Merge from 1.1.

2 files changed, 353 insertions, 139 deletions

diff --git a/src/combined/ffmpeg/ff_video_decoder.c b/src/combined/ffmpeg/ff_video_decoder.c
index c75848bee..b11967880 100644
--- a/src/combined/ffmpeg/ff_video_decoder.c
+++ b/src/combined/ffmpeg/ff_video_decoder.c
@@ -89,6 +89,7 @@ struct ff_video_decoder_s {
 
   xine_stream_t    *stream;
   int64_t           pts;
+  int64_t           last_pts;
   uint64_t          pts_tag_mask;
   uint64_t          pts_tag;
   int               pts_tag_counter;
@@ -422,6 +423,9 @@ static void init_video_codec (ff_video_decoder_t *this, unsigned int codec_type)
       break;
   }
 
+  /* dont want initial AV_NOPTS_VALUE here */
+  this->context->reordered_opaque = 0;
+
 }
 
 static void choose_speed_over_accuracy_cb(void *user_data, xine_cfg_entry_t *entry) {
@@ -1070,6 +1074,52 @@ static void ff_handle_special_buffer (ff_video_decoder_t *this, buf_element_t *b
   }
 }
 
+static uint64_t ff_tag_pts(ff_video_decoder_t *this, uint64_t pts)
+{
+  return pts | this->pts_tag;
+}
+
+static uint64_t ff_untag_pts(ff_video_decoder_t *this, uint64_t pts)
+{
+  if (this->pts_tag_mask == 0)
+    return pts; /* pts tagging inactive */
+
+  if (this->pts_tag != 0 && (pts & this->pts_tag_mask) != this->pts_tag)
+    return 0; /* reset pts if outdated while waiting for first pass (see below) */
+
+  return pts & ~this->pts_tag_mask;
+}
+
+static void ff_check_pts_tagging(ff_video_decoder_t *this, uint64_t pts)
+{
+  if (this->pts_tag_mask == 0)
+    return; /* pts tagging inactive */
+  if ((pts & this->pts_tag_mask) != this->pts_tag) {
+    this->pts_tag_stable_counter = 0;
+    return; /* pts still outdated */
+  }
+
+  /* the tag should be stable for 100 frames */
+  this->pts_tag_stable_counter++;
+
+  if (this->pts_tag != 0) {
+    if (this->pts_tag_stable_counter >= 100) {
+      /* first pass: reset pts_tag */
+      this->pts_tag = 0;
+      this->pts_tag_stable_counter = 0;
+    }
+  } else if (pts == 0)
+    return; /* cannot detect second pass */
+  else {
+    if (this->pts_tag_stable_counter >= 100) {
+      /* second pass: reset pts_tag_mask and pts_tag_counter */
+      this->pts_tag_mask = 0;
+      this->pts_tag_counter = 0;
+      this->pts_tag_stable_counter = 0;
+    }
+  }
+}
+
 static void ff_handle_mpeg12_buffer (ff_video_decoder_t *this, buf_element_t *buf) {
 
   vo_frame_t *img;
@@ -1091,6 +1141,13 @@ static void ff_handle_mpeg12_buffer (ff_video_decoder_t *this, buf_element_t *bu
     uint8_t *current;
     int next_flush;
 
+    /* apply valid pts to first frame _starting_ thereafter only */
+    if (this->pts && !this->context->reordered_opaque) {
+      this->context->reordered_opaque = 
+      this->av_frame->reordered_opaque = ff_tag_pts (this, this->pts);
+      this->pts = 0;
+    }
+
     got_picture = 0;
     if (!flush) {
       current = mpeg_parser_decode_data(this->mpeg_parser,
@@ -1172,8 +1229,11 @@ static void ff_handle_mpeg12_buffer (ff_video_decoder_t *this, buf_element_t *bu
         free_img = 0;
       }
 
-      img->pts  = this->pts;
-      this->pts = 0;
+      /* get back reordered pts */
+      img->pts = ff_untag_pts (this, this->av_frame->reordered_opaque);
+      ff_check_pts_tagging (this, this->av_frame->reordered_opaque);
+      this->av_frame->reordered_opaque = 0;
+      this->context->reordered_opaque = 0;
 
       if (this->av_frame->repeat_pict)
         img->duration = this->video_step * 3 / 2;
@@ -1214,52 +1274,6 @@ static void ff_handle_mpeg12_buffer (ff_video_decoder_t *this, buf_element_t *bu
   }
 }
 
-static uint64_t ff_tag_pts(ff_video_decoder_t *this, uint64_t pts)
-{
-  return pts | this->pts_tag;
-}
-
-static uint64_t ff_untag_pts(ff_video_decoder_t *this, uint64_t pts)
-{
-  if (this->pts_tag_mask == 0)
-    return pts; /* pts tagging inactive */
-
-  if (this->pts_tag != 0 && (pts & this->pts_tag_mask) != this->pts_tag)
-    return 0; /* reset pts if outdated while waiting for first pass (see below) */
-
-  return pts & ~this->pts_tag_mask;
-}
-
-static void ff_check_pts_tagging(ff_video_decoder_t *this, uint64_t pts)
-{
-  if (this->pts_tag_mask == 0)
-    return; /* pts tagging inactive */
-  if ((pts & this->pts_tag_mask) != this->pts_tag) {
-    this->pts_tag_stable_counter = 0;
-    return; /* pts still outdated */
-  }
-
-  /* the tag should be stable for 100 frames */
-  this->pts_tag_stable_counter++;
-
-  if (this->pts_tag != 0) {
-    if (this->pts_tag_stable_counter >= 100) {
-      /* first pass: reset pts_tag */
-      this->pts_tag = 0;
-      this->pts_tag_stable_counter = 0;
-    }
-  } else if (pts == 0)
-    return; /* cannot detect second pass */
-  else {
-    if (this->pts_tag_stable_counter >= 100) {
-      /* second pass: reset pts_tag_mask and pts_tag_counter */
-      this->pts_tag_mask = 0;
-      this->pts_tag_counter = 0;
-      this->pts_tag_stable_counter = 0;
-    }
-  }
-}
-
 static void ff_handle_buffer (ff_video_decoder_t *this, buf_element_t *buf) {
   uint8_t *chunk_buf = this->buf;
   AVRational avr00 = {0, 1};
@@ -1585,8 +1599,9 @@ static void ff_decode_data (video_decoder_t *this_gen, buf_element_t *buf) {
         ff_handle_preview_buffer(this, buf);
 
       /* decode */
-      if (buf->pts)
-	this->pts = buf->pts;
+      /* PES: each valid pts shall be used only once */
+      if (buf->pts && (buf->pts != this->last_pts))
+	this->last_pts = this->pts = buf->pts;
 
       if ((buf->type & 0xFFFF0000) == BUF_VIDEO_MPEG) {
 	ff_handle_mpeg12_buffer(this, buf);
diff --git a/src/xine-utils/color.c b/src/xine-utils/color.c
index 5c433ed59..930fb77d0 100644
--- a/src/xine-utils/color.c
+++ b/src/xine-utils/color.c
@@ -659,16 +659,31 @@ static void yuv411_to_yv12_c
 
 }
 
-#define C_YUV420_YUYV( )                                          \
-    *p_line1++ = *p_y1++; *p_line2++ = *p_y2++;                   \
-    *p_line1++ = *p_u;    *p_line2++ = (*p_u++ + *p_u2++)>>1;     \
-    *p_line1++ = *p_y1++; *p_line2++ = *p_y2++;                   \
-    *p_line1++ = *p_v;    *p_line2++ = (*p_v++ + *p_v2++)>>1;
+#define C_YUV420_YUYV_PROGRESSIVE( )                                          \
+    utmp = 3 * *p_u++;                                                        \
+    vtmp = 3 * *p_v++;                                                        \
+    *p_line1++ = *p_y1++;                *p_line2++ = *p_y2++;                \
+    *p_line1++ = (*p_ut++ + utmp) >> 2;  *p_line2++ = (utmp + *p_ub++) >> 2;  \
+    *p_line1++ = *p_y1++;                *p_line2++ = *p_y2++;                \
+    *p_line1++ = (*p_vt++ + vtmp) >> 2;  *p_line2++ = (vtmp + *p_vb++) >> 2;  \
+
+#define C_YUV420_YUYV_INTERLACED_ODD( )                                                   \
+    *p_line1++ = *p_y1++;                   *p_line2++ = *p_y2++;                         \
+    *p_line1++ = (*p_ut++ + *p_u * 7) >> 3; *p_line2++ = (*p_u++ * 5 + *p_ub++ * 3) >> 3; \
+    *p_line1++ = *p_y1++;                   *p_line2++ = *p_y2++;                         \
+    *p_line1++ = (*p_vt++ + *p_v * 7) >> 3; *p_line2++ = (*p_v++ * 5 + *p_vb++ * 3) >> 3; \
+
+#define C_YUV420_YUYV_INTERLACED_EVEN( )                                                  \
+    *p_line1++ = *p_y1++;                       *p_line2++ = *p_y2++;                     \
+    *p_line1++ = (*p_ut++ * 3 + *p_u * 5) >> 3; *p_line2++ = (*p_u++ * 7 + *p_ub++) >> 3; \
+    *p_line1++ = *p_y1++;                       *p_line2++ = *p_y2++;                     \
+    *p_line1++ = (*p_vt++ * 3 + *p_v * 5) >> 3; *p_line2++ = (*p_v++ * 7 + *p_vb++) >> 3; \
 
 /*****************************************************************************
  * I420_YUY2: planar YUV 4:2:0 to packed YUYV 4:2:2
  * original conversion routine from Videolan project
- * changed to support interlaced frames and use simple mean interpolation [MF]
+ * changed to support interlaced frames and do correct chroma upsampling with
+ * correct weighting factors and no chroma shift.
  *****************************************************************************/
 static void yv12_to_yuy2_c
   (const unsigned char *y_src, int y_src_pitch,
@@ -681,10 +696,12 @@ static void yv12_to_yuy2_c
     const uint8_t *p_y1, *p_y2 = y_src;
     const uint8_t *p_u = u_src;
     const uint8_t *p_v = v_src;
-    const uint8_t *p_u2 = u_src + u_src_pitch;
-    const uint8_t *p_v2 = v_src + v_src_pitch;
+    const uint8_t *p_ub, *p_vb;
+    const uint8_t *p_ut = u_src;
+    const uint8_t *p_vt = v_src;
 
     int i_x, i_y;
+    int utmp, vtmp;
 
     const int i_source_margin = y_src_pitch - width;
     const int i_source_u_margin = u_src_pitch - width/2;
@@ -702,28 +719,29 @@ static void yv12_to_yuy2_c
           p_y1 = p_y2;
           p_y2 += y_src_pitch;
 
+          if( i_y > 1 ) {
+            p_ub = p_u + u_src_pitch;
+            p_vb = p_v + v_src_pitch;
+          } else {
+            p_ub = p_u;
+            p_vb = p_v;
+          }
+
           for( i_x = width / 2 ; i_x-- ; )
           {
-              C_YUV420_YUYV( );
+              C_YUV420_YUYV_PROGRESSIVE( );
           }
 
           p_y2 += i_source_margin;
           p_u += i_source_u_margin;
           p_v += i_source_v_margin;
-          if( i_y > 1 ) {
-            p_u2 += i_source_u_margin;
-            p_v2 += i_source_v_margin;
-          } else {
-            p_u2 = p_u;
-            p_v2 = p_v;
-          }
+          p_ut = p_u - u_src_pitch;
+          p_vt = p_v - v_src_pitch;
           p_line2 += i_dest_margin;
       }
 
     } else {
 
-      p_u2 = u_src + 2*u_src_pitch;
-      p_v2 = v_src + 2*v_src_pitch;
       for( i_y = height / 4 ; i_y-- ; )
       {
           p_line1 = p_line2;
@@ -732,21 +750,24 @@ static void yv12_to_yuy2_c
           p_y1 = p_y2;
           p_y2 += 2 * y_src_pitch;
 
+          if( i_y > 1 ) {
+            p_ub = p_u + 2 * u_src_pitch;
+            p_vb = p_v + 2 * v_src_pitch;
+          } else {
+            p_ub = p_u;
+            p_vb = p_v;
+          }
+
           for( i_x = width / 2 ; i_x-- ; )
           {
-              C_YUV420_YUYV( );
+              C_YUV420_YUYV_INTERLACED_ODD( );
           }
 
           p_y2 += i_source_margin + y_src_pitch;
           p_u += i_source_u_margin + u_src_pitch;
           p_v += i_source_v_margin + v_src_pitch;
-          if( i_y > 1 ) {
-            p_u2 += i_source_u_margin + u_src_pitch;
-            p_v2 += i_source_v_margin + v_src_pitch;
-          } else {
-            p_u2 = p_u;
-            p_v2 = p_v;
-          }
+          p_ut = p_u - 2 * u_src_pitch;
+          p_vt = p_v - 2 * v_src_pitch;
           p_line2 += i_dest_margin + yuy2_pitch;
       }
 
@@ -754,8 +775,8 @@ static void yv12_to_yuy2_c
       p_y2 = y_src + y_src_pitch;
       p_u = u_src + u_src_pitch;
       p_v = v_src + v_src_pitch;
-      p_u2 = u_src + 3*u_src_pitch;
-      p_v2 = v_src + 3*v_src_pitch;
+      p_ut = p_u;
+      p_vt = p_v;
 
       for( i_y = height / 4 ; i_y-- ; )
       {
@@ -765,21 +786,24 @@ static void yv12_to_yuy2_c
           p_y1 = p_y2;
           p_y2 += 2 * y_src_pitch;
 
+          if( i_y > 1 ) {
+            p_ub = p_u + 2 * u_src_pitch;
+            p_vb = p_v + 2 * v_src_pitch;
+          } else {
+            p_ub = p_u;
+            p_vb = p_v;
+          }
+
           for( i_x = width / 2 ; i_x-- ; )
           {
-              C_YUV420_YUYV( );
+              C_YUV420_YUYV_INTERLACED_EVEN( );
           }
 
           p_y2 += i_source_margin + y_src_pitch;
           p_u += i_source_u_margin + u_src_pitch;
           p_v += i_source_v_margin + v_src_pitch;
-          if( i_y > 1 ) {
-            p_u2 += i_source_u_margin + u_src_pitch;
-            p_v2 += i_source_v_margin + v_src_pitch;
-          } else {
-            p_u2 = p_u;
-            p_v2 = p_v;
-          }
+          p_ut = p_u - 2 * u_src_pitch;
+          p_vt = p_v - 2 * v_src_pitch;
           p_line2 += i_dest_margin + yuy2_pitch;
       }
 
@@ -789,38 +813,204 @@ static void yv12_to_yuy2_c
 
 #if defined(ARCH_X86) || defined(ARCH_X86_64)
 
-#define MMXEXT_YUV420_YUYV( )                                                      \
+#define MMXEXT_YUV420_YUYV_PROGRESSIVE( )                                          \
 do {                                                                               \
    __asm__ __volatile__(".align 8 \n\t"                                            \
     "movq       (%0), %%mm0 \n\t"  /* Load 8 Y          y7 y6 y5 y4 y3 y2 y1 y0 */ \
     "movd       (%1), %%mm1 \n\t"  /* Load 4 Cb         00 00 00 00 u3 u2 u1 u0 */ \
     "movd       (%2), %%mm2 \n\t"  /* Load 4 Cr         00 00 00 00 v3 v2 v1 v0 */ \
-    "punpcklbw %%mm2, %%mm1 \n\t"  /*                   v3 u3 v2 u2 v1 u1 v0 u0 */ \
-    "movq      %%mm0, %%mm2 \n\t"  /*                   y7 y6 y5 y4 y3 y2 y1 y0 */ \
-    "punpcklbw %%mm1, %%mm2 \n\t"  /*                   v1 y3 u1 y2 v0 y1 u0 y0 */ \
+    "pxor      %%mm7, %%mm7 \n\t"  /*                   00 00 00 00 00 00 00 00 */ \
+    "punpcklbw %%mm7, %%mm1 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm2 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "movq      %%mm1, %%mm3 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "movq      %%mm2, %%mm4 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "psllw        $1, %%mm3 \n\t"  /* Cb * 2                                    */ \
+    "psllw        $1, %%mm4 \n\t"  /* Cr * 2                                    */ \
+    "paddw     %%mm3, %%mm1 \n\t"  /* Cb * 3                                    */ \
+    "paddw     %%mm4, %%mm2 \n\t"  /* Cr * 3                                    */ \
     :                                                                              \
     : "r" (p_y1), "r" (p_u), "r" (p_v) );                                          \
    __asm__ __volatile__(                                                           \
-    "movd       (%0), %%mm3 \n\t"  /* Load 4 Cb         00 00 00 00 u3 u2 u1 u0 */ \
-    "movd       (%1), %%mm4 \n\t"  /* Load 4 Cr         00 00 00 00 v3 v2 v1 v0 */ \
-    "punpcklbw %%mm4, %%mm3 \n\t"  /*                   v3 u3 v2 u2 v1 u1 v0 u0 */ \
-    "pavgb     %%mm1, %%mm3 \n\t"  /* (mean)            v3 u3 v2 u2 v1 u1 v0 u0 */ \
+    "movd       (%0), %%mm3 \n\t"  /* Load 4 Cbt        00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%1), %%mm4 \n\t"  /* Load 4 Crt        00 00 00 00 v3 v2 v1 v0 */ \
+    "movd       (%2), %%mm5 \n\t"  /* Load 4 Cbb        00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%3), %%mm6 \n\t"  /* Load 4 Crb        00 00 00 00 v3 v2 v1 v0 */ \
+    "punpcklbw %%mm7, %%mm3 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm4 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "punpcklbw %%mm7, %%mm5 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm6 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "paddw     %%mm1, %%mm3 \n\t"  /* Cb1 = Cbt + 3*Cb                          */ \
+    "paddw     %%mm2, %%mm4 \n\t"  /* Cr1 = Crt + 3*Cr                          */ \
+    "paddw     %%mm5, %%mm1 \n\t"  /* Cb2 = Cbb + 3*Cb                          */ \
+    "paddw     %%mm6, %%mm2 \n\t"  /* Cr2 = Crb + 3*Cr                          */ \
+    "psrlw        $2, %%mm3 \n\t"  /* Cb1 = (Cbt + 3*Cb) / 4                    */ \
+    /* either the shifts by 2 and 8 or mask off bits and shift by 6             */ \
+    "psrlw        $2, %%mm4 \n\t"  /* Cr1 = (Crt + 3*Cr) / 4                    */ \
+    "psllw        $8, %%mm4 \n\t"                                                  \
+    "por       %%mm4, %%mm3 \n\t"  /* Cr1 Cb1 interl    v3 u3 v2 u2 v1 u1 v0 u0 */ \
+    "psrlw        $2, %%mm1 \n\t"  /* Cb2 = (Cbb + 3*Cb) / 4                    */ \
+    "psrlw        $2, %%mm2 \n\t"  /* Cr2 = (Cbb + 3*Cb) / 4                    */ \
+    "psllw        $8, %%mm2 \n\t"                                                  \
+    "por       %%mm1, %%mm2 \n\t"  /* Cr2 Cb2 interl    v3 u3 v2 u2 v1 u1 v0 u0 */ \
+    "movq      %%mm0, %%mm1 \n\t"  /*                   y7 y6 y5 y4 y3 y2 y1 y0 */ \
+    "punpcklbw %%mm3, %%mm1 \n\t"  /*                   v1 y3 u1 y2 v0 y1 u0 y0 */ \
     :                                                                              \
-    : "r" (p_u2), "r" (p_v2) );                                                    \
+    : "r" (p_ut), "r" (p_vt), "r" (p_ub), "r" (p_vb) );                            \
    __asm__ __volatile__(                                                           \
-    "movntq    %%mm2, (%0)  \n\t"  /* Store low YUYV                            */ \
-    "punpckhbw %%mm1, %%mm0 \n\t"  /*                   v3 y7 u3 y6 v2 y5 u2 y4 */ \
-    "movntq    %%mm0, 8(%0) \n\t"  /* Store high YUYV                           */ \
+    "movntq    %%mm1, (%0)  \n\t"  /* Store low YUYV1                           */ \
+    "punpckhbw %%mm3, %%mm0 \n\t"  /*                   v3 y7 u3 y6 v2 y5 u2 y4 */ \
+    "movntq    %%mm0, 8(%0) \n\t"  /* Store high YUYV1                          */ \
     "movq       (%2), %%mm0 \n\t"  /* Load 8 Y          Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */ \
-    "movq      %%mm0, %%mm2 \n\t"  /*                   Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */ \
-    "punpcklbw %%mm3, %%mm2 \n\t"  /*                   v1 Y3 u1 Y2 v0 Y1 u0 Y0 */ \
-    "movntq    %%mm2, (%1)  \n\t"  /* Store low YUYV                            */ \
-    "punpckhbw %%mm3, %%mm0 \n\t"  /*                   v3 Y7 u3 Y6 v2 Y5 u2 Y4 */ \
-    "movntq    %%mm0, 8(%1) \n\t"  /* Store high YUYV                           */ \
+    "movq      %%mm0, %%mm1 \n\t"  /*                   Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */ \
+    "punpcklbw %%mm2, %%mm1 \n\t"  /*                   v1 Y3 u1 Y2 v0 Y1 u0 Y0 */ \
+    "movntq    %%mm1, (%1)  \n\t"  /* Store low YUYV2                           */ \
+    "punpckhbw %%mm2, %%mm0 \n\t"  /*                   v3 Y7 u3 Y6 v2 Y5 u2 Y4 */ \
+    "movntq    %%mm0, 8(%1) \n\t"  /* Store high YUYV2                          */ \
     :                                                                              \
     : "r" (p_line1),  "r" (p_line2),  "r" (p_y2) );                                \
   p_line1 += 16; p_line2 += 16; p_y1 += 8; p_y2 += 8; p_u += 4; p_v += 4;          \
-  p_u2 += 4; p_v2 += 4;                                                            \
+  p_ub += 4; p_vb += 4; p_ut += 4; p_vt += 4;                                      \
+} while(0)
+
+#define MMXEXT_YUV420_YUYV_INTERLACED_ODD( )                                       \
+do {                                                                               \
+   __asm__ __volatile__(".align 8 \n\t"                                            \
+    "movd       (%0), %%mm1 \n\t"  /* Load 4 Cb         00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%1), %%mm2 \n\t"  /* Load 4 Cr         00 00 00 00 v3 v2 v1 v0 */ \
+    "pxor      %%mm7, %%mm7 \n\t"  /*                   00 00 00 00 00 00 00 00 */ \
+    "punpcklbw %%mm7, %%mm1 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm2 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "movq      %%mm1, %%mm3 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "movq      %%mm2, %%mm4 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "psllw        $2, %%mm3 \n\t"  /* Cb * 4                                    */ \
+    "psllw        $2, %%mm4 \n\t"  /* Cr * 4                                    */ \
+    "paddw     %%mm3, %%mm1 \n\t"  /* Cb * 5                                    */ \
+    "paddw     %%mm4, %%mm2 \n\t"  /* Cr * 5                                    */ \
+    "psrlw        $1, %%mm3 \n\t"  /* Cb * 2                                    */ \
+    "psrlw        $1, %%mm4 \n\t"  /* Cr * 2                                    */ \
+    "paddw     %%mm1, %%mm3 \n\t"  /* Cb * 7                                    */ \
+    "paddw     %%mm2, %%mm4 \n\t"  /* Cr * 7                                    */ \
+    :                                                                              \
+    : "r" (p_u), "r" (p_v) );                                                      \
+   __asm__ __volatile__(                                                           \
+    "movd       (%1), %%mm5 \n\t"  /* Load 4 Cbt        00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%2), %%mm6 \n\t"  /* Load 4 Crt        00 00 00 00 v3 v2 v1 v0 */ \
+    "movq       (%0), %%mm0 \n\t"  /* Load 8 Y          y7 y6 y5 y4 y3 y2 y1 y0 */ \
+    "punpcklbw %%mm7, %%mm5 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm6 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "paddw     %%mm3, %%mm5 \n\t"  /* Cb1 = Cbt + 7*Cb                          */ \
+    "paddw     %%mm4, %%mm6 \n\t"  /* Cr1 = Crt + 7*Cr                          */ \
+    "psrlw        $3, %%mm5 \n\t"  /* Cb1 = (Cbt + 7*Cb) / 8                    */ \
+    /* either the shifts by 3 and 8 or mask off bits and shift by 5             */ \
+    "psrlw        $3, %%mm6 \n\t"  /* Cr1 = (Crt + 7*Cr) / 8                    */ \
+    "psllw        $8, %%mm6 \n\t"                                                  \
+    "por       %%mm5, %%mm6 \n\t"  /* Cr1 Cb1 interl    v3 u3 v2 u2 v1 u1 v0 u0 */ \
+    "movq      %%mm0, %%mm3 \n\t"  /*                   y7 y6 y5 y4 y3 y2 y1 y0 */ \
+    "punpcklbw %%mm6, %%mm3 \n\t"  /*                   v1 y3 u1 y2 v0 y1 u0 y0 */ \
+    "movntq    %%mm3, (%3)  \n\t"  /* Store low YUYV1                           */ \
+    "punpckhbw %%mm6, %%mm0 \n\t"  /*                   v3 y7 u3 y6 v2 y5 u2 y4 */ \
+    "movntq    %%mm0, 8(%3) \n\t"  /* Store high YUYV1                          */ \
+    :                                                                              \
+    : "r" (p_y1), "r" (p_ut), "r" (p_vt), "r" (p_line1) );                         \
+   __asm__ __volatile__(                                                           \
+    "movd       (%1), %%mm3 \n\t"  /* Load 4 Cbb        00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%2), %%mm4 \n\t"  /* Load 4 Crb        00 00 00 00 v3 v2 v1 v0 */ \
+    "movq       (%0), %%mm0 \n\t"  /* Load 8 Y          Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */ \
+    "punpcklbw %%mm7, %%mm3 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm4 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "movq      %%mm3, %%mm5 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "movq      %%mm4, %%mm6 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "psllw        $1, %%mm5 \n\t"  /* Cbb * 2                                   */ \
+    "psllw        $1, %%mm6 \n\t"  /* Crb * 2                                   */ \
+    "paddw     %%mm5, %%mm3 \n\t"  /* Cbb * 3                                   */ \
+    "paddw     %%mm6, %%mm4 \n\t"  /* Crb * 3                                   */ \
+    "paddw     %%mm3, %%mm1 \n\t"  /* Cb2 = 3*Cbb + 5*Cb                        */ \
+    "paddw     %%mm4, %%mm2 \n\t"  /* Cr2 = 3*Crb + 5*Cr                        */ \
+    "psrlw        $3, %%mm1 \n\t"  /* Cb2 = (3*Cbb + 5*Cb) / 8                  */ \
+    /* either the shifts by 3 and 8 or mask off bits and shift by 5             */ \
+    "psrlw        $3, %%mm2 \n\t"  /* Cr2 = (3*Crb + 5*Cr) / 8                  */ \
+    "psllw        $8, %%mm2 \n\t"                                                  \
+    "por       %%mm1, %%mm2 \n\t"  /* Cr2 Cb2 interl    v3 u3 v2 u2 v1 u1 v0 u0 */ \
+    "movq      %%mm0, %%mm1 \n\t"  /*                   Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */ \
+    "punpcklbw %%mm2, %%mm1 \n\t"  /*                   v1 Y3 u1 Y2 v0 Y1 u0 Y0 */ \
+    "movntq    %%mm1, (%3)  \n\t"  /* Store low YUYV2                           */ \
+    "punpckhbw %%mm2, %%mm0 \n\t"  /*                   v3 Y7 u3 Y6 v2 Y5 u2 Y4 */ \
+    "movntq    %%mm0, 8(%3) \n\t"  /* Store high YUYV2                          */ \
+    :                                                                              \
+    : "r" (p_y2),  "r" (p_ub), "r" (p_vb),  "r" (p_line2) );                       \
+  p_line1 += 16; p_line2 += 16; p_y1 += 8; p_y2 += 8; p_u += 4; p_v += 4;          \
+  p_ub += 4; p_vb += 4; p_ut += 4; p_vt += 4;                                      \
+} while(0)
+
+#define MMXEXT_YUV420_YUYV_INTERLACED_EVEN( )                                      \
+/* same as above, except the assembly input arguments are switched */              \
+do {                                                                               \
+   __asm__ __volatile__(".align 8 \n\t"                                            \
+    "movd       (%0), %%mm1 \n\t"  /* Load 4 Cb         00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%1), %%mm2 \n\t"  /* Load 4 Cr         00 00 00 00 v3 v2 v1 v0 */ \
+    "pxor      %%mm7, %%mm7 \n\t"  /*                   00 00 00 00 00 00 00 00 */ \
+    "punpcklbw %%mm7, %%mm1 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm2 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "movq      %%mm1, %%mm3 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "movq      %%mm2, %%mm4 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "psllw        $2, %%mm3 \n\t"  /* Cb * 4                                    */ \
+    "psllw        $2, %%mm4 \n\t"  /* Cr * 4                                    */ \
+    "paddw     %%mm3, %%mm1 \n\t"  /* Cb * 5                                    */ \
+    "paddw     %%mm4, %%mm2 \n\t"  /* Cr * 5                                    */ \
+    "psrlw        $1,  %%mm3 \n\t" /* Cb * 2                                    */ \
+    "psrlw        $1,  %%mm4 \n\t" /* Cr * 2                                    */ \
+    "paddw     %%mm1, %%mm3 \n\t"  /* Cb * 7                                    */ \
+    "paddw     %%mm2, %%mm4 \n\t"  /* Cr * 7                                    */ \
+    :                                                                              \
+    : "r" (p_u), "r" (p_v) );                                                      \
+   __asm__ __volatile__(                                                           \
+    "movd       (%1), %%mm5 \n\t"  /* Load 4 Cbt        00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%2), %%mm6 \n\t"  /* Load 4 Crt        00 00 00 00 v3 v2 v1 v0 */ \
+    "movq       (%0), %%mm0 \n\t"  /* Load 8 Y          y7 y6 y5 y4 y3 y2 y1 y0 */ \
+    "punpcklbw %%mm7, %%mm5 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm6 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "paddw     %%mm3, %%mm5 \n\t"  /* Cb1 = Cbt + 7*Cb                          */ \
+    "paddw     %%mm4, %%mm6 \n\t"  /* Cr1 = Crt + 7*Cr                          */ \
+    "psrlw        $3, %%mm5 \n\t"  /* Cb1 = (Cbt + 7*Cb) / 8                    */ \
+    /* either the shifts by 3 and 8 or mask off bits and shift by 5             */ \
+    "psrlw        $3, %%mm6 \n\t"  /* Cr1 = (Crt + 7*Cr) / 8                    */ \
+    "psllw        $8, %%mm6 \n\t"                                                  \
+    "por       %%mm5, %%mm6 \n\t"  /* Cr1 Cb1 interl    v3 u3 v2 u2 v1 u1 v0 u0 */ \
+    "movq      %%mm0, %%mm3 \n\t"  /*                   y7 y6 y5 y4 y3 y2 y1 y0 */ \
+    "punpcklbw %%mm6, %%mm3 \n\t"  /*                   v1 y3 u1 y2 v0 y1 u0 y0 */ \
+    "movntq    %%mm3, (%3)  \n\t"  /* Store low YUYV1                           */ \
+    "punpckhbw %%mm6, %%mm0 \n\t"  /*                   v3 y7 u3 y6 v2 y5 u2 y4 */ \
+    "movntq    %%mm0, 8(%3) \n\t"  /* Store high YUYV1                          */ \
+    :                                                                              \
+    : "r" (p_y2), "r" (p_ub), "r" (p_vb), "r" (p_line2) );                         \
+   __asm__ __volatile__(                                                           \
+    "movd       (%1), %%mm3 \n\t"  /* Load 4 Cbb        00 00 00 00 u3 u2 u1 u0 */ \
+    "movd       (%2), %%mm4 \n\t"  /* Load 4 Crb        00 00 00 00 v3 v2 v1 v0 */ \
+    "movq       (%0), %%mm0 \n\t"  /* Load 8 Y          Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */ \
+    "punpcklbw %%mm7, %%mm3 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "punpcklbw %%mm7, %%mm4 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "movq      %%mm3, %%mm5 \n\t"  /*                   00 u3 00 u2 00 u1 00 u0 */ \
+    "movq      %%mm4, %%mm6 \n\t"  /*                   00 v3 00 v2 00 v1 00 v0 */ \
+    "psllw        $1, %%mm5 \n\t"  /* Cbb * 2                                   */ \
+    "psllw        $1, %%mm6 \n\t"  /* Crb * 2                                   */ \
+    "paddw     %%mm5, %%mm3 \n\t"  /* Cbb * 3                                   */ \
+    "paddw     %%mm6, %%mm4 \n\t"  /* Crb * 3                                   */ \
+    "paddw     %%mm3, %%mm1 \n\t"  /* Cb2 = 3*Cbb + 5*Cb                        */ \
+    "paddw     %%mm4, %%mm2 \n\t"  /* Cr2 = 3*Crb + 5*Cr                        */ \
+    "psrlw        $3, %%mm1 \n\t"  /* Cb2 = (3*Cbb + 5*Cb) / 8                  */ \
+    /* either the shifts by 3 and 8 or mask off bits and shift by 5             */ \
+    "psrlw        $3, %%mm2 \n\t"  /* Cr2 = (3*Crb + 5*Cr) / 8                  */ \
+    "psllw        $8, %%mm2 \n\t"                                                  \
+    "por       %%mm1, %%mm2 \n\t"  /* Cr2 Cb2 interl    v3 u3 v2 u2 v1 u1 v0 u0 */ \
+    "movq      %%mm0, %%mm1 \n\t"  /*                   Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */ \
+    "punpcklbw %%mm2, %%mm1 \n\t"  /*                   v1 Y3 u1 Y2 v0 Y1 u0 Y0 */ \
+    "movntq    %%mm1, (%3)  \n\t"  /* Store low YUYV2                           */ \
+    "punpckhbw %%mm2, %%mm0 \n\t"  /*                   v3 Y7 u3 Y6 v2 Y5 u2 Y4 */ \
+    "movntq    %%mm0, 8(%3) \n\t"  /* Store high YUYV2                          */ \
+    :                                                                              \
+    : "r" (p_y1),  "r" (p_ut), "r" (p_vt),  "r" (p_line1) );                       \
+  p_line1 += 16; p_line2 += 16; p_y1 += 8; p_y2 += 8; p_u += 4; p_v += 4;          \
+  p_ub += 4; p_vb += 4; p_ut += 4; p_vt += 4;                                      \
 } while(0)
 
 #endif
@@ -836,10 +1026,12 @@ static void yv12_to_yuy2_mmxext
     const uint8_t *p_y1, *p_y2 = y_src;
     const uint8_t *p_u = u_src;
     const uint8_t *p_v = v_src;
-    const uint8_t *p_u2 = u_src + u_src_pitch;
-    const uint8_t *p_v2 = v_src + v_src_pitch;
+    const uint8_t *p_ub, *p_vb;
+    const uint8_t *p_ut = u_src;
+    const uint8_t *p_vt = v_src;
 
     int i_x, i_y;
+    int utmp, vtmp;
 
     const int i_source_margin = y_src_pitch - width;
     const int i_source_u_margin = u_src_pitch - width/2;
@@ -856,32 +1048,33 @@ static void yv12_to_yuy2_mmxext
           p_y1 = p_y2;
           p_y2 += y_src_pitch;
 
+          if( i_y > 1 ) {
+            p_ub = p_u + u_src_pitch;
+            p_vb = p_v + v_src_pitch;
+          } else {
+            p_ub = p_u;
+            p_vb = p_v;
+          }
+
           for( i_x = width / 8 ; i_x-- ; )
           {
-              MMXEXT_YUV420_YUYV( );
+              MMXEXT_YUV420_YUYV_PROGRESSIVE( );
           }
           for( i_x = (width % 8) / 2 ; i_x-- ; )
           {
-              C_YUV420_YUYV( );
+              C_YUV420_YUYV_PROGRESSIVE( );
           }
 
           p_y2 += i_source_margin;
           p_u += i_source_u_margin;
           p_v += i_source_v_margin;
-          if( i_y > 1 ) {
-            p_u2 += i_source_u_margin;
-            p_v2 += i_source_v_margin;
-          } else {
-            p_u2 = p_u;
-            p_v2 = p_v;
-          }
+          p_ut = p_u - u_src_pitch;
+          p_vt = p_v - v_src_pitch;
           p_line2 += i_dest_margin;
       }
 
     } else {
 
-      p_u2 = u_src + 2*u_src_pitch;
-      p_v2 = v_src + 2*v_src_pitch;
       for( i_y = height / 4 ; i_y-- ; )
       {
           p_line1 = p_line2;
@@ -890,25 +1083,28 @@ static void yv12_to_yuy2_mmxext
           p_y1 = p_y2;
           p_y2 += 2 * y_src_pitch;
 
+          if( i_y > 1 ) {
+            p_ub = p_u + 2 * u_src_pitch;
+            p_vb = p_v + 2 * v_src_pitch;
+          } else {
+            p_ub = p_u;
+            p_vb = p_v;
+          }
+
           for( i_x = width / 8 ; i_x-- ; )
           {
-              MMXEXT_YUV420_YUYV( );
+              MMXEXT_YUV420_YUYV_INTERLACED_ODD( );
           }
           for( i_x = (width % 8) / 2 ; i_x-- ; )
           {
-              C_YUV420_YUYV( );
+              C_YUV420_YUYV_INTERLACED_ODD( );
           }
 
           p_y2 += i_source_margin + y_src_pitch;
           p_u += i_source_u_margin + u_src_pitch;
           p_v += i_source_v_margin + v_src_pitch;
-          if( i_y > 1 ) {
-            p_u2 += i_source_u_margin + u_src_pitch;
-            p_v2 += i_source_v_margin + v_src_pitch;
-          } else {
-            p_u2 = p_u;
-            p_v2 = p_v;
-          }
+          p_ut = p_u - 2 * u_src_pitch;
+          p_vt = p_v - 2 * v_src_pitch;
           p_line2 += i_dest_margin + yuy2_pitch;
       }
 
@@ -916,9 +1112,9 @@ static void yv12_to_yuy2_mmxext
       p_y2 = y_src + y_src_pitch;
       p_u = u_src + u_src_pitch;
       p_v = v_src + v_src_pitch;
-      p_u2 = u_src + 3*u_src_pitch;
-      p_v2 = v_src + 3*v_src_pitch;
-
+      p_ut = p_u;
+      p_vt = p_v;
+ 
       for( i_y = height / 4 ; i_y-- ; )
       {
           p_line1 = p_line2;
@@ -927,25 +1123,28 @@ static void yv12_to_yuy2_mmxext
           p_y1 = p_y2;
           p_y2 += 2 * y_src_pitch;
 
+          if( i_y > 1 ) {
+            p_ub = p_u + 2 * u_src_pitch;
+            p_vb = p_v + 2 * v_src_pitch;
+          } else {
+            p_ub = p_u;
+            p_vb = p_v;
+          }
+
           for( i_x = width / 8 ; i_x-- ; )
           {
-              MMXEXT_YUV420_YUYV( );
+              MMXEXT_YUV420_YUYV_INTERLACED_EVEN( );
           }
           for( i_x = (width % 8) / 2 ; i_x-- ; )
           {
-              C_YUV420_YUYV( );
+              C_YUV420_YUYV_INTERLACED_EVEN( );
           }
 
           p_y2 += i_source_margin + y_src_pitch;
           p_u += i_source_u_margin + u_src_pitch;
           p_v += i_source_v_margin + v_src_pitch;
-          if( i_y > 1 ) {
-            p_u2 += i_source_u_margin + u_src_pitch;
-            p_v2 += i_source_v_margin + v_src_pitch;
-          } else {
-            p_u2 = p_u;
-            p_v2 = p_v;
-          }
+          p_ut = p_u - 2 * u_src_pitch;
+          p_vt = p_v - 2 * v_src_pitch;
           p_line2 += i_dest_margin + yuy2_pitch;
       }
 
@@ -1144,7 +1343,7 @@ void init_yuv_conversion(void) {
   else
     yv12_to_yuy2 = yv12_to_yuy2_c;
 
-  /* determine best YV12 -> YUY2 converter to use */
+  /* determine best YUY2 -> YV12 converter to use */
   if (xine_mm_accel() & MM_ACCEL_X86_MMXEXT)
     yuy2_to_yv12 = yuy2_to_yv12_mmxext;
   else