Merge commit 'aa2ba8c99e5708884a56aea9c1d96e014866f8a3'

* commit 'aa2ba8c99e5708884a56aea9c1d96e014866f8a3':
  swscale: Move extern declarations for tables to swscale_internal.h

Conflicts:
	libswscale/output.c
	libswscale/swscale_internal.h
	libswscale/swscale_unscaled.c
	libswscale/utils.c
	libswscale/yuv2rgb.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

diff --cc libswscale/output.c
index 2d390c7,e1d01b0..ddb0d0c
--- 1/libswscale/output.c
--- 2/libswscale/output.c
+++ b/libswscale/output.c
@@@ -36,27 -36,24 +36,27 @@@
  #include "swscale.h"
  #include "swscale_internal.h"
  
- DECLARE_ALIGNED(8, const uint8_t, dither_2x2_4)[][8]={
 -DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_4)[2][8]={
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_2x2_4)[][8] = {
  {  1,   3,   1,   3,   1,   3,   1,   3, },
  {  2,   0,   2,   0,   2,   0,   2,   0, },
 +{  1,   3,   1,   3,   1,   3,   1,   3, },
  };
  
- DECLARE_ALIGNED(8, const uint8_t, dither_2x2_8)[][8]={
 -DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_8)[2][8]={
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_2x2_8)[][8] = {
  {  6,   2,   6,   2,   6,   2,   6,   2, },
  {  0,   4,   0,   4,   0,   4,   0,   4, },
 +{  6,   2,   6,   2,   6,   2,   6,   2, },
  };
  
- DECLARE_ALIGNED(8, const uint8_t, dither_4x4_16)[][8]={
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_4x4_16)[4][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_4x4_16)[][8] = {
  {  8,   4,  11,   7,   8,   4,  11,   7, },
  {  2,  14,   1,  13,   2,  14,   1,  13, },
  { 10,   6,   9,   5,  10,   6,   9,   5, },
  {  0,  12,   3,  15,   0,  12,   3,  15, },
 +{  8,   4,  11,   7,   8,   4,  11,   7, },
  };
  
- DECLARE_ALIGNED(8, const uint8_t, dither_8x8_32)[][8]={
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_32)[8][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_32)[][8] = {
  { 17,   9,  23,  15,  16,   8,  22,  14, },
  {  5,  29,   3,  27,   4,  28,   2,  26, },
  { 21,  13,  19,  11,  20,  12,  18,  10, },
@@@ -65,10 -62,9 +65,10 @@@
  {  4,  28,   2,  26,   5,  29,   3,  27, },
  { 20,  12,  18,  10,  21,  13,  19,  11, },
  {  1,  25,   7,  31,   0,  24,   6,  30, },
 +{ 17,   9,  23,  15,  16,   8,  22,  14, },
  };
  
- DECLARE_ALIGNED(8, const uint8_t, dither_8x8_73)[][8]={
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_73)[8][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_73)[][8] = {
  {  0,  55,  14,  68,   3,  58,  17,  72, },
  { 37,  18,  50,  32,  40,  22,  54,  35, },
  {  9,  64,   5,  59,  13,  67,   8,  63, },
@@@ -81,7 -76,7 +81,7 @@@
  };
  
  #if 1
- DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[][8]={
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[8][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
  {117,  62, 158, 103, 113,  58, 155, 100, },
  { 34, 199,  21, 186,  31, 196,  17, 182, },
  {144,  89, 131,  76, 141,  86, 127,  72, },
@@@ -94,7 -88,7 +94,7 @@@
  };
  #elif 1
  // tries to correct a gamma of 1.5
- DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[][8]={
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[8][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
  {  0, 143,  18, 200,   2, 156,  25, 215, },
  { 78,  28, 125,  64,  89,  36, 138,  74, },
  { 10, 180,   3, 161,  16, 195,   8, 175, },
@@@ -107,7 -100,7 +107,7 @@@
  };
  #elif 1
  // tries to correct a gamma of 2.0
- DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[][8]={
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[8][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
  {  0, 124,   8, 193,   0, 140,  12, 213, },
  { 55,  14, 104,  42,  66,  19, 119,  52, },
  {  3, 168,   1, 145,   6, 187,   3, 162, },
@@@ -120,7 -112,7 +120,7 @@@
  };
  #else
  // tries to correct a gamma of 2.5
- DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[][8]={
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[8][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
  {  0, 107,   3, 187,   0, 125,   6, 212, },
  { 39,   7,  86,  28,  49,  11, 102,  36, },
  {  1, 158,   0, 131,   3, 180,   1, 151, },
@@@ -329,10 -314,9 +329,10 @@@ yuv2mono_X_c_template(SwsContext *c, co
                        const int16_t **alpSrc, uint8_t *dest, int dstW,
                        int y, enum AVPixelFormat target)
  {
-     const uint8_t * const d128=dither_8x8_220[y&7];
+     const uint8_t * const d128 = ff_dither_8x8_220[y&7];
      int i;
      unsigned acc = 0;
 +    int err = 0;
  
      for (i = 0; i < dstW; i += 2) {
          int j;
@@@ -440,42 -387,20 +440,42 @@@ yuv2mono_1_c_template(SwsContext *c, co
                        const int16_t *abuf0, uint8_t *dest, int dstW,
                        int uvalpha, int y, enum AVPixelFormat target)
  {
-     const uint8_t * const d128 = dither_8x8_220[y & 7];
+     const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
      int i;
  
 +    if (c->dither == SWS_DITHER_ED) {
 +        int err = 0;
 +        int acc = 0;
 +        for (i = 0; i < dstW; i +=2) {
 +            int Y;
 +
 +            Y = ((buf0[i + 0] + 64) >> 7);
 +            Y += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
 +            c->dither_error[0][i] = err;
 +            acc = 2*acc + (Y >= 128);
 +            Y -= 220*(acc&1);
 +
 +            err = ((buf0[i + 1] + 64) >> 7);
 +            err += (7*Y + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4;
 +            c->dither_error[0][i+1] = Y;
 +            acc = 2*acc + (err >= 128);
 +            err -= 220*(acc&1);
 +
 +            if ((i & 7) == 6)
 +                output_pixel(*dest++, acc);
 +        }
 +        c->dither_error[0][i] = err;
 +    } else {
      for (i = 0; i < dstW; i += 8) {
          int acc = 0;
 -
 -        accumulate_bit(acc, (buf0[i + 0] >> 7) + d128[0]);
 -        accumulate_bit(acc, (buf0[i + 1] >> 7) + d128[1]);
 -        accumulate_bit(acc, (buf0[i + 2] >> 7) + d128[2]);
 -        accumulate_bit(acc, (buf0[i + 3] >> 7) + d128[3]);
 -        accumulate_bit(acc, (buf0[i + 4] >> 7) + d128[4]);
 -        accumulate_bit(acc, (buf0[i + 5] >> 7) + d128[5]);
 -        accumulate_bit(acc, (buf0[i + 6] >> 7) + d128[6]);
 -        accumulate_bit(acc, (buf0[i + 7] >> 7) + d128[7]);
 +        accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
 +        accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
 +        accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
 +        accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
 +        accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
 +        accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
 +        accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
 +        accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
  
          output_pixel(*dest++, acc);
      }
@@@ -1188,26 -866,26 +1188,26 @@@ yuv2rgb_write(uint8_t *_dest, int i, in
          int dr1, dg1, db1, dr2, dg2, db2;
  
          if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
--            dr1 = dither_2x2_8[ y & 1     ][0];
--            dg1 = dither_2x2_4[ y & 1     ][0];
--            db1 = dither_2x2_8[(y & 1) ^ 1][0];
--            dr2 = dither_2x2_8[ y & 1     ][1];
--            dg2 = dither_2x2_4[ y & 1     ][1];
--            db2 = dither_2x2_8[(y & 1) ^ 1][1];
++            dr1 = ff_dither_2x2_8[ y & 1     ][0];
++            dg1 = ff_dither_2x2_4[ y & 1     ][0];
++            db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
++            dr2 = ff_dither_2x2_8[ y & 1     ][1];
++            dg2 = ff_dither_2x2_4[ y & 1     ][1];
++            db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
          } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
--            dr1 = dither_2x2_8[ y & 1     ][0];
--            dg1 = dither_2x2_8[ y & 1     ][1];
--            db1 = dither_2x2_8[(y & 1) ^ 1][0];
--            dr2 = dither_2x2_8[ y & 1     ][1];
--            dg2 = dither_2x2_8[ y & 1     ][0];
--            db2 = dither_2x2_8[(y & 1) ^ 1][1];
++            dr1 = ff_dither_2x2_8[ y & 1     ][0];
++            dg1 = ff_dither_2x2_8[ y & 1     ][1];
++            db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
++            dr2 = ff_dither_2x2_8[ y & 1     ][1];
++            dg2 = ff_dither_2x2_8[ y & 1     ][0];
++            db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
          } else {
-             dr1 = dither_4x4_16[ y & 3     ][0];
-             dg1 = dither_4x4_16[ y & 3     ][1];
-             db1 = dither_4x4_16[(y & 3) ^ 3][0];
-             dr2 = dither_4x4_16[ y & 3     ][1];
-             dg2 = dither_4x4_16[ y & 3     ][0];
-             db2 = dither_4x4_16[(y & 3) ^ 3][1];
+             dr1 = ff_dither_4x4_16[ y & 3     ][0];
+             dg1 = ff_dither_4x4_16[ y & 3     ][1];
+             db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
+             dr2 = ff_dither_4x4_16[ y & 3     ][1];
+             dg2 = ff_dither_4x4_16[ y & 3     ][0];
+             db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
          }
  
          dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];

diff --cc libswscale/swscale.c
index db41602,3a63fbe..1ca6934
--- 1/libswscale/swscale.c
--- 2/libswscale/swscale.c
+++ b/libswscale/swscale.c
@@@ -35,7 -35,7 +35,7 @@@
  #include "swscale_internal.h"
  #include "swscale.h"
  
- DECLARE_ALIGNED(8, const uint8_t, dither_8x8_128)[8][8] = {
 -DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_128)[8][8] = {
++DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_128)[][8] = {
      {  36, 68,  60, 92,  34, 66,  58, 90, },
      { 100,  4, 124, 28,  98,  2, 122, 26, },
      {  52, 84,  44, 76,  50, 82,  42, 74, },

diff --cc libswscale/swscale_internal.h
index 1b05461,e8245b9..8f614d3
--- 1/libswscale/swscale_internal.h
--- 2/libswscale/swscale_internal.h
+++ b/libswscale/swscale_internal.h
@@@ -815,6 -720,13 +815,15 @@@ static av_always_inline int usePal(enu
  extern const uint64_t ff_dither4[2];
  extern const uint64_t ff_dither8[2];
  
 -extern const uint8_t ff_dither_4x4_16[4][8];
 -extern const uint8_t ff_dither_8x8_32[8][8];
 -extern const uint8_t ff_dither_8x8_73[8][8];
++extern const uint8_t ff_dither_2x2_4[3][8];
++extern const uint8_t ff_dither_2x2_8[3][8];
++extern const uint8_t ff_dither_4x4_16[5][8];
++extern const uint8_t ff_dither_8x8_32[9][8];
++extern const uint8_t ff_dither_8x8_73[9][8];
+ extern const uint8_t ff_dither_8x8_128[8][8];
 -extern const uint8_t ff_dither_8x8_220[8][8];
++extern const uint8_t ff_dither_8x8_220[9][8];
+ 
+ extern const int32_t ff_yuv2rgb_coeffs[8][4];
  
  extern const AVClass sws_context_class;
  

diff --cc libswscale/utils.c
index f5826c3,408859d..3303dcb
--- 1/libswscale/utils.c
--- 2/libswscale/utils.c
+++ b/libswscale/utils.c
@@@ -226,9 -194,6 +226,7 @@@ int sws_isSupportedEndiannessConversion
             format_entries[pix_fmt].is_supported_endianness : 0;
  }
  
- extern const int32_t ff_yuv2rgb_coeffs[8][4];
- 
 +#if FF_API_SWS_FORMAT_NAME
  const char *sws_format_name(enum AVPixelFormat format)
  {
      const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(format);

diff --cc libswscale/yuv2rgb.c
index 6d54293,b510446..e447056
--- 1/libswscale/yuv2rgb.c
--- 2/libswscale/yuv2rgb.c
+++ b/libswscale/yuv2rgb.c
@@@ -33,15 -34,7 +33,8 @@@
  #include "rgb2rgb.h"
  #include "swscale.h"
  #include "swscale_internal.h"
 +#include "libavutil/pixdesc.h"
  
- extern const uint8_t dither_2x2_4[3][8];
- extern const uint8_t dither_2x2_8[3][8];
- extern const uint8_t dither_4x4_16[5][8];
- extern const uint8_t dither_8x8_32[9][8];
- extern const uint8_t dither_8x8_73[9][8];
- extern const uint8_t dither_8x8_220[9][8];
- 
  const int32_t ff_yuv2rgb_coeffs[8][4] = {
      { 117504, 138453, 13954, 34903 }, /* no sequence_display_extension */
      { 117504, 138453, 13954, 34903 }, /* ITU-R Rec. 709 (1990) */
@@@ -389,65 -382,24 +382,65 @@@ ENDYUV2RGBLINE(24, 1
      PUTBGR24(dst_2, py_2, 0);
  ENDYUV2RGBFUNC()
  
 -// This is exactly the same code as yuv2rgb_c_32 except for the types of
 -// r, g, b, dst_1, dst_2
 -YUV2RGBFUNC(yuv2rgb_c_16, uint16_t, 0)
 +YUV2RGBFUNC(yuv2rgb_c_16_ordered_dither, uint16_t, 0)
-     const uint8_t *d16 = dither_2x2_8[y & 1];
-     const uint8_t *e16 = dither_2x2_4[y & 1];
-     const uint8_t *f16 = dither_2x2_8[(y & 1)^1];
++    const uint8_t *d16 = ff_dither_2x2_8[y & 1];
++    const uint8_t *e16 = ff_dither_2x2_4[y & 1];
++    const uint8_t *f16 = ff_dither_2x2_8[(y & 1)^1];
 +
 +#define PUTRGB16(dst, src, i, o)                    \
 +    Y              = src[2 * i];                    \
 +    dst[2 * i]     = r[Y + d16[0 + o]] +            \
 +                     g[Y + e16[0 + o]] +            \
 +                     b[Y + f16[0 + o]];             \
 +    Y              = src[2 * i + 1];                \
 +    dst[2 * i + 1] = r[Y + d16[1 + o]] +            \
 +                     g[Y + e16[1 + o]] +            \
 +                     b[Y + f16[1 + o]];
      LOADCHROMA(0);
 -    PUTRGB(dst_1, py_1, 0);
 -    PUTRGB(dst_2, py_2, 0);
 +    PUTRGB16(dst_1, py_1, 0, 0);
 +    PUTRGB16(dst_2, py_2, 0, 0 + 8);
  
      LOADCHROMA(1);
 -    PUTRGB(dst_2, py_2, 1);
 -    PUTRGB(dst_1, py_1, 1);
 +    PUTRGB16(dst_2, py_2, 1, 2 + 8);
 +    PUTRGB16(dst_1, py_1, 1, 2);
  
      LOADCHROMA(2);
 -    PUTRGB(dst_1, py_1, 2);
 -    PUTRGB(dst_2, py_2, 2);
 +    PUTRGB16(dst_1, py_1, 2, 4);
 +    PUTRGB16(dst_2, py_2, 2, 4 + 8);
  
      LOADCHROMA(3);
 -    PUTRGB(dst_2, py_2, 3);
 -    PUTRGB(dst_1, py_1, 3);
 +    PUTRGB16(dst_2, py_2, 3, 6 + 8);
 +    PUTRGB16(dst_1, py_1, 3, 6);
 +CLOSEYUV2RGBFUNC(8)
 +
 +YUV2RGBFUNC(yuv2rgb_c_15_ordered_dither, uint16_t, 0)
-     const uint8_t *d16 = dither_2x2_8[y & 1];
-     const uint8_t *e16 = dither_2x2_8[(y & 1)^1];
++    const uint8_t *d16 = ff_dither_2x2_8[y & 1];
++    const uint8_t *e16 = ff_dither_2x2_8[(y & 1)^1];
 +
 +#define PUTRGB15(dst, src, i, o)                    \
 +    Y              = src[2 * i];                    \
 +    dst[2 * i]     = r[Y + d16[0 + o]] +            \
 +                     g[Y + d16[1 + o]] +            \
 +                     b[Y + e16[0 + o]];             \
 +    Y              = src[2 * i + 1];                \
 +    dst[2 * i + 1] = r[Y + d16[1 + o]] +            \
 +                     g[Y + d16[0 + o]] +            \
 +                     b[Y + e16[1 + o]];
 +    LOADCHROMA(0);
 +    PUTRGB15(dst_1, py_1, 0, 0);
 +    PUTRGB15(dst_2, py_2, 0, 0 + 8);
 +
 +    LOADCHROMA(1);
 +    PUTRGB15(dst_2, py_2, 1, 2 + 8);
 +    PUTRGB15(dst_1, py_1, 1, 2);
 +
 +    LOADCHROMA(2);
 +    PUTRGB15(dst_1, py_1, 2, 4);
 +    PUTRGB15(dst_2, py_2, 2, 4 + 8);
 +
 +    LOADCHROMA(3);
 +    PUTRGB15(dst_2, py_2, 3, 6 + 8);
 +    PUTRGB15(dst_1, py_1, 3, 6);
  CLOSEYUV2RGBFUNC(8)
  
  // r, g, b, dst_1, dst_2
@@@ -578,9 -530,9 +571,9 @@@ YUV2RGBFUNC(yuv2rgb_c_4b_ordered_dither
  CLOSEYUV2RGBFUNC(8)
  
  YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0)
-     const uint8_t *d128 = dither_8x8_220[y & 7];
+     const uint8_t *d128 = ff_dither_8x8_220[y & 7];
      char out_1 = 0, out_2 = 0;
 -    g = c->table_gU[128] + c->table_gV[128];
 +    g = c->table_gU[128 + YUVRGB_TABLE_HEADROOM] + c->table_gV[128 + YUVRGB_TABLE_HEADROOM];
  
  #define PUTRGB1(out, src, i, o)                     \
      Y    = src[2 * i];                              \