diff --git a/stepmania/src/RageDisplay.h b/stepmania/src/RageDisplay.h
index d254f7b10f..9e8751eba5 100644
--- a/stepmania/src/RageDisplay.h
+++ b/stepmania/src/RageDisplay.h
@@ -145,6 +145,7 @@ public:
 
 	virtual bool SupportsTextureFormat( PixelFormat pixfmt, bool realtime=false ) = 0;
 	virtual bool SupportsThreadedRendering() { return false; }
+	virtual bool SupportsPerVertexMatrixScale() = 0;
 
 	/* return 0 if failed or internal texture resource handle 
 	 * (unsigned in OpenGL, texture pointer in D3D) */
diff --git a/stepmania/src/RageDisplay_D3D.h b/stepmania/src/RageDisplay_D3D.h
index e3a6ea82a5..c90966578d 100644
--- a/stepmania/src/RageDisplay_D3D.h
+++ b/stepmania/src/RageDisplay_D3D.h
@@ -19,6 +19,7 @@ public:
 	void SetBlendMode( BlendMode mode );
 	bool SupportsTextureFormat( PixelFormat pixfmt, bool realtime=false );
 	bool SupportsThreadedRendering();
+	bool SupportsPerVertexMatrixScale() { return false; }
 	unsigned CreateTexture( 
 		PixelFormat pixfmt, 
 		RageSurface* img, 
diff --git a/stepmania/src/RageDisplay_Null.h b/stepmania/src/RageDisplay_Null.h
index 6694e1fb67..2c66cab366 100644
--- a/stepmania/src/RageDisplay_Null.h
+++ b/stepmania/src/RageDisplay_Null.h
@@ -16,6 +16,7 @@ public:
 	VideoModeParams GetVideoModeParams() const { return m_Params; }
 	void SetBlendMode( BlendMode mode ) { }
 	bool SupportsTextureFormat( PixelFormat pixfmt, bool realtime=false ) { return true; }
+	bool SupportsPerVertexMatrixScale() { return false; }
 	unsigned CreateTexture( 
 		PixelFormat pixfmt, 
 		RageSurface* img,
diff --git a/stepmania/src/RageDisplay_OGL.cpp b/stepmania/src/RageDisplay_OGL.cpp
index 0c3bf7ea7e..f7ce7660be 100644
--- a/stepmania/src/RageDisplay_OGL.cpp
+++ b/stepmania/src/RageDisplay_OGL.cpp
@@ -1198,7 +1198,7 @@ void RageCompiledGeometryHWOGL::Draw( int iMeshIndex ) const
 
 	if( meshInfo.m_bNeedsTextureMatrixScale )
 	{
-		if( false )	//g_bTextureMatrixShader != 0 )
+		if( g_bTextureMatrixShader != 0 )
 		{
 			/* If we're using texture matrix scales, set up that buffer, too, and enable the
 			 * vertex shader.  This shader doesn't support all OpenGL state, so only enable it
@@ -1998,6 +1998,11 @@ bool RageDisplay_OGL::SupportsTextureFormat( PixelFormat pixfmt, bool bRealtime
 	}
 }
 
+bool RageDisplay_OGL::SupportsPerVertexMatrixScale()
+{
+	return g_bTextureMatrixShader != 0;
+}
+
 void RageDisplay_OGL::SetSphereEnvironmentMapping( bool b )
 {
 	if( b )
diff --git a/stepmania/src/RageDisplay_OGL.h b/stepmania/src/RageDisplay_OGL.h
index cc28a90ebb..450b601743 100644
--- a/stepmania/src/RageDisplay_OGL.h
+++ b/stepmania/src/RageDisplay_OGL.h
@@ -19,6 +19,7 @@ public:
 	VideoModeParams GetVideoModeParams() const;
 	void SetBlendMode( BlendMode mode );
 	bool SupportsTextureFormat( PixelFormat pixfmt, bool realtime=false );
+	bool SupportsPerVertexMatrixScale();
 	unsigned CreateTexture( 
 		PixelFormat pixfmt, 
 		RageSurface* img,
diff --git a/stepmania/src/RageModelGeometry.cpp b/stepmania/src/RageModelGeometry.cpp
index 059457f6b7..87a516e0a1 100644
--- a/stepmania/src/RageModelGeometry.cpp
+++ b/stepmania/src/RageModelGeometry.cpp
@@ -61,12 +61,12 @@ void RageModelGeometry::MergeMeshes( int iFromIndex, int iToIndex )
 	msMesh& meshTo = m_Meshes[ iToIndex ];
 
 	int iShiftTriangleVertexIndicesBy = meshTo.Vertices.size();
-	int iStartShiftingTriangleAtIndex = meshTo.Triangles.size();
+	int iStartShiftingAtTriangleIndex = meshTo.Triangles.size();
 
-	meshTo.Vertices.insert( meshTo.Vertices.end(), meshFrom.Vertices.end(), meshFrom.Vertices.end() );
-	meshTo.Triangles.insert( meshTo.Triangles.end(), meshFrom.Triangles.end(), meshFrom.Triangles.end() );
+	meshTo.Vertices.insert( meshTo.Vertices.end(), meshFrom.Vertices.begin(), meshFrom.Vertices.end() );
+	meshTo.Triangles.insert( meshTo.Triangles.end(), meshFrom.Triangles.begin(), meshFrom.Triangles.end() );
 
-	for( unsigned i=iStartShiftingTriangleAtIndex; i<meshTo.Triangles.size(); i++ )
+	for( unsigned i=iStartShiftingAtTriangleIndex; i<meshTo.Triangles.size(); i++ )
 	{
 		for( int j=0; j<3; j++ )
 			meshTo.Triangles[i].nVertexIndices[j] += iShiftTriangleVertexIndicesBy;
@@ -262,6 +262,14 @@ void RageModelGeometry::LoadMilkshapeAscii( const CString& _sPath, bool bNeedsNo
 
 	OptimizeBones();
 
+	if( DISPLAY->SupportsPerVertexMatrixScale() )
+	{
+		if( m_Meshes.size() == 2  &&  m_Meshes[0].sName == m_Meshes[1].sName )
+		{
+			MergeMeshes( 1, 0 );
+		}
+	}
+
 	// send the finalized vertices to the graphics card
 	m_pCompiledGeometry->Set( m_Meshes, bNeedsNormals );