diff --git a/DirectXTex/DirectXTex.h b/DirectXTex/DirectXTex.h
index d9fc97e..9a44038 100644
--- a/DirectXTex/DirectXTex.h
+++ b/DirectXTex/DirectXTex.h
@@ -549,6 +549,10 @@ namespace DirectX
         CMSE_IGNORE_BLUE            = 0x40,
         CMSE_IGNORE_ALPHA           = 0x80,
             // Ignore the channel when computing MSE
+
+        CMSE_IMAGE1_X2_BIAS         = 0x100,
+        CMSE_IMAGE2_X2_BIAS         = 0x200,
+            // Indicates that image should be scaled and biased before comparison (i.e. UNORM -> SNORM)
     };
 
     HRESULT ComputeMSE( _In_ const Image& image1, _In_ const Image& image2, _Out_ float& mse, _Out_writes_opt_(4) float* mseV, _In_ DWORD flags = 0 );
diff --git a/DirectXTex/DirectXTexMisc.cpp b/DirectXTex/DirectXTexMisc.cpp
index ac29864..6f7b99f 100644
--- a/DirectXTex/DirectXTexMisc.cpp
+++ b/DirectXTex/DirectXTexMisc.cpp
@@ -84,6 +84,7 @@ static HRESULT _ComputeMSE( _In_ const Image& image1, _In_ const Image& image2,
     const size_t rowPitch2 = image2.rowPitch;
 
     XMVECTOR acc = g_XMZero;
+    static XMVECTORF32 two = { 2.0f, 2.0f, 2.0f, 2.0f };
 
     for( size_t h = 0; h < image1.height; ++h )
     {
@@ -102,12 +103,20 @@ static HRESULT _ComputeMSE( _In_ const Image& image1, _In_ const Image& image2,
             {
                 v1 = XMVectorPow( v1, g_Gamma22 );
             }
+            if ( flags & CMSE_IMAGE1_X2_BIAS )
+            {
+                v1 = XMVectorMultiplyAdd( v1, two, g_XMNegativeOne );
+            }
 
             XMVECTOR v2 = *(ptr2++);
             if ( flags & CMSE_IMAGE2_SRGB )
             {
                 v2 = XMVectorPow( v2, g_Gamma22 );
             }
+            if ( flags & CMSE_IMAGE2_X2_BIAS )
+            {
+                v1 = XMVectorMultiplyAdd( v2, two, g_XMNegativeOne );
+            }
 
             // sum[ (I1 - I2)^2 ]
             XMVECTOR v = XMVectorSubtract( v1, v2 );