Can CGContextClipToMask mask all non-transparent pixels with alpha=1? - ios

Question

Is it possible to make CGContextClipToMask ignore the grayscale values of the mask image and work as if it was plain black and white?
I have a grayscale image, and when I use it as a mask gray color are interpreted as an alpha channel. This is fine except for a point where I need to completely mask those pixels that are not transparent.
Short example:
UIImage *mask = [self prepareMaskImage];
UIGraphicsBeginImageContextWithOptions(mask.size, NO, mask.scale); {
// Custom code
CGContextClipToMask(UIGraphicsGetCurrentContext(), mask.size, mask.CGImage);
// Custom code
}
Is it possible to adapt this code to achieve my goal?
Long story short: I need to make a transparent grayscale image become transparent where it originally was and completely black where it's solid-colored.

Answer 1

Interesting problem! Here's code that does what I think you want in a simple sample project. Similar to above but handles scale properly. Also has option to retain the alpha in the mask image if you want. Quick hacked together test that seems to work.

Answer 2

My rough idea would be the following:
You convert your input image to readable byte data in grey + alpha format. Maybe you have to do RGBA instead due to iOS limitations.
You iterate over the byte data modifying the values in place.
To simplify access, use a
typedef struct RGBA {
UInt8 red;
UInt8 green;
UInt8 blue;
UInt8 alpha;
} RGBA;
Let's assume image your input mask.
// First step, using RGBA (because I know it works and does not harm, just writes/consumes twice the amount of memory)
CGImageRef imageRef = image.CGImage;
NSInteger rawWidth = CGImageGetWidth(imageRef);
NSInteger rawHeight = CGImageGetHeight(imageRef);
NSInteger rawBitsPerComponent = 8;
NSInteger rawBytesPerPixel = 4;
NSInteger rawBytesPerRow = rawBytesPerPixel * rawWidth;
CGRect rawRect = CGRectMake(0, 0, rawWidth, rawHeight);
CGColorSpaceRef colorSpace = CGColorSpaceCreateDeviceRGB();
UInt8 *rawImage = (UInt8 *)malloc(rawHeight * rawWidth * rawBytesPerPixel);
CGContextRef rawContext = CGBitmapContextCreate(rawImage,
rawWidth,
rawHeight,
rawBitsPerComponent,
rawBytesPerRow,
colorSpace,
kCGImageAlphaPremultipliedLast);
CGColorSpaceRelease(colorSpace);
// At this point, rawContext is ready for drawing, everything drawn will be in rawImage's byte array.
CGContextDrawImage(rawContext, rawRect, imageRef);
// Second step, crawl the byte array and do the evil work:
for (NSInteger y = 0; y < rawHeight; ++y) {
for (NSInteger x = 0; x < rawWidth; ++x) {
UInt8 *address = rawImage + x * rawBytesPerPixel + y * rawBytesPerRow;
RGBA *pixel = (RGBA *)address;
// If it is a grey input image, it does not matter what RGB channel to use - they shall all be the same
if (0 != pixel->red) {
pixel->alpha = 0;
} else {
pixel->alpha = UINT8_MAX;
}
pixel->red = 0;
pixel->green = 0;
pixel->blue = 0;
// I am still not sure if this is the transformation you are searching for, but it may give you the idea.
}
}
// Third: rawContext is ready, transformation is done. Get the image out of it
CGImageRef outputImage1 = CGBitmapContextCreateImage(rawContext);
UIImage *outputImage2 = [UIImage imageWithCGImage:outputImage1];
CGImageRelease(outputImage1);
Okay... the output is RGBA, but you can create a greyscale + alpha context and just blit your image there for conversion.

Answer 3

This piece of code helped me applying hue on non-transparent area of an image.
- (UIImage*)imageWithImage:(UIImageView*)source colorValue:(CGFloat)hue {
CGSize imageSize = [source.image size];
CGRect imageExtent = CGRectMake(0,0,imageSize.width,imageSize.height);
// Create a context containing the image.
UIGraphicsBeginImageContext(imageSize);
CGContextRef context = UIGraphicsGetCurrentContext();
[source.image drawAtPoint:CGPointMake(0, 0)];
// Setup a clip region using the image
CGContextSaveGState(context);
CGContextClipToMask(context, source.bounds, source.image.CGImage);
self.imageColor = [UIColor colorWithHue:hue saturation:1.0 brightness:1 alpha:1.0];
[self.imageColor set];
CGContextFillRect(context, source.bounds);
// Draw the hue on top of the image.
CGContextSetBlendMode(context, kCGBlendModeHue);
[self.imageColor set];
UIBezierPath *imagePath = [UIBezierPath bezierPathWithRect:imageExtent];
[imagePath fill];
CGContextRestoreGState(context); // remove clip region
// Retrieve the new image.
UIImage *result = UIGraphicsGetImageFromCurrentImageContext();
UIGraphicsEndImageContext();
return result;
}