CGImage from byte array

Question

Loading a CGImage or NSImage from a file using a standard image format (jpeg, gif, png et.) is all very simple.

However, I now need to create a CGImage from an array in bytes in memory generated using libfreetype. Its really easy to create OpenGL textures from an array of formatted bytes, and I can see how to create a CGBitmapContext to write to. But I can't seem to find an easy way to create a CGImage from a raw pixel array.

Answer 1

You can create a CGDataProvider, and let CG request the necessary data from your provider, instead of writing to an image buffer.

Here's a very simple example that generates a black CGImage of size 64x64.

CGDataProviderSequentialCallbacks callbacks;
callbacks.getBytes = getBytes;

CGDataProviderRef provider = CGDataProviderCreateSequential(NULL, &callbacks);
CGColorSpaceRef space = CGColorSpaceCreateDeviceRGB();
CGImageRef img = CGImageCreate(64,                         // width
                               64,                         // height
                               8,                          // bitsPerComponent
                               24,                         // bitsPerPixel
                               64*3,                       // bytesPerRow
                               space,                      // colorspace
                               kCGBitmapByteOrderDefault,  // bitmapInfo
                               provider,                   // CGDataProvider
                               NULL,                       // decode array
                               NO,                         // shouldInterpolate
                               kCGRenderingIntentDefault); // intent

CGColorSpaceRelease(space);
CGDataProviderRelease(provider);

// use the created CGImage

CGImageRelease(img);

and getBytes is defined like this:

size_t getBytes(void *info, void *buffer, size_t count) {
    memset(buffer, 0x00, count);
    return count;
}

of course, you will want to implement the other callbacks (skipForward, rewind, releaseInfo), and use a proper structure or object for info.

For more information, check out the CGImage and CGDataProvider references.

Answer 2

Using CFData can make the above code simpler. Here is the code.

// raw pixel data memory of 64 * 64 pixel size

UInt8 pixelData[64 * 64 * 3];

// fill the raw pixel buffer with arbitrary gray color for test

for(size_t ui = 0; ui < 64 * 64 * 3; ui++) pixelData[ui] = 210;

CGColorSpaceRef colorspace = CGColorSpaceCreateDeviceRGB();

CFDataRef rgbData = CFDataCreate(NULL, pixelData, 64 * 64 * 3);

CGDataProviderRef provider = CGDataProviderCreateWithCFData(rgbData);

CGImageRef rgbImageRef = CGImageCreate(64, 64, 8, 24, 64 * 3, colorspace, kCGBitmapByteOrderDefault, provider, NULL, true, kCGRenderingIntentDefault);

CFRelease(rgbData);

CGDataProviderRelease(provider);

CGColorSpaceRelease(colorspace);

// use the created CGImage

CGImageRelease(rgbImageRef);