diff --git a/reference/opt/shaders/legacy/vert/transpose.legacy.vert b/reference/opt/shaders/legacy/vert/transpose.legacy.vert
index d725bfbb..61ee72e7 100644
--- a/reference/opt/shaders/legacy/vert/transpose.legacy.vert
+++ b/reference/opt/shaders/legacy/vert/transpose.legacy.vert
@@ -10,6 +10,10 @@ struct Buffer
 uniform Buffer _13;
 
 attribute vec4 Position;
+varying mat4 OutputMat1;
+varying mat4 OutputMat2;
+varying mat4 OutputMat3;
+varying mat4 OutputMat4;
 
 highp mat4 spvWorkaroundRowMajor(highp mat4 wrap) { return wrap; }
 mediump mat4 spvWorkaroundRowMajorMP(mediump mat4 wrap) { return wrap; }
@@ -23,8 +27,12 @@ void main()
 {
     mat4 _55 = _13.MVPRowMajor;
     mat4 _61 = spvWorkaroundRowMajor(_13.MVPColMajor);
-    mat4 _80 = spvTranspose(_13.MVPRowMajor) * 2.0;
-    mat4 _87 = spvTranspose(_61) * 2.0;
-    gl_Position = (((((((((((spvWorkaroundRowMajor(_13.M) * (Position * _13.MVPRowMajor)) + (spvWorkaroundRowMajor(_13.M) * (spvWorkaroundRowMajor(_13.MVPColMajor) * Position))) + (spvWorkaroundRowMajor(_13.M) * (_13.MVPRowMajor * Position))) + (spvWorkaroundRowMajor(_13.M) * (Position * spvWorkaroundRowMajor(_13.MVPColMajor)))) + (_55 * Position)) + (Position * _61)) + (Position * _55)) + (_61 * Position)) + (_80 * Position)) + (_87 * Position)) + (Position * _80)) + (Position * _87);
+    mat4 _80 = _13.MVPRowMajor * 2.0;
+    mat4 _87 = _61 * 2.0;
+    OutputMat1 = spvTranspose(_13.MVPRowMajor);
+    OutputMat2 = spvWorkaroundRowMajor(_13.MVPColMajor);
+    OutputMat3 = _55;
+    OutputMat4 = spvTranspose(_61);
+    gl_Position = (((((((((((spvWorkaroundRowMajor(_13.M) * (Position * _13.MVPRowMajor)) + (spvWorkaroundRowMajor(_13.M) * (spvWorkaroundRowMajor(_13.MVPColMajor) * Position))) + (spvWorkaroundRowMajor(_13.M) * (_13.MVPRowMajor * Position))) + (spvWorkaroundRowMajor(_13.M) * (Position * spvWorkaroundRowMajor(_13.MVPColMajor)))) + (_55 * Position)) + (Position * _61)) + (Position * _55)) + (_61 * Position)) + (Position * _80)) + (Position * _87)) + (_80 * Position)) + (_87 * Position);
 }
 
diff --git a/reference/shaders/legacy/vert/transpose.legacy.vert b/reference/shaders/legacy/vert/transpose.legacy.vert
index ce5cf8b1..a271c0be 100644
--- a/reference/shaders/legacy/vert/transpose.legacy.vert
+++ b/reference/shaders/legacy/vert/transpose.legacy.vert
@@ -10,6 +10,10 @@ struct Buffer
 uniform Buffer _13;
 
 attribute vec4 Position;
+varying mat4 OutputMat1;
+varying mat4 OutputMat2;
+varying mat4 OutputMat3;
+varying mat4 OutputMat4;
 
 highp mat4 spvWorkaroundRowMajor(highp mat4 wrap) { return wrap; }
 mediump mat4 spvWorkaroundRowMajorMP(mediump mat4 wrap) { return wrap; }
@@ -29,10 +33,14 @@ void main()
     vec4 c5 = Position * spvWorkaroundRowMajor(_13.MVPColMajor);
     vec4 c6 = Position * _13.MVPRowMajor;
     vec4 c7 = spvWorkaroundRowMajor(_13.MVPColMajor) * Position;
-    vec4 c8 = (spvTranspose(_13.MVPRowMajor) * 2.0) * Position;
-    vec4 c9 = (spvTranspose(spvWorkaroundRowMajor(_13.MVPColMajor)) * 2.0) * Position;
-    vec4 c10 = Position * (spvTranspose(_13.MVPRowMajor) * 2.0);
-    vec4 c11 = Position * (spvTranspose(spvWorkaroundRowMajor(_13.MVPColMajor)) * 2.0);
+    vec4 c8 = Position * (_13.MVPRowMajor * 2.0);
+    vec4 c9 = Position * (spvWorkaroundRowMajor(_13.MVPColMajor) * 2.0);
+    vec4 c10 = (_13.MVPRowMajor * 2.0) * Position;
+    vec4 c11 = (spvWorkaroundRowMajor(_13.MVPColMajor) * 2.0) * Position;
+    OutputMat1 = spvTranspose(_13.MVPRowMajor);
+    OutputMat2 = spvWorkaroundRowMajor(_13.MVPColMajor);
+    OutputMat3 = _13.MVPRowMajor;
+    OutputMat4 = spvTranspose(spvWorkaroundRowMajor(_13.MVPColMajor));
     gl_Position = ((((((((((c0 + c1) + c2) + c3) + c4) + c5) + c6) + c7) + c8) + c9) + c10) + c11;
 }
 
diff --git a/shaders/legacy/vert/transpose.legacy.vert b/shaders/legacy/vert/transpose.legacy.vert
index 588c28d5..46f6d20b 100644
--- a/shaders/legacy/vert/transpose.legacy.vert
+++ b/shaders/legacy/vert/transpose.legacy.vert
@@ -9,6 +9,11 @@ uniform Buffer
 
 layout(location = 0) in vec4 Position;
 
+layout(location = 1) out mat4 OutputMat1;
+layout(location = 5) out mat4 OutputMat2;
+layout(location = 9) out mat4 OutputMat3;
+layout(location = 13) out mat4 OutputMat4;
+
 void main()
 {
 	vec4 c0 = M * (MVPRowMajor * Position);
@@ -21,12 +26,18 @@ void main()
 	vec4 c6 = Position * transpose(MVPRowMajor);
 	vec4 c7 = Position * transpose(MVPColMajor);
 
-	// Multiplying by scalar forces resolution of the transposition
+	// Multiplying by scalar does not affect transposition
 	vec4 c8 = (MVPRowMajor * 2.0) * Position;
 	vec4 c9 = (transpose(MVPColMajor) * 2.0) * Position;
 	vec4 c10 = Position * (MVPRowMajor * 2.0);
 	vec4 c11 = Position * (transpose(MVPColMajor) * 2.0);
 
+	// Outputting a matrix will trigger an actual transposition
+	OutputMat1 = MVPRowMajor;
+	OutputMat2 = MVPColMajor;
+	OutputMat3 = transpose(MVPRowMajor);
+	OutputMat4 = transpose(MVPColMajor);
+
 	gl_Position = c0 + c1 + c2 + c3 + c4 + c5 + c6 + c7 + c8 + c9 + c10 + c11;
 }
 
diff --git a/spirv_glsl.cpp b/spirv_glsl.cpp
index 08cf5381..13deec28 100644
--- a/spirv_glsl.cpp
+++ b/spirv_glsl.cpp
@@ -11916,8 +11916,29 @@ void CompilerGLSL::emit_instruction(const Instruction &instruction)
 		break;
 	}
 
-	case OpFMul:
 	case OpMatrixTimesScalar:
+	{
+		auto *a = maybe_get<SPIRExpression>(ops[2]);
+
+		// If the matrix need transpose, just mark the result as needing so.
+		if (a && a->need_transpose)
+		{
+			a->need_transpose = false;
+			auto expr = join(enclose_expression(to_unpacked_row_major_matrix_expression(ops[2])), " * ",
+			                 to_enclosed_unpacked_expression(ops[3]));
+			bool forward = should_forward(ops[2]) && should_forward(ops[3]);
+			auto &e = emit_op(ops[0], ops[1], expr, forward);
+			e.need_transpose = true;
+			a->need_transpose = true;
+			inherit_expression_dependencies(ops[1], ops[2]);
+			inherit_expression_dependencies(ops[1], ops[3]);
+		}
+		else
+			GLSL_BOP(*);
+		break;
+	}
+
+	case OpFMul:
 	case OpVectorTimesScalar:
 		GLSL_BOP(*);
 		break;