https://github.com/BrycensRanch/SnapX/blob/94771915abc26de19402ab690914bfb2adbfb846/SnapX.Core/SharpCapture/Linux/QImage.cs#L19
using System.Numerics;
using SixLabors.ImageSharp;
using SixLabors.ImageSharp.PixelFormats;
using SixLabors.ImageSharp.Processing;
using SnapX.Core.SharpCapture.Linux.DBus;
namespace SnapX.Core.SharpCapture.Linux;
internal static class QImage
{
public static Task<Image> LoadAsync(string imageFile, CaptureWorkspaceResult workspaceResult)
{
var format = workspaceResult.Format;
return format switch
{
// Seems like the only format returned by KWin at the moment.
QImageFormat.ARGB32Premultiplied => FromArgb32Premultiplied(imageFile, workspaceResult.Width, workspaceResult.Height),
// TODO add all possible formats that can be returned by KWin
_ => throw new NotImplementedException()
};
}
private static async Task<Image> FromArgb32Premultiplied(string imageFile, uint width, uint height)
{
var imageData = await File.ReadAllBytesAsync(imageFile);
// Q: "But isn't the format called `Argb`? Why are you loading the data as `Bgra`?"
// A: Yes, but for some reason the format provided by KWin isn't accurate to the format of the pixel data.
// Q: "Are you sure you didn't just mess up the QImageFormat enum?"
// A: The values of the enum match the documentation here: https://doc.qt.io/qt-6/qimage.html#Format-enum
var image = Image.LoadPixelData<Bgra32>(imageData, (int)width, (int)height);
image.Mutate(c => c.ProcessPixelRowsAsVector4(row =>
{
for (int col = 0; col < row.Length; col++)
{
var pixel = row[col];
var r = (byte)(pixel[2] * 255);
var g = (byte)(pixel[1] * 255);
var b = (byte)(pixel[0] * 255);
var a = (byte)(pixel[3] * 255);
if (a == 0)
{
r = 0;
g = 0;
b = 0;
row[col] = new Vector4(b, g, r, a);
}
else
{
var aFloat = a / 255f;
var rFloat = r / 255f;
var gFloat = g / 255f;
var bFloat = b / 255f;
row[col] = new Vector4(bFloat / aFloat, gFloat / aFloat, rFloat / aFloat, aFloat);
}
}
}));
return image;
}
}
