Drawing on a zoomable view

Question

I'm working on a small drawing application, which has a basic requirement of supporting zoom-in/out. I have two main issues:

1. Drawing doesn't appear crisp and clear, when view is zoomed/transformed. Is there a better approach, or is there a way to improve the drawing when the view is zoomed?

2. The drawing performance is slow, when drawing on 1200 x 1200 pts canvas (on iPhone). Any chance I can improve it for large canvas sizes?

Zooming Code:

- (void)scale:(UIPinchGestureRecognizer *)gestureRecognizer {
    [self adjustAnchorPointForGestureRecognizer:gestureRecognizer];

    UIView *canvas = [gestureRecognizer view];

    if ([gestureRecognizer state] == UIGestureRecognizerStateBegan ||
        [gestureRecognizer state] == UIGestureRecognizerStateChanged) {

        // Calculate the drawing view's size
        CGSize drawingViewSize = ...;

        // Calculate the minimum allowed tranform size
        // Developer's Note: I actually wanted to use the size 1/4th of the view
        // but self.view.frame.size doesn't return the correct (actual) width and height
        // It returns these values inverted i.e. width as height, and vice verse.
        // The reason is that the view appears to be transformed (90 degrees).
        // Since there's no work-around this, so for now, I'm just using fixed values.
        CGSize minTranformSize = CGSizeMake(100.0f, 100.0f);

        if ((minTranformSize.width > drawingViewSize.width) && (minTranformSize.height > drawingViewSize.height)) {
            minTranformSize = drawingViewSize;
        }

        // Transform the view, provided
        // 1. It won't scale more than the original size of the background image
        // 2. It won't scale less than the minimum possible transform
        CGSize transformedSize = CGSizeMake(canvas.frame.size.width * [gestureRecognizer scale],
                                            canvas.frame.size.height * [gestureRecognizer scale]);

        if ((transformedSize.width <= drawingViewSize.width) && (transformedSize.height <= drawingViewSize.height) &&
            (transformedSize.width >= minTranformSize.width) && (transformedSize.height >= minTranformSize.height)) {

            canvas.transform = CGAffineTransformScale([canvas transform],
                                                      [gestureRecognizer scale],
                                                      [gestureRecognizer scale]);
        }

        [gestureRecognizer setScale:1.0];

    } else if ([gestureRecognizer state] == UIGestureRecognizerStateEnded) {

        // Recenter the container view, if required (piece is smaller than the view and it's not aligned)
        CGSize viewSize = self.view.bounds.size;

        if ((canvas.frame.size.width < viewSize.width) ||
            (canvas.frame.size.height < viewSize.height)) {

            canvas.center = CGPointMake(viewSize.width/2, viewSize.height/2);
        }

        // Adjust the x/y coordinates, if required (piece is larger than the view and it's moving outwards from the view)
        if (((canvas.frame.origin.x > 0) || (canvas.frame.origin.y > 0)) &&
            ((canvas.frame.size.width >= viewSize.width) && (canvas.frame.size.height >= viewSize.height))) {

            canvas.frame = CGRectMake(0.0,
                                      0.0,
                                      canvas.frame.size.width,
                                      canvas.frame.size.height);
        }

        canvas.frame = CGRectIntegral(canvas.frame);
    }
}

Drawing Code

- (void)draw {
    CGContextRef context = UIGraphicsGetCurrentContext();
    CGContextSaveGState(context);

    if (self.fillColor) {
        [self.fillColor setFill];
        [self.path fill];
    }

    if ([self.strokeColor isEqual:[UIColor clearColor]]) {
        [self.path strokeWithBlendMode:kCGBlendModeClear alpha:1.0];

    } else if (self.strokeColor) {
        [self.strokeColor setStroke];
        [self.path stroke];
    }

    CGContextRestoreGState(context);
}

Answer 1

This is a pretty complicated problem, that I have struggled a lot with.

I ended up converting the drawings to vector.

1. draw all lines in one layer, draw all fills in another.
2. Convert the line drawings to Vector, using potrace (http://potrace.sourceforge.net/)
3. draw the vector using SVGKit (https://github.com/SVGKit/SVGKit) and hide the layer drawn in 1)

It is working pretty well and fairly fast, but it requires a lot of work. We have an app in our company that applies this technique:

https://itunes.apple.com/us/app/ideal-paint-hd-mormor/id569450492?mt=8.

If your only problem is performance, try taking a look at CATiledLayer. (also used in app mentioned above). It will increase performance tremendously (You can find a pretty good tutorial here http://www.cimgf.com/2011/03/01/subduing-catiledlayer/).

Good luck! :)