]> git.sesse.net Git - vlc/blobdiff - modules/video_chroma/i422_yuy2.c
Check-in modules/Makefile.am
[vlc] / modules / video_chroma / i422_yuy2.c
index c255079f318d7b669b90bab84ec0749d1802b307..315c542eafce298283355f0c0e70a1531224893a 100644 (file)
@@ -1,5 +1,5 @@
 /*****************************************************************************
- * i422_yuy2.c : YUV to YUV conversion module for vlc
+ * i422_yuy2.c : Planar YUV 4:2:2 to Packed YUV conversion module for vlc
  *****************************************************************************
  * Copyright (C) 2000, 2001 the VideoLAN team
  * $Id$
@@ -11,7 +11,7 @@
  * it under the terms of the GNU General Public License as published by
  * the Free Software Foundation; either version 2 of the License, or
  * (at your option) any later version.
- * 
+ *
  * This program is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 /*****************************************************************************
  * Preamble
  *****************************************************************************/
-#include <string.h>                                            /* strerror() */
-#include <stdlib.h>                                      /* malloc(), free() */
+
+#ifdef HAVE_CONFIG_H
+# include "config.h"
+#endif
 
 #include <vlc/vlc.h>
 #include <vlc_vout.h>
@@ -53,7 +55,6 @@ static void I422_cyuv           ( vout_thread_t *, picture_t *, picture_t * );
 #if defined (MODULE_NAME_IS_i422_yuy2)
 static void I422_Y211           ( vout_thread_t *, picture_t *, picture_t * );
 static void I422_Y211           ( vout_thread_t *, picture_t *, picture_t * );
-static void I422_YV12           ( vout_thread_t *, picture_t *, picture_t * );
 #endif
 
 /*****************************************************************************
@@ -70,7 +71,7 @@ vlc_module_begin();
 #elif defined (MODULE_NAME_IS_i422_yuy2_sse2)
     set_description( _("SSE2 conversions from " SRC_FOURCC " to " DEST_FOURCC) );
     set_capability( "chroma", 120 );
-    add_requirement( MMX );
+    add_requirement( SSE2 );
 #endif
     set_callbacks( Activate, NULL );
 vlc_module_end();
@@ -121,10 +122,6 @@ static int Activate( vlc_object_t *p_this )
                 case VLC_FOURCC('Y','2','1','1'):
                     p_vout->chroma.pf_convert = I422_Y211;
                     break;
-
-                case VLC_FOURCC('Y','V','1','2'):
-                    p_vout->chroma.pf_convert = I422_YV12;
-                    break;
 #endif
 
                 default:
@@ -163,7 +160,7 @@ static void I422_YUY2( vout_thread_t *p_vout, picture_t *p_source,
 #if defined (MODULE_NAME_IS_i422_yuy2_sse2)
 
     if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
-        ((int)p_line|(int)p_y))) )
+        ((intptr_t)p_line|(intptr_t)p_y))) )
     {
         /* use faster SSE2 aligned fetch and store */
         for( i_y = p_vout->render.i_height ; i_y-- ; )
@@ -256,7 +253,7 @@ static void I422_YVYU( vout_thread_t *p_vout, picture_t *p_source,
 #if defined (MODULE_NAME_IS_i422_yuy2_sse2)
 
     if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
-        ((int)p_line|(int)p_y))) )
+        ((intptr_t)p_line|(intptr_t)p_y))) )
     {
         /* use faster SSE2 aligned fetch and store */
         for( i_y = p_vout->render.i_height ; i_y-- ; )
@@ -349,7 +346,7 @@ static void I422_UYVY( vout_thread_t *p_vout, picture_t *p_source,
 #if defined (MODULE_NAME_IS_i422_yuy2_sse2)
 
     if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
-        ((int)p_line|(int)p_y))) )
+        ((intptr_t)p_line|(intptr_t)p_y))) )
     {
         /* use faster SSE2 aligned fetch and store */
         for( i_y = p_vout->render.i_height ; i_y-- ; )
@@ -425,6 +422,7 @@ static void I422_UYVY( vout_thread_t *p_vout, picture_t *p_source,
 static void I422_IUYV( vout_thread_t *p_vout, picture_t *p_source,
                                               picture_t *p_dest )
 {
+    VLC_UNUSED(p_source); VLC_UNUSED(p_dest);
     /* FIXME: TODO ! */
     msg_Err( p_vout, "I422_IUYV unimplemented, please harass <sam@zoy.org>" );
 }
@@ -442,6 +440,61 @@ static void I422_cyuv( vout_thread_t *p_vout, picture_t *p_source,
 
     int i_x, i_y;
 
+    const int i_source_margin = p_source->p[0].i_pitch
+                                 - p_source->p[0].i_visible_pitch;
+    const int i_source_margin_c = p_source->p[1].i_pitch
+                                 - p_source->p[1].i_visible_pitch;
+    const int i_dest_margin = p_dest->p->i_pitch
+                               - p_dest->p->i_visible_pitch;
+
+#if defined (MODULE_NAME_IS_i422_yuy2_sse2)
+
+    if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
+        ((intptr_t)p_line|(intptr_t)p_y))) )
+    {
+        /* use faster SSE2 aligned fetch and store */
+        for( i_y = p_vout->render.i_height ; i_y-- ; )
+        {
+            p_line -= 2 * p_dest->p->i_pitch;
+
+            for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+            {
+                SSE2_CALL( SSE2_YUV422_UYVY_ALIGNED );
+            }
+            for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+            {
+                C_YUV422_UYVY( p_line, p_y, p_u, p_v );
+            }
+            p_y += i_source_margin;
+            p_u += i_source_margin_c;
+            p_v += i_source_margin_c;
+            p_line += i_dest_margin;
+        }
+    }
+    else {
+        /* use slower SSE2 unaligned fetch and store */
+        for( i_y = p_vout->render.i_height ; i_y-- ; )
+        {
+            p_line -= 2 * p_dest->p->i_pitch;
+
+            for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+            {
+                SSE2_CALL( SSE2_YUV422_UYVY_UNALIGNED );
+            }
+            for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+            {
+                C_YUV422_UYVY( p_line, p_y, p_u, p_v );
+            }
+            p_y += i_source_margin;
+            p_u += i_source_margin_c;
+            p_v += i_source_margin_c;
+            p_line += i_dest_margin;
+        }
+    }
+    SSE2_END;
+
+#else
+
     for( i_y = p_vout->render.i_height ; i_y-- ; )
     {
         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
@@ -457,12 +510,18 @@ static void I422_cyuv( vout_thread_t *p_vout, picture_t *p_source,
             MMX_CALL( MMX_YUV422_UYVY );
 #endif
         }
+        p_y += i_source_margin;
+        p_u += i_source_margin_c;
+        p_v += i_source_margin_c;
+        p_line += i_dest_margin;
     }
 #if defined (MODULE_NAME_IS_i422_yuy2_mmx)
     MMX_END;
 #elif defined (MODULE_NAME_IS_i422_yuy2_sse2)
     SSE2_END;
 #endif
+
+#endif
 }
 
 /*****************************************************************************
@@ -489,36 +548,3 @@ static void I422_Y211( vout_thread_t *p_vout, picture_t *p_source,
     }
 }
 #endif
-
-
-/*****************************************************************************
- * I422_YV12: planar YUV 4:2:2 to planar YV12
- *****************************************************************************/
-#if defined (MODULE_NAME_IS_i422_yuy2)
-static void I422_YV12( vout_thread_t *p_vout, picture_t *p_source,
-                                              picture_t *p_dest )
-{
-    uint16_t i_dpy = p_dest->p[Y_PLANE].i_pitch;
-    uint16_t i_spy = p_source->p[Y_PLANE].i_pitch;
-    uint16_t i_dpuv = p_dest->p[U_PLANE].i_pitch;
-    uint16_t i_spuv = p_source->p[U_PLANE].i_pitch;
-    uint16_t i_width = p_vout->render.i_width;
-    uint16_t i_y = p_vout->render.i_height;
-    uint8_t *p_dy = p_dest->Y_PIXELS + (i_y-1)*i_dpy;
-    uint8_t *p_y = p_source->Y_PIXELS + (i_y-1)*i_spy;
-    uint8_t *p_du = p_dest->U_PIXELS + (i_y/2-1)*i_dpuv;
-    uint8_t *p_u = p_source->U_PIXELS + (i_y-1)*i_spuv;
-    uint8_t *p_dv = p_dest->V_PIXELS + (i_y/2-1)*i_dpuv;
-    uint8_t *p_v = p_source->V_PIXELS + (i_y-1)*i_spuv;
-    i_y /= 2;
-
-    for ( ; i_y--; )
-    {
-        memcpy(p_dy, p_y, i_width); p_dy -= i_dpy; p_y -= i_spy;
-        memcpy(p_dy, p_y, i_width); p_dy -= i_dpy; p_y -= i_spy;
-        memcpy(p_du, p_u, i_width/2); p_du -= i_dpuv; p_u -= 2*i_spuv;
-        memcpy(p_dv, p_v, i_width/2); p_dv -= i_dpuv; p_v -= 2*i_spuv;
-    }
-
-}
-#endif