diff --git a/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc b/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc
index d2e39ead1817b..b70c717394894 100644
--- a/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc
+++ b/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc
@@ -239,7 +239,7 @@ DownsamplePassArgs CalculateDownsamplePassArgs(
   //     .Contains(coverage_hint.value()))
 
   std::optional<Rect> snapshot_coverage = input_snapshot.GetCoverage();
-  if (input_snapshot.transform.IsIdentity() &&
+  if (input_snapshot.transform.Equals(snapshot_entity.GetTransform()) &&
       source_expanded_coverage_hint.has_value() &&
       snapshot_coverage.has_value() &&
       snapshot_coverage->Contains(source_expanded_coverage_hint.value())) {
diff --git a/impeller/geometry/matrix.h b/impeller/geometry/matrix.h
index 1726c30151fdf..166fc11f9609f 100644
--- a/impeller/geometry/matrix.h
+++ b/impeller/geometry/matrix.h
@@ -407,6 +407,27 @@ struct Matrix {
 
   std::optional<MatrixDecomposition> Decompose() const;
 
+  bool Equals(const Matrix& matrix, Scalar epsilon = 1e-5f) const {
+    const Scalar* a = m;
+    const Scalar* b = matrix.m;
+    return ScalarNearlyEqual(a[0], b[0], epsilon) &&
+           ScalarNearlyEqual(a[1], b[1], epsilon) &&
+           ScalarNearlyEqual(a[2], b[2], epsilon) &&
+           ScalarNearlyEqual(a[3], b[3], epsilon) &&
+           ScalarNearlyEqual(a[4], b[4], epsilon) &&
+           ScalarNearlyEqual(a[5], b[5], epsilon) &&
+           ScalarNearlyEqual(a[6], b[6], epsilon) &&
+           ScalarNearlyEqual(a[7], b[7], epsilon) &&
+           ScalarNearlyEqual(a[8], b[8], epsilon) &&
+           ScalarNearlyEqual(a[9], b[9], epsilon) &&
+           ScalarNearlyEqual(a[10], b[10], epsilon) &&
+           ScalarNearlyEqual(a[11], b[11], epsilon) &&
+           ScalarNearlyEqual(a[12], b[12], epsilon) &&
+           ScalarNearlyEqual(a[13], b[13], epsilon) &&
+           ScalarNearlyEqual(a[14], b[14], epsilon) &&
+           ScalarNearlyEqual(a[15], b[15], epsilon);
+  }
+
   constexpr bool operator==(const Matrix& m) const {
     // clang-format off
     return vec[0] == m.vec[0]
diff --git a/impeller/geometry/matrix_unittests.cc b/impeller/geometry/matrix_unittests.cc
index ac160086994e3..91a2a50f2a7d4 100644
--- a/impeller/geometry/matrix_unittests.cc
+++ b/impeller/geometry/matrix_unittests.cc
@@ -25,6 +25,18 @@ TEST(MatrixTest, Multiply) {
                                         11.0, 21.0, 0.0, 1.0)));
 }
 
+TEST(MatrixTest, Equals) {
+  Matrix x;
+  Matrix y = x;
+  EXPECT_TRUE(x.Equals(y));
+}
+
+TEST(MatrixTest, NotEquals) {
+  Matrix x;
+  Matrix y = x.Translate({1, 0, 0});
+  EXPECT_FALSE(x.Equals(y));
+}
+
 TEST(MatrixTest, HasPerspective2D) {
   EXPECT_FALSE(Matrix().HasPerspective2D());
 
diff --git a/impeller/geometry/scalar.h b/impeller/geometry/scalar.h
index 2600a49c42dc5..dadc52850cb5f 100644
--- a/impeller/geometry/scalar.h
+++ b/impeller/geometry/scalar.h
@@ -22,6 +22,11 @@ constexpr T Absolute(const T& val) {
   return val >= T{} ? val : -val;
 }
 
+template <>
+constexpr Scalar Absolute<Scalar>(const float& val) {
+  return fabsf(val);
+}
+
 constexpr inline bool ScalarNearlyZero(Scalar x,
                                        Scalar tolerance = kEhCloseEnough) {
   return Absolute(x) <= tolerance;