More correct rounding for the rgb/bgr->yuv converters.
[ffmpeg.git] / libswscale / yuv2rgb.c
index e9dc5b4..5aa42c0 100644 (file)
@@ -1,14 +1,17 @@
 /*
- * yuv2rgb.c, Software YUV to RGB coverter
+ * yuv2rgb.c, Software YUV to RGB converter
  *
  *  Copyright (C) 1999, Aaron Holtzman <aholtzma@ess.engr.uvic.ca>
- *  All Rights Reserved.
  *
  *  Functions broken out from display_x11.c and several new modes
- *  added by Håkan Hjort <d95hjort@dtek.chalmers.se>
+ *  added by Håkan Hjort <d95hjort@dtek.chalmers.se>
  *
  *  15 & 16 bpp support by Franck Sicard <Franck.Sicard@solsoft.fr>
  *
+ *  MMX/MMX2 template stuff (needed for fast movntq support),
+ *  1,4,8bpp support and context / deglobalize stuff
+ *  by Michael Niedermayer (michaelni@gmx.at)
+ *
  *  This file is part of mpeg2dec, a free MPEG-2 video decoder
  *
  *  mpeg2dec is free software; you can redistribute it and/or modify
  *  GNU General Public License for more details.
  *
  *  You should have received a copy of the GNU General Public License
- *  along with GNU Make; see the file COPYING.  If not, write to
- *  the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
- *
- * MMX/MMX2 Template stuff from Michael Niedermayer (michaelni@gmx.at) (needed for fast movntq support)
- * 1,4,8bpp support by Michael Niedermayer (michaelni@gmx.at)
- * context / deglobalize stuff by Michael Niedermayer
+ *  along with mpeg2dec; if not, write to the Free Software
+ *  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
 #include <stdio.h>
 #include "swscale.h"
 #include "swscale_internal.h"
 
-#ifdef HAVE_MLIB
-#include "yuv2rgb_mlib.c"
-#endif
-
-#define DITHER1XBPP // only for mmx
+#define DITHER1XBPP // only for MMX
 
 const uint8_t  __attribute__((aligned(8))) dither_2x2_4[2][8]={
 {  1,   3,   1,   3,   1,   3,   1,   3, },
@@ -156,29 +151,17 @@ const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
 #ifdef HAVE_MMX
 
 /* hope these constant values are cache line aligned */
-static uint64_t attribute_used __attribute__((aligned(8))) mmx_00ffw = 0x00ff00ff00ff00ffULL;
-static uint64_t attribute_used __attribute__((aligned(8))) mmx_redmask = 0xf8f8f8f8f8f8f8f8ULL;
-static uint64_t attribute_used __attribute__((aligned(8))) mmx_grnmask = 0xfcfcfcfcfcfcfcfcULL;
-
-static uint64_t attribute_used __attribute__((aligned(8))) M24A=   0x00FF0000FF0000FFULL;
-static uint64_t attribute_used __attribute__((aligned(8))) M24B=   0xFF0000FF0000FF00ULL;
-static uint64_t attribute_used __attribute__((aligned(8))) M24C=   0x0000FF0000FF0000ULL;
+DECLARE_ASM_CONST(8, uint64_t, mmx_00ffw)   = 0x00ff00ff00ff00ffULL;
+DECLARE_ASM_CONST(8, uint64_t, mmx_redmask) = 0xf8f8f8f8f8f8f8f8ULL;
+DECLARE_ASM_CONST(8, uint64_t, mmx_grnmask) = 0xfcfcfcfcfcfcfcfcULL;
 
-// the volatile is required because gcc otherwise optimizes some writes away not knowing that these
-// are read in the asm block
+// The volatile is required because gcc otherwise optimizes some writes away
+// not knowing that these are read in the ASM block.
 static volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither;
 static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither;
 static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither;
 static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither;
 
-static uint64_t __attribute__((aligned(8))) dither4[2]={
-       0x0103010301030103LL,
-       0x0200020002000200LL,};
-
-static uint64_t __attribute__((aligned(8))) dither8[2]={
-       0x0602060206020602LL,
-       0x0004000400040004LL,};
-
 #undef HAVE_MMX
 
 //MMX versions
@@ -197,7 +180,7 @@ static uint64_t __attribute__((aligned(8))) dither8[2]={
 #define RENAME(a) a ## _MMX2
 #include "yuv2rgb_template.c"
 
-#endif /* defined(ARCH_X86) */
+#endif /* HAVE_MMX */
 
 const int32_t Inverse_Table_6_9[8][4] = {
     {117504, 138453, 13954, 34903}, /* no sequence_display_extension */
@@ -210,410 +193,459 @@ const int32_t Inverse_Table_6_9[8][4] = {
     {117579, 136230, 16907, 35559}  /* SMPTE 240M (1987) */
 };
 
-#define RGB(i)                                 \
-       U = pu[i];                              \
-       V = pv[i];                              \
-       r = (void *)c->table_rV[V];                     \
-       g = (void *)(c->table_gU[U] + c->table_gV[V]);          \
-       b = (void *)c->table_bU[U];
-
-#define DST1(i)                                        \
-       Y = py_1[2*i];                          \
-       dst_1[2*i] = r[Y] + g[Y] + b[Y];        \
-       Y = py_1[2*i+1];                        \
-       dst_1[2*i+1] = r[Y] + g[Y] + b[Y];
-
-#define DST2(i)                                        \
-       Y = py_2[2*i];                          \
-       dst_2[2*i] = r[Y] + g[Y] + b[Y];        \
-       Y = py_2[2*i+1];                        \
-       dst_2[2*i+1] = r[Y] + g[Y] + b[Y];
-
-#define DST1RGB(i)                                                     \
-       Y = py_1[2*i];                                                  \
-       dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y];    \
-       Y = py_1[2*i+1];                                                \
-       dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y];
-
-#define DST2RGB(i)                                                     \
-       Y = py_2[2*i];                                                  \
-       dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y];    \
-       Y = py_2[2*i+1];                                                \
-       dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y];
-
-#define DST1BGR(i)                                                     \
-       Y = py_1[2*i];                                                  \
-       dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y];    \
-       Y = py_1[2*i+1];                                                \
-       dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y];
-
-#define DST2BGR(i)                                                     \
-       Y = py_2[2*i];                                                  \
-       dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y];    \
-       Y = py_2[2*i+1];                                                \
-       dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y];
+#define RGB(i)                                      \
+    U = pu[i];                                      \
+    V = pv[i];                                      \
+    r = (void *)c->table_rV[V];                     \
+    g = (void *)(c->table_gU[U] + c->table_gV[V]);  \
+    b = (void *)c->table_bU[U];
+
+#define DST1(i)                         \
+    Y = py_1[2*i];                      \
+    dst_1[2*i] = r[Y] + g[Y] + b[Y];    \
+    Y = py_1[2*i+1];                    \
+    dst_1[2*i+1] = r[Y] + g[Y] + b[Y];
+
+#define DST2(i)                         \
+    Y = py_2[2*i];                      \
+    dst_2[2*i] = r[Y] + g[Y] + b[Y];    \
+    Y = py_2[2*i+1];                    \
+    dst_2[2*i+1] = r[Y] + g[Y] + b[Y];
+
+#define DST1RGB(i)                                                \
+    Y = py_1[2*i];                                                \
+    dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y];  \
+    Y = py_1[2*i+1];                                              \
+    dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y];
+
+#define DST2RGB(i)                                                \
+    Y = py_2[2*i];                                                \
+    dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y];  \
+    Y = py_2[2*i+1];                                              \
+    dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y];
+
+#define DST1BGR(i)                                                \
+    Y = py_1[2*i];                                                \
+    dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y];  \
+    Y = py_1[2*i+1];                                              \
+    dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y];
+
+#define DST2BGR(i)                                                \
+    Y = py_2[2*i];                                                \
+    dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y];  \
+    Y = py_2[2*i+1];                                              \
+    dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y];
 
 #define PROLOG(func_name, dst_type) \
 static int func_name(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, \
-             int srcSliceH, uint8_t* dst[], int dstStride[]){\
+                     int srcSliceH, uint8_t* dst[], int dstStride[]){\
     int y;\
 \
-    if(c->srcFormat == PIX_FMT_YUV422P){\
-       srcStride[1] *= 2;\
-       srcStride[2] *= 2;\
+    if (c->srcFormat == PIX_FMT_YUV422P){\
+        srcStride[1] *= 2;\
+        srcStride[2] *= 2;\
     }\
-    for(y=0; y<srcSliceH; y+=2){\
-       dst_type *dst_1= (dst_type*)(dst[0] + (y+srcSliceY  )*dstStride[0]);\
-       dst_type *dst_2= (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
-       dst_type attribute_unused *r, *b;\
-       dst_type *g;\
-       uint8_t *py_1= src[0] + y*srcStride[0];\
-       uint8_t *py_2= py_1 + srcStride[0];\
-       uint8_t *pu= src[1] + (y>>1)*srcStride[1];\
-       uint8_t *pv= src[2] + (y>>1)*srcStride[2];\
-       unsigned int h_size= c->dstW>>3;\
-       while (h_size--) {\
-           int attribute_unused U, V;\
-           int Y;\
-
-#define EPILOG(dst_delta)\
-           pu += 4;\
-           pv += 4;\
-           py_1 += 8;\
-           py_2 += 8;\
-           dst_1 += dst_delta;\
-           dst_2 += dst_delta;\
-       }\
+    for (y=0; y<srcSliceH; y+=2){\
+        dst_type *dst_1= (dst_type*)(dst[0] + (y+srcSliceY  )*dstStride[0]);\
+        dst_type *dst_2= (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
+        dst_type av_unused *r, *b;\
+        dst_type *g;\
+        uint8_t *py_1= src[0] + y*srcStride[0];\
+        uint8_t *py_2= py_1 + srcStride[0];\
+        uint8_t *pu= src[1] + (y>>1)*srcStride[1];\
+        uint8_t *pv= src[2] + (y>>1)*srcStride[2];\
+        unsigned int h_size= c->dstW>>3;\
+        while (h_size--) {\
+            int av_unused U, V;\
+            int Y;\
+
+#define EPILOG1(dst_delta)\
+            pu += 4;\
+            pv += 4;\
+            py_1 += 8;\
+            py_2 += 8;\
+            dst_1 += dst_delta;\
+            dst_2 += dst_delta;\
+        }\
+        if (c->dstW & 4) {\
+            int av_unused Y, U, V;\
+
+#define EPILOG2()\
+        }\
     }\
     return srcSliceH;\
 }
 
-PROLOG(yuv2rgb_c_32, uint32_t)
-       RGB(0);
-       DST1(0);
-       DST2(0);
-
-       RGB(1);
-       DST2(1);
-       DST1(1);
-
-       RGB(2);
-       DST1(2);
-       DST2(2);
+#define EPILOG(dst_delta)\
+    EPILOG1(dst_delta)\
+    EPILOG2()
 
-       RGB(3);
-       DST2(3);
-       DST1(3);
-EPILOG(8)
+PROLOG(yuv2rgb_c_32, uint32_t)
+    RGB(0);
+    DST1(0);
+    DST2(0);
+
+    RGB(1);
+    DST2(1);
+    DST1(1);
+
+    RGB(2);
+    DST1(2);
+    DST2(2);
+
+    RGB(3);
+    DST2(3);
+    DST1(3);
+EPILOG1(8)
+    RGB(0);
+    DST1(0);
+    DST2(0);
+
+    RGB(1);
+    DST2(1);
+    DST1(1);
+EPILOG2()
 
 PROLOG(yuv2rgb_c_24_rgb, uint8_t)
-       RGB(0);
-       DST1RGB(0);
-       DST2RGB(0);
-
-       RGB(1);
-       DST2RGB(1);
-       DST1RGB(1);
-
-       RGB(2);
-       DST1RGB(2);
-       DST2RGB(2);
-
-       RGB(3);
-       DST2RGB(3);
-       DST1RGB(3);
-EPILOG(24)
+    RGB(0);
+    DST1RGB(0);
+    DST2RGB(0);
+
+    RGB(1);
+    DST2RGB(1);
+    DST1RGB(1);
+
+    RGB(2);
+    DST1RGB(2);
+    DST2RGB(2);
+
+    RGB(3);
+    DST2RGB(3);
+    DST1RGB(3);
+EPILOG1(24)
+    RGB(0);
+    DST1RGB(0);
+    DST2RGB(0);
+
+    RGB(1);
+    DST2RGB(1);
+    DST1RGB(1);
+EPILOG2()
 
 // only trivial mods from yuv2rgb_c_24_rgb
 PROLOG(yuv2rgb_c_24_bgr, uint8_t)
-       RGB(0);
-       DST1BGR(0);
-       DST2BGR(0);
-
-       RGB(1);
-       DST2BGR(1);
-       DST1BGR(1);
-
-       RGB(2);
-       DST1BGR(2);
-       DST2BGR(2);
-
-       RGB(3);
-       DST2BGR(3);
-       DST1BGR(3);
-EPILOG(24)
+    RGB(0);
+    DST1BGR(0);
+    DST2BGR(0);
+
+    RGB(1);
+    DST2BGR(1);
+    DST1BGR(1);
+
+    RGB(2);
+    DST1BGR(2);
+    DST2BGR(2);
+
+    RGB(3);
+    DST2BGR(3);
+    DST1BGR(3);
+EPILOG1(24)
+    RGB(0);
+    DST1BGR(0);
+    DST2BGR(0);
+
+    RGB(1);
+    DST2BGR(1);
+    DST1BGR(1);
+EPILOG2()
 
 // This is exactly the same code as yuv2rgb_c_32 except for the types of
 // r, g, b, dst_1, dst_2
 PROLOG(yuv2rgb_c_16, uint16_t)
-       RGB(0);
-       DST1(0);
-       DST2(0);
+    RGB(0);
+    DST1(0);
+    DST2(0);
 
-       RGB(1);
-       DST2(1);
-       DST1(1);
+    RGB(1);
+    DST2(1);
+    DST1(1);
 
-       RGB(2);
-       DST1(2);
-       DST2(2);
+    RGB(2);
+    DST1(2);
+    DST2(2);
 
-       RGB(3);
-       DST2(3);
-       DST1(3);
+    RGB(3);
+    DST2(3);
+    DST1(3);
 EPILOG(8)
 
 // This is exactly the same code as yuv2rgb_c_32 except for the types of
 // r, g, b, dst_1, dst_2
 PROLOG(yuv2rgb_c_8, uint8_t)
-       RGB(0);
-       DST1(0);
-       DST2(0);
+    RGB(0);
+    DST1(0);
+    DST2(0);
 
-       RGB(1);
-       DST2(1);
-       DST1(1);
+    RGB(1);
+    DST2(1);
+    DST1(1);
 
-       RGB(2);
-       DST1(2);
-       DST2(2);
+    RGB(2);
+    DST1(2);
+    DST2(2);
 
-       RGB(3);
-       DST2(3);
-       DST1(3);
+    RGB(3);
+    DST2(3);
+    DST1(3);
 EPILOG(8)
 
 // r, g, b, dst_1, dst_2
 PROLOG(yuv2rgb_c_8_ordered_dither, uint8_t)
-       const uint8_t *d32= dither_8x8_32[y&7];
-       const uint8_t *d64= dither_8x8_73[y&7];
-#define DST1bpp8(i,o)                                  \
-       Y = py_1[2*i];                          \
-       dst_1[2*i] = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]];     \
-       Y = py_1[2*i+1];                        \
-       dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
-
-#define DST2bpp8(i,o)                                  \
-       Y = py_2[2*i];                          \
-       dst_2[2*i] =  r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]];    \
-       Y = py_2[2*i+1];                        \
-       dst_2[2*i+1] =  r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
-
-
-       RGB(0);
-       DST1bpp8(0,0);
-       DST2bpp8(0,0);
-
-       RGB(1);
-       DST2bpp8(1,2);
-       DST1bpp8(1,2);
-
-       RGB(2);
-       DST1bpp8(2,4);
-       DST2bpp8(2,4);
-
-       RGB(3);
-       DST2bpp8(3,6);
-       DST1bpp8(3,6);
+    const uint8_t *d32= dither_8x8_32[y&7];
+    const uint8_t *d64= dither_8x8_73[y&7];
+#define DST1bpp8(i,o)                                               \
+    Y = py_1[2*i];                                                  \
+    dst_1[2*i]   = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]];   \
+    Y = py_1[2*i+1];                                                \
+    dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
+
+#define DST2bpp8(i,o)                                               \
+    Y = py_2[2*i];                                                  \
+    dst_2[2*i]   =  r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]];  \
+    Y = py_2[2*i+1];                                                \
+    dst_2[2*i+1] =  r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
+
+
+    RGB(0);
+    DST1bpp8(0,0);
+    DST2bpp8(0,0);
+
+    RGB(1);
+    DST2bpp8(1,2);
+    DST1bpp8(1,2);
+
+    RGB(2);
+    DST1bpp8(2,4);
+    DST2bpp8(2,4);
+
+    RGB(3);
+    DST2bpp8(3,6);
+    DST1bpp8(3,6);
 EPILOG(8)
 
 
 // This is exactly the same code as yuv2rgb_c_32 except for the types of
 // r, g, b, dst_1, dst_2
 PROLOG(yuv2rgb_c_4, uint8_t)
-        int acc;
-#define DST1_4(i)                                      \
-       Y = py_1[2*i];                          \
-       acc = r[Y] + g[Y] + b[Y];       \
-       Y = py_1[2*i+1];                        \
-        acc |= (r[Y] + g[Y] + b[Y])<<4;\
-       dst_1[i] = acc; 
-
-#define DST2_4(i)                                      \
-       Y = py_2[2*i];                          \
-       acc = r[Y] + g[Y] + b[Y];       \
-       Y = py_2[2*i+1];                        \
-       acc |= (r[Y] + g[Y] + b[Y])<<4;\
-       dst_2[i] = acc; 
-       
-        RGB(0);
-       DST1_4(0);
-       DST2_4(0);
-
-       RGB(1);
-       DST2_4(1);
-       DST1_4(1);
-
-       RGB(2);
-       DST1_4(2);
-       DST2_4(2);
-
-       RGB(3);
-       DST2_4(3);
-       DST1_4(3);
+    int acc;
+#define DST1_4(i)                   \
+    Y = py_1[2*i];                  \
+    acc = r[Y] + g[Y] + b[Y];       \
+    Y = py_1[2*i+1];                \
+    acc |= (r[Y] + g[Y] + b[Y])<<4; \
+    dst_1[i] = acc;
+
+#define DST2_4(i)                   \
+    Y = py_2[2*i];                  \
+    acc = r[Y] + g[Y] + b[Y];       \
+    Y = py_2[2*i+1];                \
+    acc |= (r[Y] + g[Y] + b[Y])<<4; \
+    dst_2[i] = acc;
+
+    RGB(0);
+    DST1_4(0);
+    DST2_4(0);
+
+    RGB(1);
+    DST2_4(1);
+    DST1_4(1);
+
+    RGB(2);
+    DST1_4(2);
+    DST2_4(2);
+
+    RGB(3);
+    DST2_4(3);
+    DST1_4(3);
 EPILOG(4)
 
 PROLOG(yuv2rgb_c_4_ordered_dither, uint8_t)
-       const uint8_t *d64= dither_8x8_73[y&7];
-       const uint8_t *d128=dither_8x8_220[y&7];
-        int acc;
-
-#define DST1bpp4(i,o)                                  \
-       Y = py_1[2*i];                          \
-       acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];  \
-       Y = py_1[2*i+1];                        \
-       acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4;\
-        dst_1[i]= acc;
-
-#define DST2bpp4(i,o)                                  \
-       Y = py_2[2*i];                          \
-       acc =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]]; \
-       Y = py_2[2*i+1];                        \
-       acc |=  (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4;\
-        dst_2[i]= acc;
-
-
-       RGB(0);
-       DST1bpp4(0,0);
-       DST2bpp4(0,0);
-
-       RGB(1);
-       DST2bpp4(1,2);
-       DST1bpp4(1,2);
-
-       RGB(2);
-       DST1bpp4(2,4);
-       DST2bpp4(2,4);
-
-       RGB(3);
-       DST2bpp4(3,6);
-       DST1bpp4(3,6);
+    const uint8_t *d64= dither_8x8_73[y&7];
+    const uint8_t *d128=dither_8x8_220[y&7];
+    int acc;
+
+#define DST1bpp4(i,o)                                             \
+    Y = py_1[2*i];                                                \
+    acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];        \
+    Y = py_1[2*i+1];                                              \
+    acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4;  \
+    dst_1[i]= acc;
+
+#define DST2bpp4(i,o)                                             \
+    Y = py_2[2*i];                                                \
+    acc =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];       \
+    Y = py_2[2*i+1];                                              \
+    acc |=  (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4; \
+    dst_2[i]= acc;
+
+
+    RGB(0);
+    DST1bpp4(0,0);
+    DST2bpp4(0,0);
+
+    RGB(1);
+    DST2bpp4(1,2);
+    DST1bpp4(1,2);
+
+    RGB(2);
+    DST1bpp4(2,4);
+    DST2bpp4(2,4);
+
+    RGB(3);
+    DST2bpp4(3,6);
+    DST1bpp4(3,6);
 EPILOG(4)
 
 // This is exactly the same code as yuv2rgb_c_32 except for the types of
 // r, g, b, dst_1, dst_2
 PROLOG(yuv2rgb_c_4b, uint8_t)
-       RGB(0);
-       DST1(0);
-       DST2(0);
+    RGB(0);
+    DST1(0);
+    DST2(0);
 
-       RGB(1);
-       DST2(1);
-       DST1(1);
+    RGB(1);
+    DST2(1);
+    DST1(1);
 
-       RGB(2);
-       DST1(2);
-       DST2(2);
+    RGB(2);
+    DST1(2);
+    DST2(2);
 
-       RGB(3);
-       DST2(3);
-       DST1(3);
+    RGB(3);
+    DST2(3);
+    DST1(3);
 EPILOG(8)
 
 PROLOG(yuv2rgb_c_4b_ordered_dither, uint8_t)
-       const uint8_t *d64= dither_8x8_73[y&7];
-       const uint8_t *d128=dither_8x8_220[y&7];
+    const uint8_t *d64= dither_8x8_73[y&7];
+    const uint8_t *d128=dither_8x8_220[y&7];
 
-#define DST1bpp4b(i,o)                                 \
-       Y = py_1[2*i];                          \
-       dst_1[2*i] = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];   \
-       Y = py_1[2*i+1];                        \
-       dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
+#define DST1bpp4b(i,o)                                                \
+    Y = py_1[2*i];                                                    \
+    dst_1[2*i]   = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];   \
+    Y = py_1[2*i+1];                                                  \
+    dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
 
-#define DST2bpp4b(i,o)                                 \
-       Y = py_2[2*i];                          \
-       dst_2[2*i] =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];  \
-       Y = py_2[2*i+1];                        \
-       dst_2[2*i+1] =  r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
+#define DST2bpp4b(i,o)                                                \
+    Y = py_2[2*i];                                                    \
+    dst_2[2*i]   =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];  \
+    Y = py_2[2*i+1];                                                  \
+    dst_2[2*i+1] =  r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
 
 
-       RGB(0);
-       DST1bpp4b(0,0);
-       DST2bpp4b(0,0);
+    RGB(0);
+    DST1bpp4b(0,0);
+    DST2bpp4b(0,0);
 
-       RGB(1);
-       DST2bpp4b(1,2);
-       DST1bpp4b(1,2);
+    RGB(1);
+    DST2bpp4b(1,2);
+    DST1bpp4b(1,2);
 
-       RGB(2);
-       DST1bpp4b(2,4);
-       DST2bpp4b(2,4);
+    RGB(2);
+    DST1bpp4b(2,4);
+    DST2bpp4b(2,4);
 
-       RGB(3);
-       DST2bpp4b(3,6);
-       DST1bpp4b(3,6);
+    RGB(3);
+    DST2bpp4b(3,6);
+    DST1bpp4b(3,6);
 EPILOG(8)
 
 PROLOG(yuv2rgb_c_1_ordered_dither, uint8_t)
-       const uint8_t *d128=dither_8x8_220[y&7];
-       char out_1=0, out_2=0;
-       g= c->table_gU[128] + c->table_gV[128];
-
-#define DST1bpp1(i,o)                                  \
-       Y = py_1[2*i];                          \
-       out_1+= out_1 + g[Y+d128[0+o]]; \
-       Y = py_1[2*i+1];                        \
-       out_1+= out_1 + g[Y+d128[1+o]];
-
-#define DST2bpp1(i,o)                                  \
-       Y = py_2[2*i];                          \
-       out_2+= out_2 + g[Y+d128[8+o]]; \
-       Y = py_2[2*i+1];                        \
-       out_2+= out_2 + g[Y+d128[9+o]];
-
-       DST1bpp1(0,0);
-       DST2bpp1(0,0);
-
-       DST2bpp1(1,2);
-       DST1bpp1(1,2);
-
-       DST1bpp1(2,4);
-       DST2bpp1(2,4);
-
-       DST2bpp1(3,6);
-       DST1bpp1(3,6);
-       
-       dst_1[0]= out_1;
-       dst_2[0]= out_2;
+        const uint8_t *d128=dither_8x8_220[y&7];
+        char out_1=0, out_2=0;
+        g= c->table_gU[128] + c->table_gV[128];
+
+#define DST1bpp1(i,o)               \
+    Y = py_1[2*i];                  \
+    out_1+= out_1 + g[Y+d128[0+o]]; \
+    Y = py_1[2*i+1];                \
+    out_1+= out_1 + g[Y+d128[1+o]];
+
+#define DST2bpp1(i,o)               \
+    Y = py_2[2*i];                  \
+    out_2+= out_2 + g[Y+d128[8+o]]; \
+    Y = py_2[2*i+1];                \
+    out_2+= out_2 + g[Y+d128[9+o]];
+
+    DST1bpp1(0,0);
+    DST2bpp1(0,0);
+
+    DST2bpp1(1,2);
+    DST1bpp1(1,2);
+
+    DST1bpp1(2,4);
+    DST2bpp1(2,4);
+
+    DST2bpp1(3,6);
+    DST1bpp1(3,6);
+
+    dst_1[0]= out_1;
+    dst_2[0]= out_2;
 EPILOG(1)
 
 SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
 {
 #if defined(HAVE_MMX2) || defined(HAVE_MMX)
-    if(c->flags & SWS_CPU_CAPS_MMX2){
-       switch(c->dstFormat){
-       case PIX_FMT_RGB32: return yuv420_rgb32_MMX2;
-       case PIX_FMT_BGR24: return yuv420_rgb24_MMX2;
-       case PIX_FMT_BGR565: return yuv420_rgb16_MMX2;
-       case PIX_FMT_BGR555: return yuv420_rgb15_MMX2;
-       }
+    if (c->flags & SWS_CPU_CAPS_MMX2){
+        switch(c->dstFormat){
+        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX2;
+        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX2;
+        case PIX_FMT_RGB565: return yuv420_rgb16_MMX2;
+        case PIX_FMT_RGB555: return yuv420_rgb15_MMX2;
+        }
     }
-    if(c->flags & SWS_CPU_CAPS_MMX){
-       switch(c->dstFormat){
-       case PIX_FMT_RGB32: return yuv420_rgb32_MMX;
-       case PIX_FMT_BGR24: return yuv420_rgb24_MMX;
-       case PIX_FMT_BGR565: return yuv420_rgb16_MMX;
-       case PIX_FMT_BGR555: return yuv420_rgb15_MMX;
-       }
+    if (c->flags & SWS_CPU_CAPS_MMX){
+        switch(c->dstFormat){
+        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX;
+        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX;
+        case PIX_FMT_RGB565: return yuv420_rgb16_MMX;
+        case PIX_FMT_RGB555: return yuv420_rgb15_MMX;
+        }
     }
 #endif
-#ifdef HAVE_MLIB
+#ifdef HAVE_VIS
     {
-       SwsFunc t= yuv2rgb_init_mlib(c);
-       if(t) return t;
+        SwsFunc t= yuv2rgb_init_vis(c);
+        if (t) return t;
+    }
+#endif
+#ifdef CONFIG_MLIB
+    {
+        SwsFunc t= yuv2rgb_init_mlib(c);
+        if (t) return t;
     }
 #endif
 #ifdef HAVE_ALTIVEC
     if (c->flags & SWS_CPU_CAPS_ALTIVEC)
     {
-       SwsFunc t = yuv2rgb_init_altivec(c);
-       if(t) return t;
+        SwsFunc t = yuv2rgb_init_altivec(c);
+        if (t) return t;
+    }
+#endif
+
+#ifdef ARCH_BFIN
+    if (c->flags & SWS_CPU_CAPS_BFIN)
+    {
+        SwsFunc t = ff_bfin_yuv2rgb_get_func_ptr (c);
+        if (t) return t;
     }
 #endif
 
-    MSG_WARN("No accelerated colorspace conversion found\n");
+    av_log(c, AV_LOG_WARNING, "No accelerated colorspace conversion found.\n");
 
     switch(c->dstFormat){
+    case PIX_FMT_BGR32_1:
+    case PIX_FMT_RGB32_1:
     case PIX_FMT_BGR32:
     case PIX_FMT_RGB32: return yuv2rgb_c_32;
     case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
@@ -630,7 +662,7 @@ SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
     case PIX_FMT_BGR4_BYTE:  return yuv2rgb_c_4b_ordered_dither;
     case PIX_FMT_MONOBLACK:  return yuv2rgb_c_1_ordered_dither;
     default:
-       assert(0);
+        assert(0);
     }
     return NULL;
 }
@@ -638,16 +670,24 @@ SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
 static int div_round (int dividend, int divisor)
 {
     if (dividend > 0)
-       return (dividend + (divisor>>1)) / divisor;
+        return (dividend + (divisor>>1)) / divisor;
     else
-       return -((-dividend + (divisor>>1)) / divisor);
+        return -((-dividend + (divisor>>1)) / divisor);
 }
 
 int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange, int brightness, int contrast, int saturation)
-{  
-    const int isRgb = isBGR(c->dstFormat);
+{
+    const int isRgb =      c->dstFormat==PIX_FMT_RGB32
+                        || c->dstFormat==PIX_FMT_RGB32_1
+                        || c->dstFormat==PIX_FMT_BGR24
+                        || c->dstFormat==PIX_FMT_RGB565
+                        || c->dstFormat==PIX_FMT_RGB555
+                        || c->dstFormat==PIX_FMT_RGB8
+                        || c->dstFormat==PIX_FMT_RGB4
+                        || c->dstFormat==PIX_FMT_RGB4_BYTE
+                        || c->dstFormat==PIX_FMT_MONOBLACK;
     const int bpp = fmt_depth(c->dstFormat);
-    int i;
+    int i, base;
     uint8_t table_Y[1024];
     uint32_t *table_32 = 0;
     uint16_t *table_16 = 0;
@@ -667,11 +707,16 @@ int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange,
     int64_t oy  = 0;
 
 //printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
-    if(!fullRange){
-       cy= (cy*255) / 219;
-       oy= 16<<16;
+    if (!fullRange){
+        cy= (cy*255) / 219;
+        oy= 16<<16;
+    }else{
+        crv= (crv*224) / 255;
+        cbu= (cbu*224) / 255;
+        cgu= (cgu*224) / 255;
+        cgv= (cgv*224) / 255;
     }
-       
+
     cy = (cy *contrast             )>>16;
     crv= (crv*contrast * saturation)>>32;
     cbu= (cbu*contrast * saturation)>>32;
@@ -681,163 +726,164 @@ int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange,
     oy -= 256*brightness;
 
     for (i = 0; i < 1024; i++) {
-       int j;
+        int j;
 
-       j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
-       j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
-       table_Y[i] = j;
+        j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
+        j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
+        table_Y[i] = j;
     }
 
     switch (bpp) {
     case 32:
-       table_start= table_32 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
-
-       entry_size = sizeof (uint32_t);
-       table_r = table_32 + 197;
-       table_b = table_32 + 197 + 685;
-       table_g = table_32 + 197 + 2*682;
-
-       for (i = -197; i < 256+197; i++)
-           ((uint32_t *)table_r)[i] = table_Y[i+384] << (isRgb ? 16 : 0);
-       for (i = -132; i < 256+132; i++)
-           ((uint32_t *)table_g)[i] = table_Y[i+384] << 8;
-       for (i = -232; i < 256+232; i++)
-           ((uint32_t *)table_b)[i] = table_Y[i+384] << (isRgb ? 0 : 16);
-       break;
+        table_start= table_32 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
+        base= (c->dstFormat == PIX_FMT_RGB32_1 || c->dstFormat == PIX_FMT_BGR32_1) ? 8 : 0;
+
+        entry_size = sizeof (uint32_t);
+        table_r = table_32 + 197;
+        table_b = table_32 + 197 + 685;
+        table_g = table_32 + 197 + 2*682;
+
+        for (i = -197; i < 256+197; i++)
+            ((uint32_t *)table_r)[i] = table_Y[i+384] << ((isRgb ? 16 : 0) + base);
+        for (i = -132; i < 256+132; i++)
+            ((uint32_t *)table_g)[i] = table_Y[i+384] << (8                + base);
+        for (i = -232; i < 256+232; i++)
+            ((uint32_t *)table_b)[i] = table_Y[i+384] << ((isRgb ? 0 : 16) + base);
+        break;
 
     case 24:
-       table_start= table_8 = av_malloc ((256 + 2*232) * sizeof (uint8_t));
+        table_start= table_8 = av_malloc ((256 + 2*232) * sizeof (uint8_t));
 
-       entry_size = sizeof (uint8_t);
-       table_r = table_g = table_b = table_8 + 232;
+        entry_size = sizeof (uint8_t);
+        table_r = table_g = table_b = table_8 + 232;
 
-       for (i = -232; i < 256+232; i++)
-           ((uint8_t * )table_b)[i] = table_Y[i+384];
-       break;
+        for (i = -232; i < 256+232; i++)
+            ((uint8_t * )table_b)[i] = table_Y[i+384];
+        break;
 
     case 15:
     case 16:
-       table_start= table_16 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
+        table_start= table_16 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
 
-       entry_size = sizeof (uint16_t);
-       table_r = table_16 + 197;
-       table_b = table_16 + 197 + 685;
-       table_g = table_16 + 197 + 2*682;
+        entry_size = sizeof (uint16_t);
+        table_r = table_16 + 197;
+        table_b = table_16 + 197 + 685;
+        table_g = table_16 + 197 + 2*682;
 
-       for (i = -197; i < 256+197; i++) {
-           int j = table_Y[i+384] >> 3;
+        for (i = -197; i < 256+197; i++) {
+            int j = table_Y[i+384] >> 3;
 
-           if (isRgb)
-               j <<= ((bpp==16) ? 11 : 10);
+            if (isRgb)
+                j <<= ((bpp==16) ? 11 : 10);
 
-           ((uint16_t *)table_r)[i] = j;
-       }
-       for (i = -132; i < 256+132; i++) {
-           int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
+            ((uint16_t *)table_r)[i] = j;
+        }
+        for (i = -132; i < 256+132; i++) {
+            int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
 
-           ((uint16_t *)table_g)[i] = j << 5;
-       }
-       for (i = -232; i < 256+232; i++) {
-           int j = table_Y[i+384] >> 3;
+            ((uint16_t *)table_g)[i] = j << 5;
+        }
+        for (i = -232; i < 256+232; i++) {
+            int j = table_Y[i+384] >> 3;
 
-           if (!isRgb)
-               j <<= ((bpp==16) ? 11 : 10);
+            if (!isRgb)
+                j <<= ((bpp==16) ? 11 : 10);
 
-           ((uint16_t *)table_b)[i] = j;
-       }
-       break;
+            ((uint16_t *)table_b)[i] = j;
+        }
+        break;
 
     case 8:
-       table_start= table_332 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
+        table_start= table_332 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
 
-       entry_size = sizeof (uint8_t);
-       table_r = table_332 + 197;
-       table_b = table_332 + 197 + 685;
-       table_g = table_332 + 197 + 2*682;
+        entry_size = sizeof (uint8_t);
+        table_r = table_332 + 197;
+        table_b = table_332 + 197 + 685;
+        table_g = table_332 + 197 + 2*682;
 
-       for (i = -197; i < 256+197; i++) {
-           int j = (table_Y[i+384 - 16] + 18)/36;
+        for (i = -197; i < 256+197; i++) {
+            int j = (table_Y[i+384 - 16] + 18)/36;
 
-           if (isRgb)
-               j <<= 5;
+            if (isRgb)
+                j <<= 5;
 
-           ((uint8_t *)table_r)[i] = j;
-       }
-       for (i = -132; i < 256+132; i++) {
-           int j = (table_Y[i+384 - 16] + 18)/36;
+            ((uint8_t *)table_r)[i] = j;
+        }
+        for (i = -132; i < 256+132; i++) {
+            int j = (table_Y[i+384 - 16] + 18)/36;
 
-           if (!isRgb)
-               j <<= 1;
+            if (!isRgb)
+                j <<= 1;
 
-           ((uint8_t *)table_g)[i] = j << 2;
-       }
-       for (i = -232; i < 256+232; i++) {
-           int j = (table_Y[i+384 - 37] + 43)/85;
+            ((uint8_t *)table_g)[i] = j << 2;
+        }
+        for (i = -232; i < 256+232; i++) {
+            int j = (table_Y[i+384 - 37] + 43)/85;
 
-           if (!isRgb)
-               j <<= 6;
+            if (!isRgb)
+                j <<= 6;
 
-           ((uint8_t *)table_b)[i] = j;
-       }
-       break;
+            ((uint8_t *)table_b)[i] = j;
+        }
+        break;
     case 4:
     case 4|128:
-       table_start= table_121 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
+        table_start= table_121 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
 
-       entry_size = sizeof (uint8_t);
-       table_r = table_121 + 197;
-       table_b = table_121 + 197 + 685;
-       table_g = table_121 + 197 + 2*682;
+        entry_size = sizeof (uint8_t);
+        table_r = table_121 + 197;
+        table_b = table_121 + 197 + 685;
+        table_g = table_121 + 197 + 2*682;
 
-       for (i = -197; i < 256+197; i++) {
-           int j = table_Y[i+384 - 110] >> 7;
+        for (i = -197; i < 256+197; i++) {
+            int j = table_Y[i+384 - 110] >> 7;
 
-           if (isRgb)
-               j <<= 3;
+            if (isRgb)
+                j <<= 3;
 
-           ((uint8_t *)table_r)[i] = j;
-       }
-       for (i = -132; i < 256+132; i++) {
-           int j = (table_Y[i+384 - 37]+ 43)/85;
+            ((uint8_t *)table_r)[i] = j;
+        }
+        for (i = -132; i < 256+132; i++) {
+            int j = (table_Y[i+384 - 37]+ 43)/85;
 
-           ((uint8_t *)table_g)[i] = j << 1;
-       }
-       for (i = -232; i < 256+232; i++) {
-           int j =table_Y[i+384 - 110] >> 7;
+            ((uint8_t *)table_g)[i] = j << 1;
+        }
+        for (i = -232; i < 256+232; i++) {
+            int j =table_Y[i+384 - 110] >> 7;
 
-           if (!isRgb)
-               j <<= 3;
+            if (!isRgb)
+                j <<= 3;
 
-           ((uint8_t *)table_b)[i] = j;
-       }
-       break;
+            ((uint8_t *)table_b)[i] = j;
+        }
+        break;
 
     case 1:
-       table_start= table_1 = av_malloc (256*2 * sizeof (uint8_t));
+        table_start= table_1 = av_malloc (256*2 * sizeof (uint8_t));
 
-       entry_size = sizeof (uint8_t);
-       table_g = table_1;
-       table_r = table_b = NULL;
+        entry_size = sizeof (uint8_t);
+        table_g = table_1;
+        table_r = table_b = NULL;
 
-       for (i = 0; i < 256+256; i++) {
-           int j = table_Y[i + 384 - 110]>>7;
+        for (i = 0; i < 256+256; i++) {
+            int j = table_Y[i + 384 - 110]>>7;
 
-           ((uint8_t *)table_g)[i] = j;
-       }
-       break;
+            ((uint8_t *)table_g)[i] = j;
+        }
+        break;
 
     default:
-       table_start= NULL;
-       MSG_ERR("%ibpp not supported by yuv2rgb\n", bpp);
-       //free mem?
-       return -1;
+        table_start= NULL;
+        av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);
+        //free mem?
+        return -1;
     }
 
     for (i = 0; i < 256; i++) {
-       c->table_rV[i] = (uint8_t *)table_r + entry_size * div_round (crv * (i-128), 76309);
-       c->table_gU[i] = (uint8_t *)table_g + entry_size * div_round (cgu * (i-128), 76309);
-       c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
-       c->table_bU[i] = (uint8_t *)table_b + entry_size * div_round (cbu * (i-128), 76309);
+        c->table_rV[i] = (uint8_t *)table_r + entry_size * div_round (crv * (i-128), 76309);
+        c->table_gU[i] = (uint8_t *)table_g + entry_size * div_round (cgu * (i-128), 76309);
+        c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
+        c->table_bU[i] = (uint8_t *)table_b + entry_size * div_round (cbu * (i-128), 76309);
     }
 
     av_free(c->yuvTable);