I want to use partially transparent images in drag/drop operations. This is all set up and works fine, but the actual transformation to transparency has a weird side effect. For some reason, the pixels seem to be blended against a black background.
The following image describes the problem:
Figure a) is the original bitmap.
Figure b) is what is produced after alpha blending has been performed. Obviously this is a lot darker than the intended 50% alpha filter intended.
Figure c) is the desired effect, image a) with 50% transparency (added to the composition with a drawing program).
The code I use to produce the trasparent image is the following:
Bitmap bmpNew = new Bitmap(bmpOriginal.Width, bmpOriginal.Height);
Graphics g = Graphics.FromImage(bmpNew);
// Making the bitmap 50% transparent:
float[][] ptsArray ={
new float[] {1, 0, 0, 0, 0}, // Red
new float[] {0, 1, 0, 0, 0}, // Green
new float[] {0, 0, 1, 0, 0}, // Blue
new float[] {0, 0, 0, 0.5f, 0}, // Alpha
new float[] {0, 0, 0, 0, 1} // Brightness
};
ColorMatrix clrMatrix = new ColorMatrix(ptsArray);
ImageAttributes imgAttributes = new ImageAttributes();
imgAttributes.SetColorMatrix(clrMatrix, ColorMatrixFlag.Default, ColorAdjustType.Bitmap);
g.DrawImage(bmpOriginal, new Rectangle(0, 0, bmpOriginal.Width, bmpOriginal.Height), 0, 0, bmpOriginal.Width, bmpOriginal.Height, GraphicsUnit.Pixel, imgAttributes);
Cursors.Default.Draw(g, new Rectangle(bmpOriginal.Width / 2 - 8, bmpOriginal.Height / 2 - 8, 32, 32));
g.Dispose();
imgAttributes.Dispose();
return bmpNew;
Does anyone know why the alpha blending does not work?
Update I:
For clarity, the code does work if I'm alphablending on top of a drawn surface. The problem is that I want to create a completely semitransparent image from an existing image and use this as a dynamic cursor during drag/drop operations. Even skipping the above and only painting a filled rectangle of color 88ffffff yields a dark grey color. Something fishy is going on with the icon.
Update II:
Since I've reseached a whole lot and believe this has got something to do with the Cursor creation, I'm gonna include that code below too. If I GetPixel-sample the bitmap just before the CreateIconIndirect call, the four color values seem to be intact. Thus I have a feeling the culprits might be the hbmColor or the hbmMask members of the IconInfo structure.
Here's the IconInfo structure:
public struct IconInfo { // http://msdn.microsoft.com/en-us/library/ms648052(VS.85).aspx
public bool fIcon; // Icon or cursor. True = Icon, False = Cursor
public int xHotspot;
public int yHotspot;
public IntPtr hbmMask; // Specifies the icon bitmask bitmap. If this structure defines a black and white icon,
// this bitmask is formatted so that the upper half is the icon AND bitmask and the lower
// half is the icon XOR bitmask. Under this condition, the height should be an even multiple of two.
// If this structure defines a color icon, this mask only defines the AND bitmask of the icon.
public IntPtr hbmColor; // Handle to the icon color bitmap. This member can be optional if this structure defines a black
// and white icon. The AND bitmask of hbmMask is applied with the SRCAND flag to the destination;
// subsequently, the color bitmap is applied (using XOR) to the destination by using the SRCINVERT flag.
}
And here is the code that actually creates the Cursor:
public static Cursor CreateCursor(Bitmap bmp, int xHotSpot, int yHotSpot) {
IconInfo iconInfo = new IconInfo();
GetIconInfo(bmp.GetHicon(), ref iconInfo);
iconInfo.hbmColor = (IntPtr)0;
iconInfo.hbmMask = bmp.GetHbitmap();
iconInfo.xHotspot = xHotSpot;
iconInfo.yHotspot = yHotSpot;
iconInfo.fIcon = false;
return new Cursor(CreateIconIndirect(ref iconInfo));
}
The two external functions are defined as follows:
[DllImport("user32.dll", EntryPoint = "CreateIconIndirect")]
public static extern IntPtr CreateIconIndirect(ref IconInfo icon);
[DllImport("user32.dll")]
[return: MarshalAs(UnmanagedType.Bool)]
public static extern bool GetIconInfo(IntPtr hIcon, ref IconInfo pIconInfo);
GDI+ has a number of problems related to alpha blending when doing interop with GDI (and Win32). In this case, the call to bmp.GetHbitmap() will blend your image with a black background. An article on CodeProject gives more detail on the problem, and a solution that was used for adding images to an image list.
You should be able to use similar code to get the HBITMAP to use for the mask:
[DllImport("kernel32.dll")]
public static extern bool RtlMoveMemory(IntPtr dest, IntPtr source, int dwcount);
[DllImport("gdi32.dll")]
public static extern IntPtr CreateDIBSection(IntPtr hdc, [In, MarshalAs(UnmanagedType.LPStruct)]BITMAPINFO pbmi, uint iUsage, out IntPtr ppvBits, IntPtr hSection, uint dwOffset);
public static IntPtr GetBlendedHBitmap(Bitmap bitmap)
{
BITMAPINFO bitmapInfo = new BITMAPINFO();
bitmapInfo.biSize = 40;
bitmapInfo.biBitCount = 32;
bitmapInfo.biPlanes = 1;
bitmapInfo.biWidth = bitmap.Width;
bitmapInfo.biHeight = -bitmap.Height;
IntPtr pixelData;
IntPtr hBitmap = CreateDIBSection(
IntPtr.Zero, bitmapInfo, 0, out pixelData, IntPtr.Zero, 0);
Rectangle bounds = new Rectangle(0, 0, bitmap.Width, bitmap.Height);
BitmapData bitmapData = bitmap.LockBits(
bounds, ImageLockMode.ReadOnly, PixelFormat.Format32bppArgb );
RtlMoveMemory(
pixelData, bitmapData.Scan0, bitmap.Height * bitmapData.Stride);
bitmap.UnlockBits(bitmapData);
return hBitmap;
}
A while ago, I read this problem arises out of a requirement for pre-multiplied alpha channels in the bitmaps. I'm not sure if this was an issue with Windows cursors or GDI, and for the life of me, I cannot find documentation regarding this. So, while this explanation may or may not be correct, the following code does indeed do what you want, using a pre-multiplied alpha channel in the cursor bitmap.
public class CustomCursor
{
// alphaLevel is a value between 0 and 255. For 50% transparency, use 128.
public Cursor CreateCursorFromBitmap(Bitmap bitmap, byte alphaLevel, Point hotSpot)
{
Bitmap cursorBitmap = null;
External.ICONINFO iconInfo = new External.ICONINFO();
Rectangle rectangle = new Rectangle(0, 0, bitmap.Width, bitmap.Height);
try
{
// Here, the premultiplied alpha channel is specified
cursorBitmap = new Bitmap(bitmap.Width, bitmap.Height, PixelFormat.Format32bppPArgb);
// I'm assuming the source bitmap can be locked in a 24 bits per pixel format
BitmapData bitmapData = bitmap.LockBits(rectangle, ImageLockMode.ReadOnly, PixelFormat.Format24bppRgb);
BitmapData cursorBitmapData = cursorBitmap.LockBits(rectangle, ImageLockMode.WriteOnly, cursorBitmap.PixelFormat);
// Use either SafeCopy() or UnsafeCopy() to set the bitmap contents
SafeCopy(bitmapData, cursorBitmapData, alphaLevel);
//UnsafeCopy(bitmapData, cursorBitmapData, alphaLevel);
cursorBitmap.UnlockBits(cursorBitmapData);
bitmap.UnlockBits(bitmapData);
if (!External.GetIconInfo(cursorBitmap.GetHicon(), out iconInfo))
throw new Exception("GetIconInfo() failed.");
iconInfo.xHotspot = hotSpot.X;
iconInfo.yHotspot = hotSpot.Y;
iconInfo.IsIcon = false;
IntPtr cursorPtr = External.CreateIconIndirect(ref iconInfo);
if (cursorPtr == IntPtr.Zero)
throw new Exception("CreateIconIndirect() failed.");
return (new Cursor(cursorPtr));
}
finally
{
if (cursorBitmap != null)
cursorBitmap.Dispose();
if (iconInfo.ColorBitmap != IntPtr.Zero)
External.DeleteObject(iconInfo.ColorBitmap);
if (iconInfo.MaskBitmap != IntPtr.Zero)
External.DeleteObject(iconInfo.MaskBitmap);
}
}
private void SafeCopy(BitmapData srcData, BitmapData dstData, byte alphaLevel)
{
for (int y = 0; y < srcData.Height; y++)
for (int x = 0; x < srcData.Width; x++)
{
byte b = Marshal.ReadByte(srcData.Scan0, y * srcData.Stride + x * 3);
byte g = Marshal.ReadByte(srcData.Scan0, y * srcData.Stride + x * 3 + 1);
byte r = Marshal.ReadByte(srcData.Scan0, y * srcData.Stride + x * 3 + 2);
Marshal.WriteByte(dstData.Scan0, y * dstData.Stride + x * 4, b);
Marshal.WriteByte(dstData.Scan0, y * dstData.Stride + x * 4 + 1, g);
Marshal.WriteByte(dstData.Scan0, y * dstData.Stride + x * 4 + 2, r);
Marshal.WriteByte(dstData.Scan0, y * dstData.Stride + x * 4 + 3, alphaLevel);
}
}
private unsafe void UnsafeCopy(BitmapData srcData, BitmapData dstData, byte alphaLevel)
{
for (int y = 0; y < srcData.Height; y++)
{
byte* srcRow = (byte*)srcData.Scan0 + (y * srcData.Stride);
byte* dstRow = (byte*)dstData.Scan0 + (y * dstData.Stride);
for (int x = 0; x < srcData.Width; x++)
{
dstRow[x * 4] = srcRow[x * 3];
dstRow[x * 4 + 1] = srcRow[x * 3 + 1];
dstRow[x * 4 + 2] = srcRow[x * 3 + 2];
dstRow[x * 4 + 3] = alphaLevel;
}
}
}
}
The pinvoke declarations are found in the External class, shown here:
public class External
{
[StructLayout(LayoutKind.Sequential)]
public struct ICONINFO
{
public bool IsIcon;
public int xHotspot;
public int yHotspot;
public IntPtr MaskBitmap;
public IntPtr ColorBitmap;
};
[DllImport("user32.dll")]
public static extern bool GetIconInfo(IntPtr hIcon, out ICONINFO piconinfo);
[DllImport("user32.dll")]
public static extern IntPtr CreateIconIndirect([In] ref ICONINFO piconinfo);
[DllImport("gdi32.dll")]
public static extern bool DeleteObject(IntPtr hObject);
[DllImport("gdi32.dll")]
public static extern IntPtr CreateBitmap(int nWidth, int nHeight, uint cPlanes, uint cBitsPerPel, IntPtr lpvBits);
}
A few notes on the code:
- To use the unsafe method, UnsafeCopy(), you must compile with the /unsafe flag.
- The bitmap copying methods are ugly, especially the safe method, which uses Marshal.ReadByte()/Marshal.WriteByte() calls. There must be a faster way to copy bitmap data while also inserting alpha bytes.
- I do assume that the source bitmap is able to be locked in a 24 bits per pixel format. This should not be a problem, though.
try lowering Blue's value to .7 or .6 and see if that is closer to what you want.
Here's a good site that explains ColorMatrix:
When I run your code to modify an image in a picturebox with a background grid image, I get the effect you desired without changing your code. Perhaps your image is being drawn over the top of something that has a dark color...
Forgive me if my suggestion is too simplistic (I'm still new to C#) but I found this on the MSDN site and maybe this might point you in the right direction?
/matt