diff --git a/Source/Engine/Graphics/Shaders/GPUVertexLayout.cpp b/Source/Engine/Graphics/Shaders/GPUVertexLayout.cpp
index 585568d60..23382673f 100644
--- a/Source/Engine/Graphics/Shaders/GPUVertexLayout.cpp
+++ b/Source/Engine/Graphics/Shaders/GPUVertexLayout.cpp
@@ -216,20 +216,21 @@ GPUVertexLayout* GPUVertexLayout::Get(const Span<GPUVertexLayout*>& layouts)
     return result;
 }
 
-GPUVertexLayout* GPUVertexLayout::Merge(GPUVertexLayout* base, GPUVertexLayout* reference, bool removeUnused, bool addMissing, int32 missingSlotOverride)
+GPUVertexLayout* GPUVertexLayout::Merge(GPUVertexLayout* base, GPUVertexLayout* reference, bool removeUnused, bool addMissing, int32 missingSlotOverride, bool referenceOrder)
 {
     GPUVertexLayout* result = base ? base : reference;
     if (base && reference && base != reference)
     {
         bool elementsModified = false;
         Elements newElements = base->GetElements();
+        const Elements& refElements = reference->GetElements();
         if (removeUnused)
         {
             for (int32 i = newElements.Count() - 1; i >= 0; i--)
             {
                 bool missing = true;
                 const VertexElement& e = newElements.Get()[i];
-                for (const VertexElement& ee : reference->GetElements())
+                for (const VertexElement& ee : refElements)
                 {
                     if (ee.Type == e.Type)
                     {
@@ -247,7 +248,7 @@ GPUVertexLayout* GPUVertexLayout::Merge(GPUVertexLayout* base, GPUVertexLayout*
         }
         if (addMissing)
         {
-            for (const VertexElement& e : reference->GetElements())
+            for (const VertexElement& e : refElements)
             {
                 bool missing = true;
                 for (const VertexElement& ee : base->GetElements())
@@ -282,6 +283,32 @@ GPUVertexLayout* GPUVertexLayout::Merge(GPUVertexLayout* base, GPUVertexLayout*
                 }
             }
         }
+        if (referenceOrder)
+        {
+            for (int32 i = 0, j = 0; i < newElements.Count() && j < refElements.Count(); j++)
+            {
+                if (newElements[i].Type == refElements[j].Type)
+                {
+                    // Elements match so move forward
+                    i++;
+                    continue;
+                }
+
+                // Find reference element in a new list
+                for (int32 k = i + 1; k < newElements.Count(); k++)
+                {
+                    if (newElements[k].Type == refElements[j].Type)
+                    {
+                        // Move matching element to the reference position
+                        VertexElement e = newElements[k];
+                        newElements.RemoveAt(k);
+                        newElements.Insert(i, e);
+                        i++;
+                        break;
+                    }
+                }
+            }
+        }
         if (elementsModified)
             result = Get(newElements, true);
     }
diff --git a/Source/Engine/Graphics/Shaders/GPUVertexLayout.h b/Source/Engine/Graphics/Shaders/GPUVertexLayout.h
index 9d33566fb..04815565b 100644
--- a/Source/Engine/Graphics/Shaders/GPUVertexLayout.h
+++ b/Source/Engine/Graphics/Shaders/GPUVertexLayout.h
@@ -84,8 +84,9 @@ public:
     /// <param name="removeUnused">True to remove elements from base layout that don't exist in a reference layout.</param>
     /// <param name="addMissing">True to add missing elements to base layout that exist in a reference layout.</param>
     /// <param name="missingSlotOverride">Allows to override the input slot for missing elements. Use value -1 to inherit slot from the reference layout.</param>
+    /// <param name="referenceOrder">True to reorder result elements to match the reference layout. For example, if input vertex buffer layout is different than vertex shader then it can match those.</param>
     /// <returns>Vertex layout object. Doesn't need to be cleared as it's cached for an application lifetime.</returns>
-    static GPUVertexLayout* Merge(GPUVertexLayout* base, GPUVertexLayout* reference, bool removeUnused = false, bool addMissing = true, int32 missingSlotOverride = -1);
+    static GPUVertexLayout* Merge(GPUVertexLayout* base, GPUVertexLayout* reference, bool removeUnused = false, bool addMissing = true, int32 missingSlotOverride = -1, bool referenceOrder = false);
 
 public:
     // [GPUResource]