From 49d20566a366e1bbda1df9986ff8ffa784172b06 Mon Sep 17 00:00:00 2001
From: James Jackson-South <james_south@hotmail.com>
Date: Sat, 25 Jul 2026 23:32:25 +1000
Subject: [PATCH] Use tensor addition for histogram offsets

---
 .../Common/Helpers/TensorPrimitives.cs        | 12 +++++++++
 .../Formats/Png/Filters/IPngFilterOperator.cs |  1 +
 .../Formats/Webp/Lossless/Vp8LHistogram.cs    |  1 -
 .../HistogramEqualizationProcessor{TPixel}.cs |  6 ++---
 .../Common/TensorPrimitivesTests.cs           | 27 +++++++++++++++++++
 5 files changed, 42 insertions(+), 5 deletions(-)

diff --git a/src/ImageSharp/Common/Helpers/TensorPrimitives.cs b/src/ImageSharp/Common/Helpers/TensorPrimitives.cs
index 7f97776ab..bcfd16bb9 100644
--- a/src/ImageSharp/Common/Helpers/TensorPrimitives.cs
+++ b/src/ImageSharp/Common/Helpers/TensorPrimitives.cs
@@ -137,6 +137,18 @@ internal static class TensorPrimitives_
         where T : IAdditionOperators<T, T, T>, IAdditiveIdentity<T, T>
         => InvokeSpanSpanIntoSpan<T, AddOperator<T>>(x, y, destination);
 
+    /// <summary>
+    /// Computes the element-wise sum of the values in <paramref name="x"/> and the scalar <paramref name="y"/>.
+    /// </summary>
+    /// <typeparam name="T">The element type.</typeparam>
+    /// <param name="x">The first addends.</param>
+    /// <param name="y">The scalar second addend.</param>
+    /// <param name="destination">The destination for the sums.</param>
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    public static void Add<T>(ReadOnlySpan<T> x, T y, Span<T> destination)
+        where T : IAdditionOperators<T, T, T>, IAdditiveIdentity<T, T>
+        => InvokeSpanScalarIntoSpan<T, AddOperator<T>>(x, y, destination);
+
     /// <summary>
     /// Computes the element-wise result of dividing the values in <paramref name="x"/> by <paramref name="y"/>.
     /// </summary>
diff --git a/src/ImageSharp/Formats/Png/Filters/IPngFilterOperator.cs b/src/ImageSharp/Formats/Png/Filters/IPngFilterOperator.cs
index c81187775..bae1ac477 100644
--- a/src/ImageSharp/Formats/Png/Filters/IPngFilterOperator.cs
+++ b/src/ImageSharp/Formats/Png/Filters/IPngFilterOperator.cs
@@ -3,6 +3,7 @@
 
 using System.Runtime.CompilerServices;
 using System.Runtime.Intrinsics;
+using System.Runtime.Intrinsics.Arm;
 using System.Runtime.Intrinsics.X86;
 using SixLabors.ImageSharp.Common.Helpers;
 
diff --git a/src/ImageSharp/Formats/Webp/Lossless/Vp8LHistogram.cs b/src/ImageSharp/Formats/Webp/Lossless/Vp8LHistogram.cs
index c59235e70..eca60dd3f 100644
--- a/src/ImageSharp/Formats/Webp/Lossless/Vp8LHistogram.cs
+++ b/src/ImageSharp/Formats/Webp/Lossless/Vp8LHistogram.cs
@@ -533,7 +533,6 @@ internal abstract unsafe class Vp8LHistogram
 
         return cost;
     }
-
 }
 
 internal sealed unsafe class OwnedVp8LHistogram : Vp8LHistogram, IDisposable
diff --git a/src/ImageSharp/Processing/Processors/Normalization/HistogramEqualizationProcessor{TPixel}.cs b/src/ImageSharp/Processing/Processors/Normalization/HistogramEqualizationProcessor{TPixel}.cs
index fd01bd8dd..b624d5b0e 100644
--- a/src/ImageSharp/Processing/Processors/Normalization/HistogramEqualizationProcessor{TPixel}.cs
+++ b/src/ImageSharp/Processing/Processors/Normalization/HistogramEqualizationProcessor{TPixel}.cs
@@ -4,6 +4,7 @@
 using System.Numerics;
 using System.Runtime.CompilerServices;
 using System.Runtime.InteropServices;
+using SixLabors.ImageSharp.Common.Helpers;
 using SixLabors.ImageSharp.PixelFormats;
 
 namespace SixLabors.ImageSharp.Processing.Processors.Normalization;
@@ -115,10 +116,7 @@ internal abstract class HistogramEqualizationProcessor<TPixel> : ImageProcessor<
         int addToEachBin = sumOverClip > 0 ? (int)MathF.Floor(sumOverClip / this.luminanceLevelsFloat) : 0;
         if (addToEachBin > 0)
         {
-            for (nuint i = 0; i < (uint)histogram.Length; i++)
-            {
-                Unsafe.Add(ref histogramBase, i) += addToEachBin;
-            }
+            TensorPrimitives_.Add(histogram, addToEachBin, histogram);
         }
 
         int residual = sumOverClip - (addToEachBin * this.LuminanceLevels);
diff --git a/tests/ImageSharp.Tests/Common/TensorPrimitivesTests.cs b/tests/ImageSharp.Tests/Common/TensorPrimitivesTests.cs
index 1cc5b943f..375fc000d 100644
--- a/tests/ImageSharp.Tests/Common/TensorPrimitivesTests.cs
+++ b/tests/ImageSharp.Tests/Common/TensorPrimitivesTests.cs
@@ -90,6 +90,33 @@ public class TensorPrimitivesTests
         Assert.Equal(expected, x);
     }
 
+    /// <summary>
+    /// Verifies that scalar integer addition produces identical results for separate and in-place destinations.
+    /// </summary>
+    /// <param name="length">The input length.</param>
+    [Theory]
+    [MemberData(nameof(SpanLengths))]
+    public void AddScalarInt32MatchesScalarFormula(int length)
+    {
+        int[] source = new int[length];
+        int[] expected = new int[length];
+        const int addend = 17;
+
+        for (int i = 0; i < length; i++)
+        {
+            source[i] = (i * 37) - 200;
+            expected[i] = source[i] + addend;
+        }
+
+        int[] destination = new int[length];
+        TensorPrimitives_.Add(source, addend, destination);
+        Assert.Equal(expected, destination);
+
+        int[] inPlace = (int[])source.Clone();
+        TensorPrimitives_.Add(inPlace, addend, inPlace);
+        Assert.Equal(expected, inPlace);
+    }
+
     /// <summary>
     /// Verifies that integer clamping produces identical results for separate and in-place destinations.
     /// </summary>