:
private void ExampleOfUse()
{
Border border = new Border()
{
Width = 100,
Height = 100
};
ContainerVisual container = new ContainerVisual();
container.Children.Add(border);
container.Transform = new ScaleTransform(2d, 2d);
Rect transformedBounds = GetTransformedVisualBounds(container, this, VisualTreeHelper.GetDescendantBounds(container));
// This should print a rectangle with a size of 200x200;
Debug.Print($"Transformed bounds: {transformedBounds}");
}
This is a very old question, but it was useful for me and I hope it is still useful for someone else as well.
public static Rect GetRelativePlacement(this UIElement element, UIElement relativeTo)
{
var absolutePos = element.PointToScreen(new Point(0, 0));
var posRelativeTo = relativeTo.PointToScreen(new Point(0, 0));
var topLeft = new Point(absolutePos.X - posRelativeTo.X, absolutePos.Y - posRelativeTo.Y);
var bottomRight = element.PointToScreen(new Point(element.RenderSize.Width, element.RenderSize.Height));
var bounds = Rect.Empty;
bounds.Union(topLeft);
bounds.Union(bottomRight);
return bounds;
}

Databound Triangle Creation?

I've got a requirement to create several shapes based on a supplied size (all of them have the same height/width) and have their sizes be databound to that supplied property on the datacontext.
Most of the shapes are easy: Circle (ellipse with height/width bound), square (rectangle with height/width bound), diamond (same as square, then use a RotateTransform), + (two lines), X (two lines).
But I'm trying to figure out how to do it for a triangle and I can't figure it out. It needs to be a filled object, so I can't just do it with three lines.
But all of the ways i've seen to do it (w/ a Path or a Polygon) end up taking Point objects (StartPoint, EndPoint, etc). And you can't bind to the X or Y values of the Point object.
Am I missing something? Or do I need to write my own custom shape or something?
Edit: To add a little bit of clarity... the type of triangle I'm creating doesn't really matter. It can be equilateral or isosceles. I was targeting an isosceles, so that it would have a base with the databound width and the top "tip" of the triangle will be at the mid-point of the databound width and at Y=0. That was just an optimization for simplicity's sake
The behavior class:
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Windows;
using System.Windows.Shapes;
using System.Windows.Media;
namespace WpfApplication1
{
public enum ShapeType
{
Rectangle,
Isosceles,
Ellipse,
Dice,
Hexagon
}
public class PathControl
{
public static readonly DependencyProperty ShapeTypeProperty =
DependencyProperty.RegisterAttached("ShapeType",
typeof(ShapeType?),
typeof(DependencyObject),
new PropertyMetadata(null,
new PropertyChangedCallback((sender, args) =>
{
Path path = sender as Path;
ShapeType? shapeType = (ShapeType?)args.NewValue;
//todo: use a WeakEvent
path.SizeChanged +=
(pathSender, pathArgs) =>
{
PathControl.InvalidatePath((Path)sender, shapeType);
};
})));
private static void InvalidatePath(Path path, ShapeType? shapeType)
{
if (path != null
&& shapeType.HasValue)
{
string source = null;
double netWidth = path.Width - 2 * path.StrokeThickness,
netHeight = path.Height - 2 * path.StrokeThickness;
if (shapeType == ShapeType.Rectangle)
{
source = string.Format("M0,0 h{0} v{1} h-{0} z",
new object[2]
{
netWidth,
netHeight
});
}
else if (shapeType == ShapeType.Isosceles)
{
source = string.Format("M0,{1} l{0},-{1} {0},{1} z",
new object[2]
{
netWidth / 2,
netHeight
});
}
else
{
throw new NotImplementedException(shapeType.ToString());
}
path.Data = Geometry.Parse(source);
}
}
public static void SetShapeType(DependencyObject o, ShapeType e)
{
o.SetValue(PathControl.ShapeTypeProperty, e);
}
public static ShapeType? GetShapeType(DependencyObject o)
{
return (ShapeType)o.GetValue(PathControl.ShapeTypeProperty);
}
}
}
The XAML:
<Window x:Class="WpfApplication1.MainWindow"
xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
Title="MainWindow" Height="350" Width="525"
xmlns:local="clr-namespace:WpfApplication1">
<Grid>
<Path Width="100" Height="100" Stroke="Green" StrokeThickness="2" Fill="Yellow"
local:PathControl.ShapeType="Isosceles">
<Path.RenderTransform>
<RotateTransform Angle="90"></RotateTransform>
</Path.RenderTransform>
</Path>
</Grid>
</Window>
Binding to the Points is the best/only way. The X and X properties of a Point cannot be bound to because they do not raise the PropertyChanged event. The Point is a structure and structures should be read-only.
The PointCollection class raises the correct events so you can bind to it. This allows you to manipulate the triangles by modifying the collection of point by replacing the points. Do not change the point but replace them so the proper events will be raised.

Simple WPF UI databinding

I'm trying to write a simple WPF app that has two ellipses, joined by a line, like you might see in a network graph. When the ellipses are animated, I just want the joining line to automagically 'stick' to the canvas locations of the two ellipses that the line joins. The XAML is just a canvas:
<Window x:Class="UIDataBindingDemo.Window1"
xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
Title="Window1" Height="400" Width="400">
<Grid>
<Canvas x:Name="cnvExample" />
</Grid>
...and I'm just doing some really simple stuff in the constructor here:
using System;
using System.Windows;
using System.Windows.Controls;
using System.Windows.Data;
using System.Windows.Media;
using System.Windows.Media.Animation;
using System.Windows.Shapes;
namespace UIDataBindingDemo
{
/// <summary>
/// Interaction logic for Window1.xaml
/// </summary>
public partial class Window1 : Window
{
public Window1()
{
InitializeComponent();
// create 2 ellipses, one next to the other, and add them to the canvas
Ellipse el1 = new Ellipse();
Canvas.SetTop(el1, 100);
Canvas.SetLeft(el1, 100);
el1.Width = 20;
el1.Height = 20;
el1.Fill = Brushes.Red;
el1.Stroke = Brushes.Black;
Ellipse el2 = new Ellipse();
Canvas.SetTop(el2, 100);
Canvas.SetLeft(el2, 200);
el2.Width = 20;
el2.Height = 20;
el2.Fill = Brushes.Blue;
el2.Stroke = Brushes.Black;
cnvExample.Children.Add(el1);
cnvExample.Children.Add(el2);
// create a line that connects the 2 ellipses. Bind the two points that define this line to the
// locations of our ellipses, so the line always connects them, through animations, drag and drop
// operations, whatever.
Line line = new Line();
line.StrokeThickness = 3;
line.Stroke = Brushes.Black;
line.SetBinding(Line.X1Property, new Binding("(Canvas.Left)") { Source = el1 });
line.SetBinding(Line.X1Property, new Binding("(Canvas.Top)") { Source = el1 });
line.SetBinding(Line.X1Property, new Binding("(Canvas.Left)") { Source = el2 });
line.SetBinding(Line.X1Property, new Binding("(Canvas.Top)") { Source = el2 });
cnvExample.Children.Add(line);
// animate the second ellipse, so it moves down and to the right, nice and slow
var moveTheBlueOne = new DoubleAnimation(300, TimeSpan.FromSeconds(10));
el2.BeginAnimation(Canvas.LeftProperty, moveTheBlueOne);
el2.BeginAnimation(Canvas.TopProperty, moveTheBlueOne);
}
}
I'm pretty new to WPF, and I'm sure I'm missing something simple. Why am I not seeing the line?
I don't know if it's a cut and paste error but youre assigning each binding to the same DependencyProperty "Line.X1Property", you should use all four X and Y properties to define a starting point and an ending one for a line.
line.SetBinding(Line.X1Property, new Binding("(Canvas.Left)") { Source = el1 });
line.SetBinding(Line.Y1Property, new Binding("(Canvas.Top)") { Source = el1 });
line.SetBinding(Line.X2Property, new Binding("(Canvas.Left)") { Source = el2 });
line.SetBinding(Line.Y2Property, new Binding("(Canvas.Top)") { Source = el2 });
this way it works for me.

Resources