Gdiplus::Bitmap to BYTE array?

Question

Here's my attempt (ugly GDI+ and GDI mix...)

// ...
BYTE               pixels[BMP_WIDTH * BMP_HEIGHT * BMP_BPP];
HBITMAP            hBitmap;
Gdiplus::Bitmap    cBitmap(BMP_WIDTH, BMP_HEIGHT, PixelFormat32bppRGB);
Gdiplus::Graphics  cGraphics(&cBitmap);
Gdiplus::Pen       cPen(Gdiplus::Color(255, 255, 0, 0));

cGraphics.DrawRectangle(&cPen, 0, 0, cBitmap.GetWidth() - 1, cBitmap.GetHeight() - 1);

// and here it get's real ugly, I'd like to change that...
cBitmap.GetHBITMAP(Gdiplus::Color(255, 255, 255), &hBitmap);
GetBitmapBits(hBitmap, sizeof(pixels), pixels);
// ...

Someone told me to use LockBits but I really didn't understand how. I tried it, but I failed so I'm not going to post that attempt, too.

Solution 1

You could use Bitmap::LockBits to get access to a raw array of data. Here you could read about how to use Bitmap::LockBits.

Solution 2

Here's something I wrote that returns a vector of vectors (with the contained vectors representing columns of pixels in the image) when passed a file path:

#include <vector>
std::vector<std::vector<unsigned>> getPixels(const wchar_t *filename, int &width, int &height) {
    Gdiplus::Bitmap bitmap(filename);

    //Pass up the width and height, as these are useful for accessing pixels in the vector o' vectors.
    width = bitmap.GetWidth();
    height = bitmap.GetHeight();

    auto *bitmapData = new Gdiplus::BitmapData;

    //Lock the whole bitmap so we can read pixel data easily.
    Gdiplus::Rect rect(0, 0, width, height);
    bitmap.LockBits(&rect, Gdiplus::ImageLockModeRead, PixelFormat32bppARGB, bitmapData);

    //Get the individual pixels from the locked area.
    auto *pixels = static_cast<unsigned *>(bitmapData->Scan0);

    //Vector of vectors; each vector is a column.
    std::vector<std::vector<unsigned>> resultPixels(width, std::vector<unsigned>(height));

    const int stride = abs(bitmapData->Stride);
    for(int x = 0; x < width; x++) {
        for(int y = 0; y < height; y++) {
            //Get the pixel colour from the pixels array which we got earlier.
            const unsigned pxColor = pixels[y * stride / 4 + x];

            //Get each individual colour component. Bitmap colours are in reverse order.
            const unsigned red = (pxColor & 0xFF0000) >> 16;
            const unsigned green = (pxColor & 0xFF00) >> 8;
            const unsigned blue = pxColor & 0xFF;

            //Combine the values in a more typical RGB format (as opposed to the bitmap way).
            const int rgbValue = RGB(red, green, blue);

            //Assign this RGB value to the pixel location in the vector o' vectors.
            resultPixels[x][y] = rgbValue;
        }
    }

    //Unlock the bits that we locked before.
    bitmap.UnlockBits(bitmapData);
    return resultPixels;
}

Solution 3

Here is how I would do it using GDIPlus Bitmap.LockBits method defined in the header GdiPlusBitmap.h:

Notice that since bitmaps are usually DWORD aligned you may want to discard this unused data that was needed for the alignment, as malat correctly commented..

    Gdiplus::BitmapData bitmapData;
    Gdiplus::Rect rect(0, 0, bitmap.GetWidth(), bitmap.GetHeight());

    //get the bitmap data
    if(Gdiplus::Ok == bitmap.LockBits(
                        &rect, //A rectangle structure that specifies the portion of the Bitmap to lock.
                        Gdiplus::ImageLockModeRead | Gdiplus::ImageLockModeWrite, //ImageLockMode values that specifies the access level (read/write) for the Bitmap.
                        bitmap.GetPixelFormat(),// PixelFormat values that specifies the data format of the Bitmap.
                        &bitmapData //BitmapData that will contain the information about the lock operation.
                        ))
    {
         //get the lenght of the bitmap data in bytes
         int len = bitmapData.Height * std::abs(bitmapData.Stride);

         BYTE* buffer = new BYTE[len];
         memcpy(bitmapData.Scan0, buffer, len);//copy it to an array of BYTEs

         //... 

         //cleanup
         bitmap.UnlockBits(&bitmapData);        
         delete []buffer;
    }

Solution 4

Have you tried supplying the bytes when you create the bitmap:

int width = BMP_WIDTH;
int height = BMP_HEIGHT;
int stride = 4 * width;
BYTE bytes[stride * height];

Gdiplus::Bitmap  cBitmap(width, height, stride, PixelFormat32bppRGB, bytes);

View more solutions

Comments

[malat]

No this will not work, bitmapData.Stride is different from bitmapData.Width * sizeof( bitmapData.PixelFormat), since it contains also unused data. This requires a for loop to discard unused data.

[malat]

This is an easy case, one should pay attention that stride % 4 == 0 in the general case (not all 3 components RGB works).

[Gray Programmerz]

code crashes at memcpy(...)

[Gray Programmerz]

how do that, when BITMAP is created using BITMAP::FromHBITMAP(...) ?