Im using glReadPixels to grab screen shots of my opengl scene and then turning them into a video using AVAssetWriter on IOS 4. My problem is i need to pass the alpha channel to the video which only accepts kCVPixelFormatType_32ARGB and glReadPixels on retrieves RGBA. So basically i need a way to convert my RGBA to ARGB, in other words put the alpha bytes first.
int depth = 4;
unsigned char buffer[width * height * depth];
glReadPixels(0,0,width, height, GL_RGBA, GL_UNSIGNED_BYTE, &buffer);
CGDataProviderRef ref = CGDataProviderCreateWithData(NULL, &buffer), width*height*depth, NULL );
CGBitmapInfo bitmapInfo = kCGBitmapByteOrderDefault | kCGImageAlphaPremultipliedLast;
CGImageRef image = CGImageCreate(width, height, 8, 32, width*depth, CGColorSpaceCreateDeviceRGB(), bitmapInfo, ref, NULL, true, kCGRenderingIntentDefault);
UIWindow* parentWindow = [self window];
NSDictionary *options = [NSDictionary dictionaryWithObjectsAndKeys:[NSNumber numberWithBool:YES], kCVPixelBufferCGImageCompatibilityKey, [NSNumber numberWithBool:YES], kCVPixelBufferCGBitmapContextCompatibilityKey, nil];
CVPixelBufferRef pxbuffer = NULL;
CVReturn status = CVPixelBufferCreate(kCFAllocatorDefault, width, height, kCVPixelFormatType_32ARGB, (CFDictionaryRef) options, &pxbuffer);
NSParameterAssert(status == kCVReturnSuccess);
NSParameterAssert(pxbuffer != NULL);
CVPixelBufferLockBaseAddress(pxbuffer, 0);
void *pxdata = CVPixelBufferGetBaseAddress(pxbuffer);
NSParameterAssert(pxdata != NULL);
CGColorSpaceRef rgbColorSpace = CGColorSpaceCreateDeviceRGB();
CGContextRef context = CGBitmapContextCreate(pxdata, width, height, 8, depth*width, rgbColorSpace, kCGImageAlphaPremultipliedFirst);
NSParameterAssert(context);
CGContextConcatCTM(context, parentWindow.transform);
CGContextDrawImage(context, CGRectMake(0, 0, width, height), image);
CGColorSpaceRelease(rgbColorSpace);
CGContextRelease(context);
CVPixelBufferUnlockBaseAddress(pxbuffer, 0);
return pxbuffer; // chuck pixel buffer into AVAssetWriter
Thought i would post the whole code as i may help someone else.
Cheers
Note: I'm assuming 8 bits per channel. Adjust accordingly if this is not the case.
To move the alpha bits last, you need to perform rotation. This is usually expressed most easily through bit shifting.
In this case, you want to move the RGB bits 8 bits right, and the A bits 24 bits left. These two values should then be put together using bitwise OR, so that becomes argb = (rgba >> 8) | (rgba << 24)
.