More Drawing

More 100 Days of SwiftUI done today. I've used CGAffineTransform a little bit, but I'm still not really comfortable with 2d transforms (much less 3d). I guess if I wanna do a lot of drawing with code, I'll have to get comfortable with this.

I don't think I'll have much use for ImagePaint or even-odd fills (except for debugging), but CGAffineTransform and .drawingGroup() will definitely be a part of my toolkit going forward. I still have yet to watch the WWDC video on Canvas - Add rich graphics to your SwiftUI app - so I'll try to get to that over the weekend.

Meanwhile, I implemented another sketch in Swift/SwiftUI from the Generative Design book today, Color Palettes from Images. After re-implementing the p5js sketch in Swift, I kept going with the idea and wired up the camera to it, processing frames from the camera and then sorting pixels by hsb values. Here's a self-portrait, with sampled colors sorted by brightness.

I'm not sure my approach is totally optimal, I'm swapping between UIImage, CGImage, and CIimage formats and I worry that's moving the pixels from the GPU to the CPU and back again. Not really a problem for a one-off photo process, but working with video frames means I really need to pay attention to performance. I know next to nothing about Core Image but I wouldn't be surprised if there's a way with CISampling and CIFilter to sample the color from the pixels of an image into an array for sorting. More stuff to add to the list!