diff --git a/README.chromium b/README.chromium
index 611f0d54c..963ecad44 100644
--- a/README.chromium
+++ b/README.chromium
@@ -1,6 +1,6 @@
 Name: libyuv
 URL: http://code.google.com/p/libyuv/
-Version: 538
+Version: 539
 License: BSD
 License File: LICENSE
 
diff --git a/include/libyuv/row.h b/include/libyuv/row.h
index 459c19f74..abb9e1efe 100644
--- a/include/libyuv/row.h
+++ b/include/libyuv/row.h
@@ -119,6 +119,7 @@ extern "C" {
 #if !defined(YUV_DISABLE_ASM) && defined(_M_IX86)
 #define HAS_ARGBCOLORTABLEROW_X86
 #define HAS_ARGBMULTIPLYROW_SSE2
+#define HAS_ARGBTOUV444ROW_SSSE3
 #endif
 
 // The following are Yasm x86 only.
@@ -492,6 +493,13 @@ void ARGB1555ToUVRow_C(const uint8* src_argb1555, int src_stride_argb1555,
 void ARGB4444ToUVRow_C(const uint8* src_argb4444, int src_stride_argb4444,
                        uint8* dst_u, uint8* dst_v, int width);
 
+void ARGBToUV444Row_SSSE3(const uint8* src_argb,
+                          uint8* dst_u, uint8* dst_v, int width);
+void ARGBToUV444Row_Unaligned_SSSE3(const uint8* src_argb,
+                                    uint8* dst_u, uint8* dst_v, int width);
+void ARGBToUV444Row_Any_SSSE3(const uint8* src_argb,
+                              uint8* dst_u, uint8* dst_v, int width);
+
 void ARGBToUV422Row_SSSE3(const uint8* src_argb,
                           uint8* dst_u, uint8* dst_v, int width);
 void ARGBToUV422Row_Unaligned_SSSE3(const uint8* src_argb,
diff --git a/include/libyuv/version.h b/include/libyuv/version.h
index 14e001081..8e4eb36f5 100644
--- a/include/libyuv/version.h
+++ b/include/libyuv/version.h
@@ -11,6 +11,6 @@
 #ifndef INCLUDE_LIBYUV_VERSION_H_  // NOLINT
 #define INCLUDE_LIBYUV_VERSION_H_
 
-#define LIBYUV_VERSION 538
+#define LIBYUV_VERSION 539
 
 #endif  // INCLUDE_LIBYUV_VERSION_H_  NOLINT
diff --git a/source/convert_from_argb.cc b/source/convert_from_argb.cc
index a9a287f47..7038f652b 100644
--- a/source/convert_from_argb.cc
+++ b/source/convert_from_argb.cc
@@ -1,875 +1,887 @@
-/*
- *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS.  All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-#include "libyuv/convert_from_argb.h"
-
-#include "libyuv/basic_types.h"
-#include "libyuv/cpu_id.h"
-#include "libyuv/format_conversion.h"
-#include "libyuv/planar_functions.h"
-#include "libyuv/row.h"
-
-#ifdef __cplusplus
-namespace libyuv {
-extern "C" {
-#endif
-
-// ARGB little endian (bgra in memory) to I444
-LIBYUV_API
-int ARGBToI444(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-  void (*ARGBToUV444Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV444Row_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    ARGBToUV444Row = ARGBToUV444Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-      ARGBToUV444Row = ARGBToUV444Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV444Row(src_argb, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-// ARGB little endian (bgra in memory) to I422
-LIBYUV_API
-int ARGBToI422(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV422Row_C;
-#if defined(HAS_ARGBTOUV422ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
-      }
-    }
-  }
-#endif
-
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
+/*
+ *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "libyuv/convert_from_argb.h"
+
+#include "libyuv/basic_types.h"
+#include "libyuv/cpu_id.h"
+#include "libyuv/format_conversion.h"
+#include "libyuv/planar_functions.h"
+#include "libyuv/row.h"
+
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
+
+// ARGB little endian (bgra in memory) to I444
+LIBYUV_API
+int ARGBToI444(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_y, int dst_stride_y,
+               uint8* dst_u, int dst_stride_u,
+               uint8* dst_v, int dst_stride_v,
+               int width, int height) {
+  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+  void (*ARGBToUV444Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
+                         int pix) = ARGBToUV444Row_C;
+#if defined(HAS_ARGBTOUV444ROW_SSSE3)
+    if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+      ARGBToUV444Row = ARGBToUV444Row_Any_SSSE3;
       if (IS_ALIGNED(width, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV422Row(src_argb, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-// ARGB little endian (bgra in memory) to I411
-LIBYUV_API
-int ARGBToI411(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_u, int dst_stride_u,
-               uint8* dst_v, int dst_stride_v,
-               int width, int height) {
-  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUV411Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV411Row_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 32) {
-      ARGBToUV411Row = ARGBToUV411Row_Any_NEON;
-      if (IS_ALIGNED(width, 32)) {
-        ARGBToUV411Row = ARGBToUV411Row_NEON;
-      }
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV411Row(src_argb, dst_u, dst_v, width);
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-    dst_u += dst_stride_u;
-    dst_v += dst_stride_v;
-  }
-  return 0;
-}
-
-LIBYUV_API
-int ARGBToNV12(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_uv, int dst_stride_uv,
-               int width, int height) {
-  if (!src_argb ||
-      !dst_y || !dst_uv ||
-      width <= 0 || height == 0 ||
-      width > kMaxStride) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUVRow = ARGBToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          ARGBToYRow = ARGBToYRow_SSSE3;
+        ARGBToUV444Row = ARGBToUV444Row_Unaligned_SSSE3;
+        if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+          ARGBToUV444Row = ARGBToUV444Row_SSSE3;
         }
       }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUVRow = ARGBToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUVRow = ARGBToUVRow_NEON;
-      }
-    }
   }
 #endif
-  int halfwidth = (width + 1) >> 1;
-  void (*MergeUVRow_)(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                      int width) = MergeUVRow_C;
-#if defined(HAS_MERGEUVROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_SSE2;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_uv, 16) && IS_ALIGNED(dst_stride_uv, 16)) {
-        MergeUVRow_ = MergeUVRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
-    MergeUVRow_ = MergeUVRow_Any_AVX2;
-    if (IS_ALIGNED(halfwidth, 32)) {
-      MergeUVRow_ = MergeUVRow_Unaligned_AVX2;
-      if (IS_ALIGNED(dst_uv, 32) && IS_ALIGNED(dst_stride_uv, 32)) {
-        MergeUVRow_ = MergeUVRow_AVX2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_NEON;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_NEON;
-    }
-  }
-#endif
-
-  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
-  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
-
-  for (int y = 0; y < height - 1; y += 2) {
-    ARGBToUVRow(src_argb, src_stride_argb, row_u, row_v, width);
-    MergeUVRow_(row_u, row_v, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-    ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
-    src_argb += src_stride_argb * 2;
-    dst_y += dst_stride_y * 2;
-    dst_uv += dst_stride_uv;
-  }
-  if (height & 1) {
-    ARGBToUVRow(src_argb, 0, row_u, row_v, width);
-    MergeUVRow_(row_u, row_v, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-    ARGBToYRow(src_argb + 0, dst_y + dst_stride_y, width);
-  }
-  return 0;
-}
-
-// Same as NV12 but U and V swapped.
-LIBYUV_API
-int ARGBToNV21(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               uint8* dst_uv, int dst_stride_uv,
-               int width, int height) {
-  if (!src_argb ||
-      !dst_y || !dst_uv ||
-      width <= 0 || height == 0 ||
-      width > kMaxStride) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
-                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUVRow = ARGBToUVRow_SSSE3;
-        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-          ARGBToYRow = ARGBToYRow_SSSE3;
-        }
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUVRow = ARGBToUVRow_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUVRow = ARGBToUVRow_NEON;
-      }
-    }
-  }
-#endif
-  int halfwidth = (width + 1) >> 1;
-  void (*MergeUVRow_)(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
-                      int width) = MergeUVRow_C;
-#if defined(HAS_MERGEUVROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_SSE2;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_Unaligned_SSE2;
-      if (IS_ALIGNED(dst_uv, 16) && IS_ALIGNED(dst_stride_uv, 16)) {
-        MergeUVRow_ = MergeUVRow_SSE2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_AVX2)
-  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
-    MergeUVRow_ = MergeUVRow_Any_AVX2;
-    if (IS_ALIGNED(halfwidth, 32)) {
-      MergeUVRow_ = MergeUVRow_Unaligned_AVX2;
-      if (IS_ALIGNED(dst_uv, 32) && IS_ALIGNED(dst_stride_uv, 32)) {
-        MergeUVRow_ = MergeUVRow_AVX2;
-      }
-    }
-  }
-#endif
-#if defined(HAS_MERGEUVROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
-    MergeUVRow_ = MergeUVRow_Any_NEON;
-    if (IS_ALIGNED(halfwidth, 16)) {
-      MergeUVRow_ = MergeUVRow_NEON;
-    }
-  }
-#endif
-
-  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
-  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
-
-  for (int y = 0; y < height - 1; y += 2) {
-    ARGBToUVRow(src_argb, src_stride_argb, row_u, row_v, width);
-    MergeUVRow_(row_v, row_u, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-    ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
-    src_argb += src_stride_argb * 2;
-    dst_y += dst_stride_y * 2;
-    dst_uv += dst_stride_uv;
-  }
-  if (height & 1) {
-    ARGBToUVRow(src_argb, 0, row_u, row_v, width);
-    MergeUVRow_(row_v, row_u, dst_uv, halfwidth);
-    ARGBToYRow(src_argb, dst_y, width);
-    ARGBToYRow(src_argb + 0, dst_y + dst_stride_y, width);
-  }
-  return 0;
-}
-
-// Convert ARGB to YUY2.
-LIBYUV_API
-int ARGBToYUY2(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_yuy2, int dst_stride_yuy2,
-               int width, int height) {
-  if (!src_argb || !dst_yuy2 ||
-      width <= 0 || height == 0 ||
-      width > kMaxStride) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_yuy2 = dst_yuy2 + (height - 1) * dst_stride_yuy2;
-    dst_stride_yuy2 = -dst_stride_yuy2;
-  }
-  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV422Row_C;
-#if defined(HAS_ARGBTOUV422ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
-      }
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_NEON;
-      }
-    }
-  }
-#endif
-
-  void (*I422ToYUY2Row)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_yuy2, int width) =
-      I422ToYUY2Row_C;
-#if defined(HAS_I422TOYUY2ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16 &&
-      IS_ALIGNED(dst_yuy2, 16) && IS_ALIGNED(dst_stride_yuy2, 16)) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOYUY2ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToYUY2Row = I422ToYUY2Row_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToYUY2Row = I422ToYUY2Row_NEON;
-    }
-  }
-#endif
-  SIMD_ALIGNED(uint8 row_y[kMaxStride]);
-  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
-  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV422Row(src_argb, row_u, row_v, width);
-    ARGBToYRow(src_argb, row_y, width);
-    I422ToYUY2Row(row_y, row_u, row_v, dst_yuy2, width);
-    src_argb += src_stride_argb;
-    dst_yuy2 += dst_stride_yuy2;
-  }
-  return 0;
-}
-
-// Convert ARGB to UYVY.
-LIBYUV_API
-int ARGBToUYVY(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_uyvy, int dst_stride_uyvy,
-               int width, int height) {
-  if (!src_argb || !dst_uyvy ||
-      width <= 0 || height == 0 ||
-      width > kMaxStride) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    dst_uyvy = dst_uyvy + (height - 1) * dst_stride_uyvy;
-    dst_stride_uyvy = -dst_stride_uyvy;
-  }
-  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
-                         int pix) = ARGBToUV422Row_C;
-#if defined(HAS_ARGBTOUV422ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
-      }
-    }
-  }
-#endif
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-    if (width >= 16) {
-      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
-      if (IS_ALIGNED(width, 16)) {
-        ARGBToUV422Row = ARGBToUV422Row_NEON;
-      }
-    }
-  }
-#endif
-
-  void (*I422ToUYVYRow)(const uint8* src_y, const uint8* src_u,
-                        const uint8* src_v, uint8* dst_uyvy, int width) =
-      I422ToUYVYRow_C;
-#if defined(HAS_I422TOUYVYROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 16 &&
-      IS_ALIGNED(dst_uyvy, 16) && IS_ALIGNED(dst_stride_uyvy, 16)) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_SSE2;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_SSE2;
-    }
-  }
-#elif defined(HAS_I422TOUYVYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
-    I422ToUYVYRow = I422ToUYVYRow_Any_NEON;
-    if (IS_ALIGNED(width, 16)) {
-      I422ToUYVYRow = I422ToUYVYRow_NEON;
-    }
-  }
-#endif
-  SIMD_ALIGNED(uint8 row_y[kMaxStride]);
-  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
-  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToUV422Row(src_argb, row_u, row_v, width);
-    ARGBToYRow(src_argb, row_y, width);
-    I422ToUYVYRow(row_y, row_u, row_v, dst_uyvy, width);
-    src_argb += src_stride_argb;
-    dst_uyvy += dst_stride_uyvy;
-  }
-  return 0;
-}
-
-// Convert ARGB to I400.
-LIBYUV_API
-int ARGBToI400(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_y, int dst_stride_y,
-               int width, int height) {
-  if (!src_argb || !dst_y || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
-      ARGBToYRow_C;
-#if defined(HAS_ARGBTOYROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
-    ARGBToYRow = ARGBToYRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
-      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
-        ARGBToYRow = ARGBToYRow_SSSE3;
-      }
-    }
-  }
-#elif defined(HAS_ARGBTOYROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToYRow = ARGBToYRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToYRow = ARGBToYRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToYRow(src_argb, dst_y, width);
-    src_argb += src_stride_argb;
-    dst_y += dst_stride_y;
-  }
-  return 0;
-}
-
-// Convert ARGB to RGBA.
-LIBYUV_API
-int ARGBToRGBA(const uint8* src_argb, int src_stride_argb,
-               uint8* dst_rgba, int dst_stride_rgba,
-               int width, int height) {
-  if (!src_argb || !dst_rgba ||
-      width <= 0 || height == 0) {
-    return -1;
-  }
-  // Negative height means invert the image.
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToRGBARow)(const uint8* src_argb, uint8* dst_rgba, int pix) =
-      ARGBToRGBARow_C;
-#if defined(HAS_ARGBTORGBAROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) &&
-      IS_ALIGNED(width, 4) &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_rgba, 16) && IS_ALIGNED(dst_stride_rgba, 16)) {
-    ARGBToRGBARow = ARGBToRGBARow_SSSE3;
-  }
-#elif defined(HAS_ARGBTORGBAROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
-    ARGBToRGBARow = ARGBToRGBARow_NEON;
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToRGBARow(src_argb, dst_rgba, width);
-    src_argb += src_stride_argb;
-    dst_rgba += dst_stride_rgba;
-  }
-  return 0;
-}
-
-// Convert ARGB To RGB24.
-LIBYUV_API
-int ARGBToRGB24(const uint8* src_argb, int src_stride_argb,
-                uint8* dst_rgb24, int dst_stride_rgb24,
-                int width, int height) {
-  if (!src_argb || !dst_rgb24 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToRGB24Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToRGB24Row_C;
-#if defined(HAS_ARGBTORGB24ROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_rgb24, 16) && IS_ALIGNED(dst_stride_rgb24, 16)) {
-    ARGBToRGB24Row = ARGBToRGB24Row_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToRGB24Row = ARGBToRGB24Row_SSSE3;
-    }
-  }
-#elif defined(HAS_ARGBTORGB24ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToRGB24Row = ARGBToRGB24Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToRGB24Row = ARGBToRGB24Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToRGB24Row(src_argb, dst_rgb24, width);
-    src_argb += src_stride_argb;
-    dst_rgb24 += dst_stride_rgb24;
-  }
-  return 0;
-}
-
-// Convert ARGB To RAW.
-LIBYUV_API
-int ARGBToRAW(const uint8* src_argb, int src_stride_argb,
-              uint8* dst_raw, int dst_stride_raw,
-              int width, int height) {
-  if (!src_argb || !dst_raw || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToRAWRow)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToRAWRow_C;
-#if defined(HAS_ARGBTORAWROW_SSSE3)
-  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
-      IS_ALIGNED(dst_raw, 16) && IS_ALIGNED(dst_stride_raw, 16)) {
-    ARGBToRAWRow = ARGBToRAWRow_Any_SSSE3;
-    if (IS_ALIGNED(width, 16)) {
-      ARGBToRAWRow = ARGBToRAWRow_SSSE3;
-    }
-  }
-#elif defined(HAS_ARGBTORAWROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToRAWRow = ARGBToRAWRow_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToRAWRow = ARGBToRAWRow_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToRAWRow(src_argb, dst_raw, width);
-    src_argb += src_stride_argb;
-    dst_raw += dst_stride_raw;
-  }
-  return 0;
-}
-
-// Convert ARGB To RGB565.
-LIBYUV_API
-int ARGBToRGB565(const uint8* src_argb, int src_stride_argb,
-                 uint8* dst_rgb565, int dst_stride_rgb565,
-                 int width, int height) {
-  if (!src_argb || !dst_rgb565 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToRGB565Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToRGB565Row_C;
-#if defined(HAS_ARGBTORGB565ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToRGB565Row = ARGBToRGB565Row_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBToRGB565Row = ARGBToRGB565Row_SSE2;
-    }
-  }
-#elif defined(HAS_ARGBTORGB565ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToRGB565Row = ARGBToRGB565Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToRGB565Row = ARGBToRGB565Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToRGB565Row(src_argb, dst_rgb565, width);
-    src_argb += src_stride_argb;
-    dst_rgb565 += dst_stride_rgb565;
-  }
-  return 0;
-}
-
-// Convert ARGB To ARGB1555.
-LIBYUV_API
-int ARGBToARGB1555(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb1555, int dst_stride_argb1555,
-                   int width, int height) {
-  if (!src_argb || !dst_argb1555 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToARGB1555Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToARGB1555Row_C;
-#if defined(HAS_ARGBTOARGB1555ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToARGB1555Row = ARGBToARGB1555Row_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBToARGB1555Row = ARGBToARGB1555Row_SSE2;
-    }
-  }
-#elif defined(HAS_ARGBTOARGB1555ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToARGB1555Row = ARGBToARGB1555Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToARGB1555Row = ARGBToARGB1555Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToARGB1555Row(src_argb, dst_argb1555, width);
-    src_argb += src_stride_argb;
-    dst_argb1555 += dst_stride_argb1555;
-  }
-  return 0;
-}
-
-// Convert ARGB To ARGB4444.
-LIBYUV_API
-int ARGBToARGB4444(const uint8* src_argb, int src_stride_argb,
-                   uint8* dst_argb4444, int dst_stride_argb4444,
-                   int width, int height) {
-  if (!src_argb || !dst_argb4444 || width <= 0 || height == 0) {
-    return -1;
-  }
-  if (height < 0) {
-    height = -height;
-    src_argb = src_argb + (height - 1) * src_stride_argb;
-    src_stride_argb = -src_stride_argb;
-  }
-  void (*ARGBToARGB4444Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
-      ARGBToARGB4444Row_C;
-#if defined(HAS_ARGBTOARGB4444ROW_SSE2)
-  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
-      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
-    ARGBToARGB4444Row = ARGBToARGB4444Row_Any_SSE2;
-    if (IS_ALIGNED(width, 4)) {
-      ARGBToARGB4444Row = ARGBToARGB4444Row_SSE2;
-    }
-  }
-#elif defined(HAS_ARGBTOARGB4444ROW_NEON)
-  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
-    ARGBToARGB4444Row = ARGBToARGB4444Row_Any_NEON;
-    if (IS_ALIGNED(width, 8)) {
-      ARGBToARGB4444Row = ARGBToARGB4444Row_NEON;
-    }
-  }
-#endif
-
-  for (int y = 0; y < height; ++y) {
-    ARGBToARGB4444Row(src_argb, dst_argb4444, width);
-    src_argb += src_stride_argb;
-    dst_argb4444 += dst_stride_argb4444;
-  }
-  return 0;
-}
-
-#ifdef __cplusplus
-}  // extern "C"
-}  // namespace libyuv
-#endif
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
+          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
+        ARGBToYRow = ARGBToYRow_SSSE3;
+      }
+    }
+  }
+
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    ARGBToUV444Row = ARGBToUV444Row_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+      ARGBToUV444Row = ARGBToUV444Row_NEON;
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToUV444Row(src_argb, dst_u, dst_v, width);
+    ARGBToYRow(src_argb, dst_y, width);
+    src_argb += src_stride_argb;
+    dst_y += dst_stride_y;
+    dst_u += dst_stride_u;
+    dst_v += dst_stride_v;
+  }
+  return 0;
+}
+
+// ARGB little endian (bgra in memory) to I422
+LIBYUV_API
+int ARGBToI422(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_y, int dst_stride_y,
+               uint8* dst_u, int dst_stride_u,
+               uint8* dst_v, int dst_stride_v,
+               int width, int height) {
+  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
+                         int pix) = ARGBToUV422Row_C;
+#if defined(HAS_ARGBTOUV422ROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
+      }
+    }
+  }
+#endif
+
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
+          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
+        ARGBToYRow = ARGBToYRow_SSSE3;
+      }
+    }
+  }
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+    }
+    if (width >= 16) {
+      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
+      if (IS_ALIGNED(width, 16)) {
+        ARGBToUV422Row = ARGBToUV422Row_NEON;
+      }
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToUV422Row(src_argb, dst_u, dst_v, width);
+    ARGBToYRow(src_argb, dst_y, width);
+    src_argb += src_stride_argb;
+    dst_y += dst_stride_y;
+    dst_u += dst_stride_u;
+    dst_v += dst_stride_v;
+  }
+  return 0;
+}
+
+// ARGB little endian (bgra in memory) to I411
+LIBYUV_API
+int ARGBToI411(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_y, int dst_stride_y,
+               uint8* dst_u, int dst_stride_u,
+               uint8* dst_v, int dst_stride_v,
+               int width, int height) {
+  if (!src_argb || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToUV411Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
+                         int pix) = ARGBToUV411Row_C;
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
+          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
+        ARGBToYRow = ARGBToYRow_SSSE3;
+      }
+    }
+  }
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+    }
+    if (width >= 32) {
+      ARGBToUV411Row = ARGBToUV411Row_Any_NEON;
+      if (IS_ALIGNED(width, 32)) {
+        ARGBToUV411Row = ARGBToUV411Row_NEON;
+      }
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToUV411Row(src_argb, dst_u, dst_v, width);
+    ARGBToYRow(src_argb, dst_y, width);
+    src_argb += src_stride_argb;
+    dst_y += dst_stride_y;
+    dst_u += dst_stride_u;
+    dst_v += dst_stride_v;
+  }
+  return 0;
+}
+
+LIBYUV_API
+int ARGBToNV12(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_y, int dst_stride_y,
+               uint8* dst_uv, int dst_stride_uv,
+               int width, int height) {
+  if (!src_argb ||
+      !dst_y || !dst_uv ||
+      width <= 0 || height == 0 ||
+      width > kMaxStride) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
+                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+        ARGBToUVRow = ARGBToUVRow_SSSE3;
+        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
+          ARGBToYRow = ARGBToYRow_SSSE3;
+        }
+      }
+    }
+  }
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+    }
+    if (width >= 16) {
+      ARGBToUVRow = ARGBToUVRow_Any_NEON;
+      if (IS_ALIGNED(width, 16)) {
+        ARGBToUVRow = ARGBToUVRow_NEON;
+      }
+    }
+  }
+#endif
+  int halfwidth = (width + 1) >> 1;
+  void (*MergeUVRow_)(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
+                      int width) = MergeUVRow_C;
+#if defined(HAS_MERGEUVROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
+    MergeUVRow_ = MergeUVRow_Any_SSE2;
+    if (IS_ALIGNED(halfwidth, 16)) {
+      MergeUVRow_ = MergeUVRow_Unaligned_SSE2;
+      if (IS_ALIGNED(dst_uv, 16) && IS_ALIGNED(dst_stride_uv, 16)) {
+        MergeUVRow_ = MergeUVRow_SSE2;
+      }
+    }
+  }
+#endif
+#if defined(HAS_MERGEUVROW_AVX2)
+  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
+    MergeUVRow_ = MergeUVRow_Any_AVX2;
+    if (IS_ALIGNED(halfwidth, 32)) {
+      MergeUVRow_ = MergeUVRow_Unaligned_AVX2;
+      if (IS_ALIGNED(dst_uv, 32) && IS_ALIGNED(dst_stride_uv, 32)) {
+        MergeUVRow_ = MergeUVRow_AVX2;
+      }
+    }
+  }
+#endif
+#if defined(HAS_MERGEUVROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
+    MergeUVRow_ = MergeUVRow_Any_NEON;
+    if (IS_ALIGNED(halfwidth, 16)) {
+      MergeUVRow_ = MergeUVRow_NEON;
+    }
+  }
+#endif
+
+  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
+  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
+
+  for (int y = 0; y < height - 1; y += 2) {
+    ARGBToUVRow(src_argb, src_stride_argb, row_u, row_v, width);
+    MergeUVRow_(row_u, row_v, dst_uv, halfwidth);
+    ARGBToYRow(src_argb, dst_y, width);
+    ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
+    src_argb += src_stride_argb * 2;
+    dst_y += dst_stride_y * 2;
+    dst_uv += dst_stride_uv;
+  }
+  if (height & 1) {
+    ARGBToUVRow(src_argb, 0, row_u, row_v, width);
+    MergeUVRow_(row_u, row_v, dst_uv, halfwidth);
+    ARGBToYRow(src_argb, dst_y, width);
+    ARGBToYRow(src_argb + 0, dst_y + dst_stride_y, width);
+  }
+  return 0;
+}
+
+// Same as NV12 but U and V swapped.
+LIBYUV_API
+int ARGBToNV21(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_y, int dst_stride_y,
+               uint8* dst_uv, int dst_stride_uv,
+               int width, int height) {
+  if (!src_argb ||
+      !dst_y || !dst_uv ||
+      width <= 0 || height == 0 ||
+      width > kMaxStride) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
+                      uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+        ARGBToUVRow = ARGBToUVRow_SSSE3;
+        if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
+          ARGBToYRow = ARGBToYRow_SSSE3;
+        }
+      }
+    }
+  }
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+    }
+    if (width >= 16) {
+      ARGBToUVRow = ARGBToUVRow_Any_NEON;
+      if (IS_ALIGNED(width, 16)) {
+        ARGBToUVRow = ARGBToUVRow_NEON;
+      }
+    }
+  }
+#endif
+  int halfwidth = (width + 1) >> 1;
+  void (*MergeUVRow_)(const uint8* src_u, const uint8* src_v, uint8* dst_uv,
+                      int width) = MergeUVRow_C;
+#if defined(HAS_MERGEUVROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2) && halfwidth >= 16) {
+    MergeUVRow_ = MergeUVRow_Any_SSE2;
+    if (IS_ALIGNED(halfwidth, 16)) {
+      MergeUVRow_ = MergeUVRow_Unaligned_SSE2;
+      if (IS_ALIGNED(dst_uv, 16) && IS_ALIGNED(dst_stride_uv, 16)) {
+        MergeUVRow_ = MergeUVRow_SSE2;
+      }
+    }
+  }
+#endif
+#if defined(HAS_MERGEUVROW_AVX2)
+  if (TestCpuFlag(kCpuHasAVX2) && halfwidth >= 32) {
+    MergeUVRow_ = MergeUVRow_Any_AVX2;
+    if (IS_ALIGNED(halfwidth, 32)) {
+      MergeUVRow_ = MergeUVRow_Unaligned_AVX2;
+      if (IS_ALIGNED(dst_uv, 32) && IS_ALIGNED(dst_stride_uv, 32)) {
+        MergeUVRow_ = MergeUVRow_AVX2;
+      }
+    }
+  }
+#endif
+#if defined(HAS_MERGEUVROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && halfwidth >= 16) {
+    MergeUVRow_ = MergeUVRow_Any_NEON;
+    if (IS_ALIGNED(halfwidth, 16)) {
+      MergeUVRow_ = MergeUVRow_NEON;
+    }
+  }
+#endif
+
+  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
+  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
+
+  for (int y = 0; y < height - 1; y += 2) {
+    ARGBToUVRow(src_argb, src_stride_argb, row_u, row_v, width);
+    MergeUVRow_(row_v, row_u, dst_uv, halfwidth);
+    ARGBToYRow(src_argb, dst_y, width);
+    ARGBToYRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width);
+    src_argb += src_stride_argb * 2;
+    dst_y += dst_stride_y * 2;
+    dst_uv += dst_stride_uv;
+  }
+  if (height & 1) {
+    ARGBToUVRow(src_argb, 0, row_u, row_v, width);
+    MergeUVRow_(row_v, row_u, dst_uv, halfwidth);
+    ARGBToYRow(src_argb, dst_y, width);
+    ARGBToYRow(src_argb + 0, dst_y + dst_stride_y, width);
+  }
+  return 0;
+}
+
+// Convert ARGB to YUY2.
+LIBYUV_API
+int ARGBToYUY2(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_yuy2, int dst_stride_yuy2,
+               int width, int height) {
+  if (!src_argb || !dst_yuy2 ||
+      width <= 0 || height == 0 ||
+      width > kMaxStride) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    dst_yuy2 = dst_yuy2 + (height - 1) * dst_stride_yuy2;
+    dst_stride_yuy2 = -dst_stride_yuy2;
+  }
+  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
+                         int pix) = ARGBToUV422Row_C;
+#if defined(HAS_ARGBTOUV422ROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
+      }
+    }
+  }
+#endif
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+        ARGBToYRow = ARGBToYRow_SSSE3;
+      }
+    }
+  }
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+    }
+    if (width >= 16) {
+      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
+      if (IS_ALIGNED(width, 16)) {
+        ARGBToUV422Row = ARGBToUV422Row_NEON;
+      }
+    }
+  }
+#endif
+
+  void (*I422ToYUY2Row)(const uint8* src_y, const uint8* src_u,
+                        const uint8* src_v, uint8* dst_yuy2, int width) =
+      I422ToYUY2Row_C;
+#if defined(HAS_I422TOYUY2ROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2) && width >= 16 &&
+      IS_ALIGNED(dst_yuy2, 16) && IS_ALIGNED(dst_stride_yuy2, 16)) {
+    I422ToYUY2Row = I422ToYUY2Row_Any_SSE2;
+    if (IS_ALIGNED(width, 16)) {
+      I422ToYUY2Row = I422ToYUY2Row_SSE2;
+    }
+  }
+#elif defined(HAS_I422TOYUY2ROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
+    I422ToYUY2Row = I422ToYUY2Row_Any_NEON;
+    if (IS_ALIGNED(width, 16)) {
+      I422ToYUY2Row = I422ToYUY2Row_NEON;
+    }
+  }
+#endif
+  SIMD_ALIGNED(uint8 row_y[kMaxStride]);
+  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
+  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToUV422Row(src_argb, row_u, row_v, width);
+    ARGBToYRow(src_argb, row_y, width);
+    I422ToYUY2Row(row_y, row_u, row_v, dst_yuy2, width);
+    src_argb += src_stride_argb;
+    dst_yuy2 += dst_stride_yuy2;
+  }
+  return 0;
+}
+
+// Convert ARGB to UYVY.
+LIBYUV_API
+int ARGBToUYVY(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_uyvy, int dst_stride_uyvy,
+               int width, int height) {
+  if (!src_argb || !dst_uyvy ||
+      width <= 0 || height == 0 ||
+      width > kMaxStride) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    dst_uyvy = dst_uyvy + (height - 1) * dst_stride_uyvy;
+    dst_stride_uyvy = -dst_stride_uyvy;
+  }
+  void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
+                         int pix) = ARGBToUV422Row_C;
+#if defined(HAS_ARGBTOUV422ROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+        ARGBToUV422Row = ARGBToUV422Row_SSSE3;
+      }
+    }
+  }
+#endif
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+        ARGBToYRow = ARGBToYRow_SSSE3;
+      }
+    }
+  }
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+    }
+    if (width >= 16) {
+      ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
+      if (IS_ALIGNED(width, 16)) {
+        ARGBToUV422Row = ARGBToUV422Row_NEON;
+      }
+    }
+  }
+#endif
+
+  void (*I422ToUYVYRow)(const uint8* src_y, const uint8* src_u,
+                        const uint8* src_v, uint8* dst_uyvy, int width) =
+      I422ToUYVYRow_C;
+#if defined(HAS_I422TOUYVYROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2) && width >= 16 &&
+      IS_ALIGNED(dst_uyvy, 16) && IS_ALIGNED(dst_stride_uyvy, 16)) {
+    I422ToUYVYRow = I422ToUYVYRow_Any_SSE2;
+    if (IS_ALIGNED(width, 16)) {
+      I422ToUYVYRow = I422ToUYVYRow_SSE2;
+    }
+  }
+#elif defined(HAS_I422TOUYVYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 16) {
+    I422ToUYVYRow = I422ToUYVYRow_Any_NEON;
+    if (IS_ALIGNED(width, 16)) {
+      I422ToUYVYRow = I422ToUYVYRow_NEON;
+    }
+  }
+#endif
+  SIMD_ALIGNED(uint8 row_y[kMaxStride]);
+  SIMD_ALIGNED(uint8 row_u[kMaxStride / 2]);
+  SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToUV422Row(src_argb, row_u, row_v, width);
+    ARGBToYRow(src_argb, row_y, width);
+    I422ToUYVYRow(row_y, row_u, row_v, dst_uyvy, width);
+    src_argb += src_stride_argb;
+    dst_uyvy += dst_stride_uyvy;
+  }
+  return 0;
+}
+
+// Convert ARGB to I400.
+LIBYUV_API
+int ARGBToI400(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_y, int dst_stride_y,
+               int width, int height) {
+  if (!src_argb || !dst_y || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
+      ARGBToYRow_C;
+#if defined(HAS_ARGBTOYROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
+    ARGBToYRow = ARGBToYRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
+      if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
+          IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
+        ARGBToYRow = ARGBToYRow_SSSE3;
+      }
+    }
+  }
+#elif defined(HAS_ARGBTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToYRow = ARGBToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToYRow = ARGBToYRow_NEON;
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToYRow(src_argb, dst_y, width);
+    src_argb += src_stride_argb;
+    dst_y += dst_stride_y;
+  }
+  return 0;
+}
+
+// Convert ARGB to RGBA.
+LIBYUV_API
+int ARGBToRGBA(const uint8* src_argb, int src_stride_argb,
+               uint8* dst_rgba, int dst_stride_rgba,
+               int width, int height) {
+  if (!src_argb || !dst_rgba ||
+      width <= 0 || height == 0) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToRGBARow)(const uint8* src_argb, uint8* dst_rgba, int pix) =
+      ARGBToRGBARow_C;
+#if defined(HAS_ARGBTORGBAROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) &&
+      IS_ALIGNED(width, 4) &&
+      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
+      IS_ALIGNED(dst_rgba, 16) && IS_ALIGNED(dst_stride_rgba, 16)) {
+    ARGBToRGBARow = ARGBToRGBARow_SSSE3;
+  }
+#elif defined(HAS_ARGBTORGBAROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && IS_ALIGNED(width, 8)) {
+    ARGBToRGBARow = ARGBToRGBARow_NEON;
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToRGBARow(src_argb, dst_rgba, width);
+    src_argb += src_stride_argb;
+    dst_rgba += dst_stride_rgba;
+  }
+  return 0;
+}
+
+// Convert ARGB To RGB24.
+LIBYUV_API
+int ARGBToRGB24(const uint8* src_argb, int src_stride_argb,
+                uint8* dst_rgb24, int dst_stride_rgb24,
+                int width, int height) {
+  if (!src_argb || !dst_rgb24 || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToRGB24Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
+      ARGBToRGB24Row_C;
+#if defined(HAS_ARGBTORGB24ROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16 &&
+      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
+      IS_ALIGNED(dst_rgb24, 16) && IS_ALIGNED(dst_stride_rgb24, 16)) {
+    ARGBToRGB24Row = ARGBToRGB24Row_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToRGB24Row = ARGBToRGB24Row_SSSE3;
+    }
+  }
+#elif defined(HAS_ARGBTORGB24ROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToRGB24Row = ARGBToRGB24Row_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToRGB24Row = ARGBToRGB24Row_NEON;
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToRGB24Row(src_argb, dst_rgb24, width);
+    src_argb += src_stride_argb;
+    dst_rgb24 += dst_stride_rgb24;
+  }
+  return 0;
+}
+
+// Convert ARGB To RAW.
+LIBYUV_API
+int ARGBToRAW(const uint8* src_argb, int src_stride_argb,
+              uint8* dst_raw, int dst_stride_raw,
+              int width, int height) {
+  if (!src_argb || !dst_raw || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToRAWRow)(const uint8* src_argb, uint8* dst_rgb, int pix) =
+      ARGBToRAWRow_C;
+#if defined(HAS_ARGBTORAWROW_SSSE3)
+  if (TestCpuFlag(kCpuHasSSSE3) && width >= 16 &&
+      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
+      IS_ALIGNED(dst_raw, 16) && IS_ALIGNED(dst_stride_raw, 16)) {
+    ARGBToRAWRow = ARGBToRAWRow_Any_SSSE3;
+    if (IS_ALIGNED(width, 16)) {
+      ARGBToRAWRow = ARGBToRAWRow_SSSE3;
+    }
+  }
+#elif defined(HAS_ARGBTORAWROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToRAWRow = ARGBToRAWRow_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToRAWRow = ARGBToRAWRow_NEON;
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToRAWRow(src_argb, dst_raw, width);
+    src_argb += src_stride_argb;
+    dst_raw += dst_stride_raw;
+  }
+  return 0;
+}
+
+// Convert ARGB To RGB565.
+LIBYUV_API
+int ARGBToRGB565(const uint8* src_argb, int src_stride_argb,
+                 uint8* dst_rgb565, int dst_stride_rgb565,
+                 int width, int height) {
+  if (!src_argb || !dst_rgb565 || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToRGB565Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
+      ARGBToRGB565Row_C;
+#if defined(HAS_ARGBTORGB565ROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
+      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+    ARGBToRGB565Row = ARGBToRGB565Row_Any_SSE2;
+    if (IS_ALIGNED(width, 4)) {
+      ARGBToRGB565Row = ARGBToRGB565Row_SSE2;
+    }
+  }
+#elif defined(HAS_ARGBTORGB565ROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToRGB565Row = ARGBToRGB565Row_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToRGB565Row = ARGBToRGB565Row_NEON;
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToRGB565Row(src_argb, dst_rgb565, width);
+    src_argb += src_stride_argb;
+    dst_rgb565 += dst_stride_rgb565;
+  }
+  return 0;
+}
+
+// Convert ARGB To ARGB1555.
+LIBYUV_API
+int ARGBToARGB1555(const uint8* src_argb, int src_stride_argb,
+                   uint8* dst_argb1555, int dst_stride_argb1555,
+                   int width, int height) {
+  if (!src_argb || !dst_argb1555 || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToARGB1555Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
+      ARGBToARGB1555Row_C;
+#if defined(HAS_ARGBTOARGB1555ROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
+      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+    ARGBToARGB1555Row = ARGBToARGB1555Row_Any_SSE2;
+    if (IS_ALIGNED(width, 4)) {
+      ARGBToARGB1555Row = ARGBToARGB1555Row_SSE2;
+    }
+  }
+#elif defined(HAS_ARGBTOARGB1555ROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToARGB1555Row = ARGBToARGB1555Row_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToARGB1555Row = ARGBToARGB1555Row_NEON;
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToARGB1555Row(src_argb, dst_argb1555, width);
+    src_argb += src_stride_argb;
+    dst_argb1555 += dst_stride_argb1555;
+  }
+  return 0;
+}
+
+// Convert ARGB To ARGB4444.
+LIBYUV_API
+int ARGBToARGB4444(const uint8* src_argb, int src_stride_argb,
+                   uint8* dst_argb4444, int dst_stride_argb4444,
+                   int width, int height) {
+  if (!src_argb || !dst_argb4444 || width <= 0 || height == 0) {
+    return -1;
+  }
+  if (height < 0) {
+    height = -height;
+    src_argb = src_argb + (height - 1) * src_stride_argb;
+    src_stride_argb = -src_stride_argb;
+  }
+  void (*ARGBToARGB4444Row)(const uint8* src_argb, uint8* dst_rgb, int pix) =
+      ARGBToARGB4444Row_C;
+#if defined(HAS_ARGBTOARGB4444ROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2) && width >= 4 &&
+      IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
+    ARGBToARGB4444Row = ARGBToARGB4444Row_Any_SSE2;
+    if (IS_ALIGNED(width, 4)) {
+      ARGBToARGB4444Row = ARGBToARGB4444Row_SSE2;
+    }
+  }
+#elif defined(HAS_ARGBTOARGB4444ROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
+    ARGBToARGB4444Row = ARGBToARGB4444Row_Any_NEON;
+    if (IS_ALIGNED(width, 8)) {
+      ARGBToARGB4444Row = ARGBToARGB4444Row_NEON;
+    }
+  }
+#endif
+
+  for (int y = 0; y < height; ++y) {
+    ARGBToARGB4444Row(src_argb, dst_argb4444, width);
+    src_argb += src_stride_argb;
+    dst_argb4444 += dst_stride_argb4444;
+  }
+  return 0;
+}
+
+#ifdef __cplusplus
+}  // extern "C"
+}  // namespace libyuv
+#endif
diff --git a/source/row_any.cc b/source/row_any.cc
index ce9f352c1..1f1df54c1 100644
--- a/source/row_any.cc
+++ b/source/row_any.cc
@@ -287,6 +287,23 @@ UVANY(UYVYToUVRow_Any_NEON, UYVYToUVRow_NEON, UYVYToUVRow_C, 2)
 #endif
 #undef UVANY
 
+#define UV444ANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, BPP, MASK)                  \
+    void NAMEANY(const uint8* src_uv,                                          \
+                 uint8* dst_u, uint8* dst_v, int width) {                      \
+      int n = width & ~MASK;                                                   \
+      ANYTOUV_SIMD(src_uv, dst_u, dst_v, n);                                   \
+      ANYTOUV_C(src_uv  + n * BPP,                                             \
+                dst_u + n,                                                     \
+                dst_v + n,                                                     \
+                width & MASK);                                                 \
+    }
+
+#ifdef HAS_ARGBTOUV444ROW_SSSE3
+UV444ANY(ARGBToUV444Row_Any_SSSE3, ARGBToUV444Row_Unaligned_SSSE3,
+         ARGBToUV444Row_C, 4, 15)
+#endif
+#undef UV444ANY
+
 #define UV422ANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, BPP, MASK, SHIFT)           \
     void NAMEANY(const uint8* src_uv,                                          \
                  uint8* dst_u, uint8* dst_v, int width) {                      \
diff --git a/source/row_win.cc b/source/row_win.cc
index 82d77027a..f988312ed 100644
--- a/source/row_win.cc
+++ b/source/row_win.cc
@@ -1101,6 +1101,122 @@ __asm {
   }
 }
 
+__declspec(naked) __declspec(align(16))
+void ARGBToUV444Row_SSSE3(const uint8* src_argb0,
+                          uint8* dst_u, uint8* dst_v, int width) {
+__asm {
+    push       edi
+    mov        eax, [esp + 4 + 4]   // src_argb
+    mov        edx, [esp + 4 + 8]   // dst_u
+    mov        edi, [esp + 4 + 12]  // dst_v
+    mov        ecx, [esp + 4 + 16]  // pix
+    movdqa     xmm7, kARGBToU
+    movdqa     xmm6, kARGBToV
+    movdqa     xmm5, kAddUV128
+    sub        edi, edx             // stride from u to v
+
+    align      16
+ convertloop:
+    /* convert to U and V */
+    movdqa     xmm0, [eax]          // U
+    movdqa     xmm1, [eax + 16]
+    movdqa     xmm2, [eax + 32]
+    movdqa     xmm3, [eax + 48]
+    pmaddubsw  xmm0, xmm7
+    pmaddubsw  xmm1, xmm7
+    pmaddubsw  xmm2, xmm7
+    pmaddubsw  xmm3, xmm7
+    phaddw     xmm0, xmm1
+    phaddw     xmm2, xmm3
+    psrlw      xmm0, 8
+    psrlw      xmm2, 8
+    packuswb   xmm0, xmm2
+    paddb      xmm0, xmm5
+    sub        ecx,  16
+    movdqa     [edx], xmm0
+
+    movdqa     xmm0, [eax]          // V
+    movdqa     xmm1, [eax + 16]
+    movdqa     xmm2, [eax + 32]
+    movdqa     xmm3, [eax + 48]
+    pmaddubsw  xmm0, xmm6
+    pmaddubsw  xmm1, xmm6
+    pmaddubsw  xmm2, xmm6
+    pmaddubsw  xmm3, xmm6
+    phaddw     xmm0, xmm1
+    phaddw     xmm2, xmm3
+    psrlw      xmm0, 8
+    psrlw      xmm2, 8
+    packuswb   xmm0, xmm2
+    paddb      xmm0, xmm5
+    lea        eax,  [eax + 64]
+    movdqa     [edx + edi], xmm0
+    lea        edx,  [edx + 16]
+    jg         convertloop
+
+    pop        edi
+    ret
+  }
+}
+
+__declspec(naked) __declspec(align(16))
+void ARGBToUV444Row_Unaligned_SSSE3(const uint8* src_argb0,
+                                    uint8* dst_u, uint8* dst_v, int width) {
+__asm {
+    push       edi
+    mov        eax, [esp + 4 + 4]   // src_argb
+    mov        edx, [esp + 4 + 8]   // dst_u
+    mov        edi, [esp + 4 + 12]  // dst_v
+    mov        ecx, [esp + 4 + 16]  // pix
+    movdqa     xmm7, kARGBToU
+    movdqa     xmm6, kARGBToV
+    movdqa     xmm5, kAddUV128
+    sub        edi, edx             // stride from u to v
+
+    align      16
+ convertloop:
+    /* convert to U and V */
+    movdqu     xmm0, [eax]          // U
+    movdqu     xmm1, [eax + 16]
+    movdqu     xmm2, [eax + 32]
+    movdqu     xmm3, [eax + 48]
+    pmaddubsw  xmm0, xmm7
+    pmaddubsw  xmm1, xmm7
+    pmaddubsw  xmm2, xmm7
+    pmaddubsw  xmm3, xmm7
+    phaddw     xmm0, xmm1
+    phaddw     xmm2, xmm3
+    psrlw      xmm0, 8
+    psrlw      xmm2, 8
+    packuswb   xmm0, xmm2
+    paddb      xmm0, xmm5
+    sub        ecx,  16
+    movdqu     [edx], xmm0
+
+    movdqu     xmm0, [eax]          // V
+    movdqu     xmm1, [eax + 16]
+    movdqu     xmm2, [eax + 32]
+    movdqu     xmm3, [eax + 48]
+    pmaddubsw  xmm0, xmm6
+    pmaddubsw  xmm1, xmm6
+    pmaddubsw  xmm2, xmm6
+    pmaddubsw  xmm3, xmm6
+    phaddw     xmm0, xmm1
+    phaddw     xmm2, xmm3
+    psrlw      xmm0, 8
+    psrlw      xmm2, 8
+    packuswb   xmm0, xmm2
+    paddb      xmm0, xmm5
+    lea        eax,  [eax + 64]
+    movdqu     [edx + edi], xmm0
+    lea        edx,  [edx + 16]
+    jg         convertloop
+
+    pop        edi
+    ret
+  }
+}
+
 __declspec(naked) __declspec(align(16))
 void ARGBToUV422Row_SSSE3(const uint8* src_argb0,
                           uint8* dst_u, uint8* dst_v, int width) {