After searching for days, it suddenly hit me that the answer had been staring me in the face all the time! I was creating a GDI+ bitmap from a pointer to a byte array. Then trying to create an HBITMAP using the same pointer. But I could just as easily create the HBITMAP first and use the pointer from that to create the GDI+ bitmap.
It works like a charm! You can mix GDI and GDI+ operations however you like. The image is both plain GDI and GDI+ at once. Instead of using DrawImage, you can BitBlt from the exact same pixel data!
Here's the code:
// Create the HBITMAP
BITMAPINFO binfo = new BITMAPINFO();
binfo.biSize = (uint)Marshal.SizeOf(typeof(BITMAPINFO));
binfo.biWidth = width;
binfo.biHeight = height;
binfo.biBitCount = (ushort)Image.GetPixelFormatSize(pixelFormat);
binfo.biPlanes = 1;
binfo.biCompression = 0;
hDC = CreateCompatibleDC(IntPtr.Zero);
IntPtr pointer;
hBitmap = CreateDIBSection(hDC, ref binfo, 0, out pointer, IntPtr.Zero, 0);
// Create the GDI+ bitmap using the pointer returned from CreateDIBSection
gdiBitmap = new Bitmap(width, height, width * binfo.biBitCount >> 3, pixelFormat, pointer);
Bitmap
object, shouldn't you just useGraphics.DrawImage
, where theGraphics
object wraps the intended target DC? Why go through anHBitmap
at all? – Photothermic