Use ImageMagick for AVIF reference decoding

Register AVIF with the existing reference decoder and retain 10/12-bit
sample precision when expanding native samples to Rgba64.

Release/net11 verification used VerifyEncoder with exact comparisons for
8-bit color/alpha and 10/12-bit source pixels. This checkpoint does not
establish complete AV1 conformance or independent ICC parity.

tests/ImageSharp.Tests/TestUtilities/ReferenceCodecs/MagickReferenceDecoder.cs

@@ -7,6 +7,7 @@ using ImageMagick.Formats;
 using SixLabors.ImageSharp.Formats;
 using SixLabors.ImageSharp.Formats.Bmp;
 using SixLabors.ImageSharp.Formats.Exr;
+using SixLabors.ImageSharp.Formats.Heif;
 using SixLabors.ImageSharp.Formats.Jpeg;
 using SixLabors.ImageSharp.Formats.Png;
 using SixLabors.ImageSharp.Formats.Tiff;
@@ -45,6 +46,8 @@ public class MagickReferenceDecoder : ImageDecoder
 
     public static MagickReferenceDecoder Exr { get; } = new(ExrFormat.Instance);
 
+    public static MagickReferenceDecoder Heif { get; } = new(HeifFormat.Instance);
+
     protected override Image<TPixel> Decode<TPixel>(DecoderOptions options, Stream stream, CancellationToken cancellationToken)
     {
         ImageMetadata metadata = new();
@@ -83,13 +86,13 @@ public class MagickReferenceDecoder : ImageDecoder
                 (int)magicFrame.Height);
 
             using IUnsafePixelCollection<ushort> pixels = magicFrame.GetPixelsUnsafe();
-            if (magicFrame.Depth is 12 or 10 or 8 or 6 or 5 or 4 or 3 or 2 or 1)
+            if (magicFrame.Depth is 8 or 6 or 5 or 4 or 3 or 2 or 1)
             {
                 byte[] data = pixels.ToByteArray(PixelMapping.RGBA);
 
                 FromRgba32Bytes(configuration, data, buffer);
             }
-            else if (magicFrame.Depth is 14 or 16 or 32)
+            else if (magicFrame.Depth is 10 or 12 or 14 or 16 or 32)
             {
                 if (this.imageFormat is PngFormat png)
                 {
@@ -97,6 +100,21 @@ public class MagickReferenceDecoder : ImageDecoder
                 }
 
                 ushort[] data = pixels.ToShortArray(PixelMapping.RGBA);
+                if (this.imageFormat is HeifFormat && magicFrame.Depth is 10 or 12)
+                {
+                    // The HEIF reader left-aligns native samples in its 16-bit channels. Recover the declared
+                    // sample values before expanding their full range to Rgba64; a shift alone leaves white
+                    // below 65535 and changes every nonzero component's normalized value.
+                    int depth = (int)magicFrame.Depth;
+                    int shift = 16 - depth;
+                    int maximum = (1 << depth) - 1;
+                    for (int i = 0; i < data.Length; i++)
+                    {
+                        int sample = data[i] >> shift;
+                        data[i] = (ushort)(((sample * 65535) + (maximum / 2)) / maximum);
+                    }
+                }
+
                 Span<byte> bytes = MemoryMarshal.Cast<ushort, byte>(data.AsSpan());
                 FromRgba64Bytes(configuration, bytes, buffer);
             }

tests/ImageSharp.Tests/TestUtilities/TestEnvironment.Formats.cs

@@ -85,6 +85,12 @@ public static partial class TestEnvironment
             bmpEncoder,
             new BmpImageFormatDetector());
 
+        cfg.ConfigureCodecs(
+            HeifFormat.Instance,
+            MagickReferenceDecoder.Heif,
+            new HeifEncoder(),
+            new HeifImageFormatDetector());
+
         return cfg;
     }
 }