Converting CVImageBufferRef to PNG

Question

Hi, I want to write all frames to disk using QTKit.framework for camera input. But all the images I get are half transparent and without some colors, maybe colorspace problem? ;( They seem good in the preview View but when I write them "something" happens. I wonder what it that.

-(void)savePNGImage:(CGImageRef)imageRef path:(NSString *)path {
    NSURL *outURL = [[NSURL alloc] initFileURLWithPath:path]; 
    CGImageDestinationRef dr = CGImageDestinationCreateWithURL ((CFURLRef)outURL, (CFStringRef)@"public.png" , 1, NULL);
    CGImageDestinationAddImage(dr, imageRef, NULL);
    CGImageDestinationFinalize(dr);
}
-(void)saveJPEGImage:(CGImageRef)imageRef path:(NSString *)path {
    CFMutableDictionaryRef mSaveMetaAndOpts = CFDictionaryCreateMutable(nil, 0,
                                                                    &kCFTypeDictionaryKeyCallBacks,  &kCFTypeDictionaryValueCallBacks);
    CFDictionarySetValue(mSaveMetaAndOpts, kCGImageDestinationLossyCompressionQuality, 
                         [NSNumber numberWithFloat:1.0]);   // set the compression quality here
    NSURL *outURL = [[NSURL alloc] initFileURLWithPath:path];
    CGImageDestinationRef dr = CGImageDestinationCreateWithURL ((CFURLRef)outURL, (CFStringRef)@"public.jpeg" , 1, NULL);
    CGImageDestinationAddImage(dr, imageRef, mSaveMetaAndOpts);
    CGImageDestinationFinalize(dr);
}


- (void)captureOutput:(QTCaptureOutput *)captureOutput 
  didOutputVideoFrame:(CVImageBufferRef)videoFrame 
     withSampleBuffer:(QTSampleBuffer *)sampleBuffer 
       fromConnection:(QTCaptureConnection *)connection{

    CVPixelBufferLockBaseAddress(videoFrame,0); 
    uint8_t *baseAddress = (uint8_t *)CVPixelBufferGetBaseAddress(videoFrame); 
    size_t bytesPerRow = CVPixelBufferGetBytesPerRow(videoFrame); 
    size_t width = CVPixelBufferGetWidth(videoFrame); 
    size_t height = CVPixelBufferGetHeight(videoFrame); 
    CVPixelBufferUnlockBaseAddress(videoFrame,0);

    CGColorSpaceRef colorSpace = CGColorSpaceCreateDeviceRGB(); 
    CGContextRef newContext = CGBitmapContextCreate(baseAddress, 
                                                        width, height, 8, 
                                                        bytesPerRow, 
                                                        colorSpace, kCGBitmapByteOrder32Little | kCGImageAlphaPremultipliedFirst); 
        CGImageRef frame = CGBitmapContextCreateImage(newContext); 

    CGContextRelease(newContext); 
    CGColorSpaceRelease(colorSpace);

    [self savePNGImage:frame path:[NSString stringWithFormat:@"/Users/nacho4d/Desktop/framesCam/frame%003d.png", frameNum++]];
[self saveJPEGImage:frame path:[NSString stringWithFormat:@"/Users/nacho4d/Desktop/framesCam/frame%003d.jpeg", frameNum++]];

    CGImageRelease(frame);
}

Capturer attributes are just frame size and pixel format like so: [NSNumber numberWithUnsignedInt:kCVPixelFormatType_32ARGB], (id)kCVPixelBufferPixelFormatTypeKey

Update:

I have tried with JPEG also and I get the same kind of image, like there are some channels lacking in the image and all written frames have a white background. (since JPEG does not allow transparence ??)

The original and the written (JPEG) one:

(I don't show the PNG since is transparent and is difficult to see in a browser)

Thanks in advance.