Merge "Provide option to swap bytes for 16 bit colormodes"

configure.ac

@@ -226,6 +226,19 @@ if test "$target_os" = "mingw32"; then
   fi
 fi
 
+dnl === If --enable-swap-16bit-csp is defined, add -DWEBP_SWAP_16BIT_CSP
+
+USE_SWAP_16BIT_CSP=""
+AC_MSG_CHECKING(if --enable-swap-16bit-csp option is specified)
+AC_ARG_ENABLE([swap-16bit-csp],
+              AS_HELP_STRING([--enable-swap-16bit-csp],
+                             [Enable byte swap for 16 bit colorspaces]))
+if test "$enable_swap_16bit_csp" = "yes"; then
+  USE_SWAP_16BIT_CSP="-DWEBP_SWAP_16BIT_CSP"
+fi
+AC_MSG_RESULT(${enable_swap_16bit_csp-no})
+AC_SUBST(USE_SWAP_16BIT_CSP)
+
 dnl === If --enable-experimental is defined, add -DWEBP_EXPERIMENTAL_FEATURES
 
 USE_EXPERIMENTAL_CODE=""

makefile.unix

@@ -53,6 +53,9 @@ endif
 # Extra flags to enable experimental features and code
 # EXTRA_FLAGS += -DWEBP_EXPERIMENTAL_FEATURES
 
+# Extra flags to enable byte swap for 16 bit colorspaces.
+# EXTRA_FLAGS += -DWEBP_SWAP_16BIT_CSP
+
 # Extra flags to enable multi-threading
 EXTRA_FLAGS += -DWEBP_USE_THREAD
 EXTRA_LIBS += -lpthread

src/dsp/Makefile.am

@@ -23,5 +23,5 @@ noinst_HEADERS += ../dec/decode_vp8.h
 noinst_HEADERS += ../webp/decode.h
 
 libwebpdsp_la_LDFLAGS = -lm
-libwebpdsp_la_CPPFLAGS = $(USE_EXPERIMENTAL_CODE)
+libwebpdsp_la_CPPFLAGS = $(USE_EXPERIMENTAL_CODE) $(USE_SWAP_16BIT_CSP)
 libwebpdspincludedir = $(includedir)/webp

src/dsp/lossless.c

@@ -1033,8 +1033,15 @@ static void ConvertBGRAToRGBA4444(const uint32_t* src,
   const uint32_t* const src_end = src + num_pixels;
   while (src < src_end) {
     const uint32_t argb = *src++;
-    *dst++ = ((argb >> 16) & 0xf0) | ((argb >> 12) & 0xf);
+    const uint8_t rg = ((argb >> 16) & 0xf0) | ((argb >> 12) & 0xf);
-    *dst++ = ((argb >>  0) & 0xf0) | ((argb >> 28) & 0xf);
+    const uint8_t ba = ((argb >>  0) & 0xf0) | ((argb >> 28) & 0xf);
+#ifdef WEBP_SWAP_16BIT_CSP
+    *dst++ = ba;
+    *dst++ = rg;
+#else
+    *dst++ = rg;
+    *dst++ = ba;
+#endif
   }
 }
 
@@ -1043,8 +1050,15 @@ static void ConvertBGRAToRGB565(const uint32_t* src,
   const uint32_t* const src_end = src + num_pixels;
   while (src < src_end) {
     const uint32_t argb = *src++;
-    *dst++ = ((argb >> 16) & 0xf8) | ((argb >> 13) & 0x7);
+    const uint8_t rg = ((argb >> 16) & 0xf8) | ((argb >> 13) & 0x7);
-    *dst++ = ((argb >>  5) & 0xe0) | ((argb >>  3) & 0x1f);
+    const uint8_t gb = ((argb >>  5) & 0xe0) | ((argb >>  3) & 0x1f);
+#ifdef WEBP_SWAP_16BIT_CSP
+    *dst++ = gb;
+    *dst++ = rg;
+#else
+    *dst++ = rg;
+    *dst++ = gb;
+#endif
   }
 }
 

src/dsp/yuv.h

@@ -74,10 +74,17 @@ static WEBP_INLINE void VP8YuvToRgb565(uint8_t y, uint8_t u, uint8_t v,
   const int r_off = VP8kVToR[v];
   const int g_off = (VP8kVToG[v] + VP8kUToG[u]) >> YUV_FIX;
   const int b_off = VP8kUToB[u];
-  rgb[0] = ((VP8kClip[y + r_off - YUV_RANGE_MIN] & 0xf8) |
+  const uint8_t rg = ((VP8kClip[y + r_off - YUV_RANGE_MIN] & 0xf8) |
-            (VP8kClip[y + g_off - YUV_RANGE_MIN] >> 5));
+                      (VP8kClip[y + g_off - YUV_RANGE_MIN] >> 5));
-  rgb[1] = (((VP8kClip[y + g_off - YUV_RANGE_MIN] << 3) & 0xe0) |
+  const uint8_t gb = (((VP8kClip[y + g_off - YUV_RANGE_MIN] << 3) & 0xe0) |
-            (VP8kClip[y + b_off - YUV_RANGE_MIN] >> 3));
+                      (VP8kClip[y + b_off - YUV_RANGE_MIN] >> 3));
+#ifdef WEBP_SWAP_16BIT_CSP
+  rgb[0] = gb;
+  rgb[1] = rg;
+#else
+  rgb[0] = rg;
+  rgb[1] = gb;
+#endif
 }
 
 static WEBP_INLINE void VP8YuvToArgb(uint8_t y, uint8_t u, uint8_t v,
@@ -91,10 +98,16 @@ static WEBP_INLINE void VP8YuvToRgba4444(uint8_t y, uint8_t u, uint8_t v,
   const int r_off = VP8kVToR[v];
   const int g_off = (VP8kVToG[v] + VP8kUToG[u]) >> YUV_FIX;
   const int b_off = VP8kUToB[u];
-  // Don't update alpha (last 4 bits of argb[1])
+  const uint8_t rg = ((VP8kClip4Bits[y + r_off - YUV_RANGE_MIN] << 4) |
-  argb[0] = ((VP8kClip4Bits[y + r_off - YUV_RANGE_MIN] << 4) |
+                      VP8kClip4Bits[y + g_off - YUV_RANGE_MIN]);
-             VP8kClip4Bits[y + g_off - YUV_RANGE_MIN]);
+  const uint8_t ba = (VP8kClip4Bits[y + b_off - YUV_RANGE_MIN] << 4) | 0x0f;
-  argb[1] = 0x0f | (VP8kClip4Bits[y + b_off - YUV_RANGE_MIN] << 4);
+#ifdef WEBP_SWAP_16BIT_CSP
+  argb[0] = ba;
+  argb[1] = rg;
+#else
+  argb[0] = rg;
+  argb[1] = ba;
+#endif
 }
 
 static WEBP_INLINE void VP8YuvToBgr(uint8_t y, uint8_t u, uint8_t v,

src/webp/decode.h

@@ -130,7 +130,14 @@ WEBP_EXTERN(uint8_t*) WebPDecodeYUVInto(
 // Note: the naming describes the byte-ordering of packed samples in memory.
 // For instance, MODE_BGRA relates to samples ordered as B,G,R,A,B,G,R,A,...
 // Non-capital names (e.g.:MODE_Argb) relates to pre-multiplied RGB channels.
-// RGB-565 and RGBA-4444 are also endian-agnostic and byte-oriented.
+// RGBA-4444 and RGB-565 colorspaces are represented by following byte-order:
+// RGBA-4444: [r3 r2 r1 r0 g3 g2 g1 g0], [b3 b2 b1 b0 a3 a2 a1 a0], ...
+// RGB-565: [r4 r3 r2 r1 r0 g5 g4 g3], [g2 g1 g0 b4 b3 b2 b1 b0], ...
+// In the case WEBP_SWAP_16BITS_CSP is defined, the bytes are swapped for
+// these two modes:
+// RGBA-4444: [b3 b2 b1 b0 a3 a2 a1 a0], [r3 r2 r1 r0 g3 g2 g1 g0], ...
+// RGB-565: [g2 g1 g0 b4 b3 b2 b1 b0], [r4 r3 r2 r1 r0 g5 g4 g3], ...
+
 enum WEBP_CSP_MODE {
   MODE_RGB = 0, MODE_RGBA = 1,
   MODE_BGR = 2, MODE_BGRA = 3,