A simple Swift wrapper around the TinyJPEG library.
It makes the interface very Swifty as seen here:
// this is all just generating simple, raw image data
let imageDimensions = (x: 1275, y: 800)
let imageRowStride = 1280 * 4
func offset(for point: Point, channelCount: Int) -> Int {
(point.y * imageRowStride) + point.x * channelCount
}
let buffer = UnsafeMutableRawBufferPointer.allocate(byteCount: imageRowStride * imageDimensions.y, alignment: 32)
for y in 0..<imageDimensions.y {
for x in 0..<imageDimensions.x {
let redIndex = offset(for: (x, y), channelCount: 4)
let red = (Double(x) / Double(imageDimensions.x)) * 255
let green = (Double(y) / Double(imageDimensions.y)) * 255
let blue = 255 - red
buffer[redIndex] = UInt8(red)
buffer[redIndex + 1] = UInt8(green)
buffer[redIndex + 2] = UInt8(blue)
buffer[redIndex + 3] = 255
}
}
let pixelData = Data(bytesNoCopy: buffer.baseAddress!, count: buffer.count, deallocator: .free)
// now that that's done (you'll probably get this from CoreGraphics or any other library
providing raw, 8 bits/channel RGB or RGBA interleaved data), here's the actual magic:
let jpeg = try TinySwiftJPEG.encodeJPEG( // this is the actual magic
from: pixelData,
rowStride: imageRowStride,
width: imageDimensions.x,
height: imageDimensions.y,
channels: .rgba)
try jpeg.write(to: URL(filePath: "/path/to/output.jpg")I made a small modification to the TinyJPEG library to allow for a different count of bytes per row
than simply just width * channelCount (channel count being either 3 or 4, depending on alpha) as
that's a very common situation in CoreGraphics (and probably elsewhere). That should save us an
unecessary byte copy!