Thalia
Thalia

Reputation: 14615

c++: Convert 24bpp to 8 bpp or 1bpp image

I have to convert a 24bpp image to a 1bpp image or 8bpp image based on color table. The caller expects a unsigned char* in either case (which would be further processed or maybe for now debug output by sending the BITMAPINFOHEADER.biBitCount to its proper value, 8 or 1).

I have code to extract the color index into the palette (colorIndexArray is from color conversion or dithering algorithms)... I can get the info for an 8bpp bitmap...

But my problem is, I don't know how to put this info into a 1bpp bitmap

typedef struct {
    unsigned int size;
    unsigned char* pixels;
} ColorIndexArray;

unsigned char* convertImage(const ColorIndexArray& colorIndexArray, unsigned int paletteSize)
{
    unsigned char* outputImage;
    if (paleteSize > 2)
    {
        outputImage = (unsigned char*)LocalAlloc(LPTR, colorIndexArray.size);
        for (int i=0; i<colorIndexArray.size; i++)
            *(outputImage+i) = colorIndexArray.pixels[i];  
        // this works great              
    }
    else  // monochrome, caller has palette colors likely b/w (or purple/magenta or anything), must be 1bpp
    {
        outputImage = (unsigned char*)LocalAlloc(LPTR, colorIndexArray.size / 8);
        // how can i place the unsigned char* info (which is already 
        // determined based on desired algorithm, representing index in 
        // color table) into the output image inside a single bit ?
        // (obviously its value for a monochrome image would be 0 or 1 but    
        // it is saved as unsigned char* at the algorithm output) 
        // And how do I advance the pointer ?
        // Will it be type safe ? Aligned to byte ? or do I have to fill 
        // with something at the end to make multiple of 8 bits ?
    }

    return outputImage;
}

Trying this after comment suggestion:

#include <GdiPlus.h>
....
else {
    Gdiplus::Bitmap monoBitmap(w, h, PixelFormat1bppIndexed);
    Gdiplus::BitmapData monoBitmapData;
    Gdiplus::Rect rect(0, 0, w, h);
    monoBitmap.LockBits(&rect, Gdiplus::ImageLockModeWrite, PixelFormat1bppIndexed, &monoBitmapData);
    outputImage = (unsigned char*)monoBitmapData.Scan0;

    for (unsigned int y = 0; y < h; y++)
    {
        for (unsigned int x = 0; x < w; x++)
        {
            if (colorIndexArray.pixels[x + y * w])
                outputImage[y*monoBitmapData.Stride + x / 8] |= (unsigned char)(0x80 >> (x % 8));
        }           
    }
    monoBitmap.UnlockBits(&monoBitmapData); 
}
return outputImage;

(Also need to allocate the memory for outputImage)

Upvotes: 1

Views: 1742

Answers (1)

Thalia
Thalia

Reputation: 14615

Based on the example suggested by Hans Passant (thank you also for pointing out how important the stride is), I wrote this little conversion

unsigned long stride = (((w + 31) & ~31) >> 3); 

outputImage = (unsigned char*)LocalAlloc(LPTR, stride * h);

for (unsigned int y = 0; y < h; y++)
{           
    unsigned char* b = (unsigned char*)LocalAlloc(LPTR, stride);
    for (unsigned int x = 0; x < w; x++)
        if (colorIndexArray.pixels[x + y * w])
            b[x / 8] |= (unsigned char)(0x80 >> (x % 8));               
    CopyMemory(outputImage + stride * y, b, stride);
}

Upvotes: 1

Related Questions