git.sesse.net Git - vlc/blob - modules/video_chroma/i420_yuy2.c

   1 /*****************************************************************************
   2  * i420_yuy2.c : YUV to YUV conversion module for vlc
   3  *****************************************************************************
   4  * Copyright (C) 2000, 2001 VideoLAN
   5  * $Id$
   6  *
   7  * Authors: Samuel Hocevar <sam@zoy.org>
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with this program; if not, write to the Free Software
  21  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111, USA.
  22  *****************************************************************************/
  23
  24 /*****************************************************************************
  25  * Preamble
  26  *****************************************************************************/
  27 #include <string.h>                                            /* strerror() */
  28 #include <stdlib.h>                                      /* malloc(), free() */
  29
  30 #include <vlc/vlc.h>
  31 #include <vlc/vout.h>
  32
  33 #ifdef HAVE_ALTIVEC_H
  34 #   include <altivec.h>
  35 #endif
  36
  37 #include "i420_yuy2.h"
  38
  39 #define SRC_FOURCC  "I420,IYUV,YV12"
  40
  41 #if defined (MODULE_NAME_IS_i420_yuy2)
  42 #    define DEST_FOURCC "YUY2,YUNV,YVYU,UYVY,UYNV,Y422,IUYV,cyuv,Y211"
  43 #elif defined (MODULE_NAME_IS_i420_yuy2_mmx)
  44 #    define DEST_FOURCC "YUY2,YUNV,YVYU,UYVY,UYNV,Y422,IUYV,cyuv"
  45 #elif defined (MODULE_NAME_IS_i420_yuy2_altivec)
  46 #    define DEST_FOURCC "YUY2,YUNV"
  47 #endif
  48
  49 /*****************************************************************************
  50  * Local and extern prototypes.
  51  *****************************************************************************/
  52 static int  Activate ( vlc_object_t * );
  53
  54 static void I420_YUY2           ( vout_thread_t *, picture_t *, picture_t * );
  55 #if !defined (MODULE_NAME_IS_i420_yuy2_altivec)
  56 static void I420_YVYU           ( vout_thread_t *, picture_t *, picture_t * );
  57 static void I420_UYVY           ( vout_thread_t *, picture_t *, picture_t * );
  58 static void I420_IUYV           ( vout_thread_t *, picture_t *, picture_t * );
  59 static void I420_cyuv           ( vout_thread_t *, picture_t *, picture_t * );
  60 #endif
  61 #if defined (MODULE_NAME_IS_i420_yuy2)
  62 static void I420_Y211           ( vout_thread_t *, picture_t *, picture_t * );
  63 #endif
  64
  65 #ifdef MODULE_NAME_IS_i420_yuy2_mmx
  66 static uint64_t i_00ffw;
  67 static uint64_t i_80w;
  68 #endif
  69
  70 /*****************************************************************************
  71  * Module descriptor.
  72  *****************************************************************************/
  73 vlc_module_begin();
  74 #if defined (MODULE_NAME_IS_i420_yuy2)
  75     set_description( _("Conversions from " SRC_FOURCC " to " DEST_FOURCC) );
  76     set_capability( "chroma", 80 );
  77 #elif defined (MODULE_NAME_IS_i420_yuy2_mmx)
  78     set_description( _("MMX conversions from " SRC_FOURCC " to " DEST_FOURCC) );
  79     set_capability( "chroma", 100 );
  80     add_requirement( MMX );
  81     /* Initialize MMX-specific constants */
  82     i_00ffw = 0x00ff00ff00ff00ffULL;
  83     i_80w   = 0x0000000080808080ULL;
  84 #elif defined (MODULE_NAME_IS_i420_yuy2_altivec)
  85     set_description(
  86             _("AltiVec conversions from " SRC_FOURCC " to " DEST_FOURCC) );
  87     set_capability( "chroma", 100 );
  88     add_requirement( ALTIVEC );
  89 #endif
  90     set_callbacks( Activate, NULL );
  91 vlc_module_end();
  92
  93 /*****************************************************************************
  94  * Activate: allocate a chroma function
  95  *****************************************************************************
  96  * This function allocates and initializes a chroma function
  97  *****************************************************************************/
  98 static int Activate( vlc_object_t *p_this )
  99 {
 100     vout_thread_t *p_vout = (vout_thread_t *)p_this;
 101
 102     if( p_vout->render.i_width & 1 || p_vout->render.i_height & 1 )
 103     {
 104         return -1;
 105     }
 106
 107     switch( p_vout->render.i_chroma )
 108     {
 109         case VLC_FOURCC('Y','V','1','2'):
 110         case VLC_FOURCC('I','4','2','0'):
 111         case VLC_FOURCC('I','Y','U','V'):
 112             switch( p_vout->output.i_chroma )
 113             {
 114                 case VLC_FOURCC('Y','U','Y','2'):
 115                 case VLC_FOURCC('Y','U','N','V'):
 116                     p_vout->chroma.pf_convert = I420_YUY2;
 117                     break;
 118
 119 #if !defined (MODULE_NAME_IS_i420_yuy2_altivec)
 120                 case VLC_FOURCC('Y','V','Y','U'):
 121                     p_vout->chroma.pf_convert = I420_YVYU;
 122                     break;
 123
 124                 case VLC_FOURCC('U','Y','V','Y'):
 125                 case VLC_FOURCC('U','Y','N','V'):
 126                 case VLC_FOURCC('Y','4','2','2'):
 127                     p_vout->chroma.pf_convert = I420_UYVY;
 128                     break;
 129
 130                 case VLC_FOURCC('I','U','Y','V'):
 131                     p_vout->chroma.pf_convert = I420_IUYV;
 132                     break;
 133
 134                 case VLC_FOURCC('c','y','u','v'):
 135                     p_vout->chroma.pf_convert = I420_cyuv;
 136                     break;
 137 #endif
 138
 139 #if defined (MODULE_NAME_IS_i420_yuy2)
 140                 case VLC_FOURCC('Y','2','1','1'):
 141                     p_vout->chroma.pf_convert = I420_Y211;
 142                     break;
 143 #endif
 144
 145                 default:
 146                     return -1;
 147             }
 148             break;
 149
 150         default:
 151             return -1;
 152     }
 153
 154     return 0;
 155 }
 156
 157 /* Following functions are local */
 158
 159 /*****************************************************************************
 160  * I420_YUY2: planar YUV 4:2:0 to packed YUYV 4:2:2
 161  *****************************************************************************/
 162 static void I420_YUY2( vout_thread_t *p_vout, picture_t *p_source,
 163                                               picture_t *p_dest )
 164 {
 165     uint8_t *p_line1, *p_line2 = p_dest->p->p_pixels;
 166     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 167     uint8_t *p_u = p_source->U_PIXELS;
 168     uint8_t *p_v = p_source->V_PIXELS;
 169
 170     int i_x, i_y;
 171
 172 #if defined (MODULE_NAME_IS_i420_yuy2_altivec)
 173 #define VEC_NEXT_LINES( ) \
 174     p_line1  = p_line2; \
 175     p_line2 += p_dest->p->i_pitch; \
 176     p_y1     = p_y2; \
 177     p_y2    += p_source->p[Y_PLANE].i_pitch;
 178
 179 #define VEC_LOAD_UV( ) \
 180     u_vec = vec_ld( 0, p_u ); p_u += 16; \
 181     v_vec = vec_ld( 0, p_v ); p_v += 16;
 182
 183 #define VEC_MERGE( a ) \
 184     uv_vec = a( u_vec, v_vec ); \
 185     y_vec = vec_ld( 0, p_y1 ); p_y1 += 16; \
 186     vec_st( vec_mergeh( y_vec, uv_vec ), 0, p_line1 ); p_line1 += 16; \
 187     vec_st( vec_mergel( y_vec, uv_vec ), 0, p_line1 ); p_line1 += 16; \
 188     y_vec = vec_ld( 0, p_y2 ); p_y2 += 16; \
 189     vec_st( vec_mergeh( y_vec, uv_vec ), 0, p_line2 ); p_line2 += 16; \
 190     vec_st( vec_mergel( y_vec, uv_vec ), 0, p_line2 ); p_line2 += 16;
 191
 192     vector unsigned char u_vec;
 193     vector unsigned char v_vec;
 194     vector unsigned char uv_vec;
 195     vector unsigned char y_vec;
 196
 197     if( !( ( p_vout->render.i_width % 32 ) |
 198            ( p_vout->render.i_height % 2 ) ) )
 199     {
 200         /* Width is a multiple of 32, we take 2 lines at a time */
 201         for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 202         {
 203             VEC_NEXT_LINES( );
 204             for( i_x = p_vout->render.i_width / 32 ; i_x-- ; )
 205             {
 206                 VEC_LOAD_UV( );
 207                 VEC_MERGE( vec_mergeh );
 208                 VEC_MERGE( vec_mergel );
 209             }
 210         }
 211     }
 212     else if( !( ( p_vout->render.i_width % 16 ) |
 213                 ( p_vout->render.i_height % 4 ) ) )
 214     {
 215         /* Width is only a multiple of 16, we take 4 lines at a time */
 216         for( i_y = p_vout->render.i_height / 4 ; i_y-- ; )
 217         {
 218             /* Line 1 and 2, pixels 0 to ( width - 16 ) */
 219             VEC_NEXT_LINES( );
 220             for( i_x = p_vout->render.i_width / 32 ; i_x-- ; )
 221             {
 222                 VEC_LOAD_UV( );
 223                 VEC_MERGE( vec_mergeh );
 224                 VEC_MERGE( vec_mergel );
 225             }
 226
 227             /* Line 1 and 2, pixels ( width - 16 ) to ( width ) */
 228             VEC_LOAD_UV( );
 229             VEC_MERGE( vec_mergeh );
 230
 231             /* Line 3 and 4, pixels 0 to 16 */
 232             VEC_NEXT_LINES( );
 233             VEC_MERGE( vec_mergel );
 234
 235             /* Line 3 and 4, pixels 16 to ( width ) */
 236             for( i_x = p_vout->render.i_width / 32 ; i_x-- ; )
 237             {
 238                 VEC_LOAD_UV( );
 239                 VEC_MERGE( vec_mergeh );
 240                 VEC_MERGE( vec_mergel );
 241             }
 242         }
 243     }
 244     else
 245     {
 246         /* Crap, use the C version */
 247 #undef VEC_NEXT_LINES
 248 #undef VEC_LOAD_UV
 249 #undef VEC_MERGE
 250 #endif
 251
 252     const int i_source_margin = p_source->p->i_pitch
 253                                  - p_source->p->i_visible_pitch;
 254     const int i_dest_margin = p_dest->p->i_pitch
 255                                - p_dest->p->i_visible_pitch;
 256
 257     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 258     {
 259         p_line1 = p_line2;
 260         p_line2 += p_dest->p->i_pitch;
 261
 262         p_y1 = p_y2;
 263         p_y2 += p_source->p[Y_PLANE].i_pitch;
 264
 265 #if !defined (MODULE_NAME_IS_i420_yuy2_mmx)
 266         for( i_x = p_vout->render.i_width / 2 ; i_x-- ; )
 267         {
 268             C_YUV420_YUYV( );
 269         }
 270 #else
 271         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 272         {
 273             MMX_CALL( MMX_YUV420_YUYV );
 274         }
 275         for( i_x = ( p_vout->render.i_width % 8 ) / 2; i_x-- ; )
 276         {
 277             C_YUV420_YUYV( );
 278         }
 279 #endif
 280
 281         p_y1 += i_source_margin;
 282         p_y2 += i_source_margin;
 283         p_line1 += i_dest_margin;
 284         p_line2 += i_dest_margin;
 285     }
 286
 287 #if defined (MODULE_NAME_IS_i420_yuy2_altivec)
 288     }
 289 #endif
 290 }
 291
 292 /*****************************************************************************
 293  * I420_YVYU: planar YUV 4:2:0 to packed YVYU 4:2:2
 294  *****************************************************************************/
 295 #if !defined (MODULE_NAME_IS_i420_yuy2_altivec)
 296 static void I420_YVYU( vout_thread_t *p_vout, picture_t *p_source,
 297                                               picture_t *p_dest )
 298 {
 299     uint8_t *p_line1, *p_line2 = p_dest->p->p_pixels;
 300     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 301     uint8_t *p_u = p_source->U_PIXELS;
 302     uint8_t *p_v = p_source->V_PIXELS;
 303
 304     int i_x, i_y;
 305
 306     const int i_source_margin = p_source->p->i_pitch
 307                                  - p_source->p->i_visible_pitch;
 308     const int i_dest_margin = p_dest->p->i_pitch
 309                                - p_dest->p->i_visible_pitch;
 310
 311     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 312     {
 313         p_line1 = p_line2;
 314         p_line2 += p_dest->p->i_pitch;
 315
 316         p_y1 = p_y2;
 317         p_y2 += p_source->p[Y_PLANE].i_pitch;
 318
 319         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 320         {
 321 #if defined (MODULE_NAME_IS_i420_yuy2)
 322             C_YUV420_YVYU( );
 323             C_YUV420_YVYU( );
 324             C_YUV420_YVYU( );
 325             C_YUV420_YVYU( );
 326 #else
 327             MMX_CALL( MMX_YUV420_YVYU );
 328 #endif
 329         }
 330
 331         p_y1 += i_source_margin;
 332         p_y2 += i_source_margin;
 333         p_line1 += i_dest_margin;
 334         p_line2 += i_dest_margin;
 335     }
 336 }
 337
 338 /*****************************************************************************
 339  * I420_UYVY: planar YUV 4:2:0 to packed UYVY 4:2:2
 340  *****************************************************************************/
 341 static void I420_UYVY( vout_thread_t *p_vout, picture_t *p_source,
 342                                               picture_t *p_dest )
 343 {
 344     uint8_t *p_line1, *p_line2 = p_dest->p->p_pixels;
 345     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 346     uint8_t *p_u = p_source->U_PIXELS;
 347     uint8_t *p_v = p_source->V_PIXELS;
 348
 349     int i_x, i_y;
 350
 351     const int i_source_margin = p_source->p->i_pitch
 352                                  - p_source->p->i_visible_pitch;
 353     const int i_dest_margin = p_dest->p->i_pitch
 354                                - p_dest->p->i_visible_pitch;
 355
 356     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 357     {
 358         p_line1 = p_line2;
 359         p_line2 += p_dest->p->i_pitch;
 360
 361         p_y1 = p_y2;
 362         p_y2 += p_source->p[Y_PLANE].i_pitch;
 363
 364         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 365         {
 366 #if defined (MODULE_NAME_IS_i420_yuy2)
 367             C_YUV420_UYVY( );
 368             C_YUV420_UYVY( );
 369             C_YUV420_UYVY( );
 370             C_YUV420_UYVY( );
 371 #else
 372             MMX_CALL( MMX_YUV420_UYVY );
 373 #endif
 374         }
 375
 376         p_y1 += i_source_margin;
 377         p_y2 += i_source_margin;
 378         p_line1 += i_dest_margin;
 379         p_line2 += i_dest_margin;
 380     }
 381 }
 382
 383 /*****************************************************************************
 384  * I420_IUYV: planar YUV 4:2:0 to interleaved packed UYVY 4:2:2
 385  *****************************************************************************/
 386 static void I420_IUYV( vout_thread_t *p_vout, picture_t *p_source,
 387                                               picture_t *p_dest )
 388 {
 389     /* FIXME: TODO ! */
 390     msg_Err( p_vout, "I420_IUYV unimplemented, please harass <sam@zoy.org>" );
 391 }
 392
 393 /*****************************************************************************
 394  * I420_cyuv: planar YUV 4:2:0 to upside-down packed UYVY 4:2:2
 395  *****************************************************************************/
 396 static void I420_cyuv( vout_thread_t *p_vout, picture_t *p_source,
 397                                               picture_t *p_dest )
 398 {
 399     uint8_t *p_line1 = p_dest->p->p_pixels +
 400                        p_dest->p->i_visible_lines * p_dest->p->i_pitch
 401                        + p_dest->p->i_pitch;
 402     uint8_t *p_line2 = p_dest->p->p_pixels +
 403                        p_dest->p->i_visible_lines * p_dest->p->i_pitch;
 404     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 405     uint8_t *p_u = p_source->U_PIXELS;
 406     uint8_t *p_v = p_source->V_PIXELS;
 407
 408     int i_x, i_y;
 409
 410     const int i_source_margin = p_source->p->i_pitch
 411                                  - p_source->p->i_visible_pitch;
 412     const int i_dest_margin = p_dest->p->i_pitch
 413                                - p_dest->p->i_visible_pitch;
 414
 415     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 416     {
 417         p_line1 -= 3 * p_dest->p->i_pitch;
 418         p_line2 -= 3 * p_dest->p->i_pitch;
 419
 420         p_y1 = p_y2;
 421         p_y2 += p_source->p[Y_PLANE].i_pitch;
 422
 423         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 424         {
 425 #if defined (MODULE_NAME_IS_i420_yuy2)
 426             C_YUV420_UYVY( );
 427             C_YUV420_UYVY( );
 428             C_YUV420_UYVY( );
 429             C_YUV420_UYVY( );
 430 #else
 431             MMX_CALL( MMX_YUV420_UYVY );
 432 #endif
 433         }
 434
 435         p_y1 += i_source_margin;
 436         p_y2 += i_source_margin;
 437         p_line1 += i_dest_margin;
 438         p_line2 += i_dest_margin;
 439     }
 440 }
 441 #endif // !defined (MODULE_NAME_IS_i420_yuy2_altivec)
 442
 443 /*****************************************************************************
 444  * I420_Y211: planar YUV 4:2:0 to packed YUYV 2:1:1
 445  *****************************************************************************/
 446 #if defined (MODULE_NAME_IS_i420_yuy2)
 447 static void I420_Y211( vout_thread_t *p_vout, picture_t *p_source,
 448                                               picture_t *p_dest )
 449 {
 450     uint8_t *p_line1, *p_line2 = p_dest->p->p_pixels;
 451     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 452     uint8_t *p_u = p_source->U_PIXELS;
 453     uint8_t *p_v = p_source->V_PIXELS;
 454
 455     int i_x, i_y;
 456
 457     const int i_source_margin = p_source->p->i_pitch
 458                                  - p_source->p->i_visible_pitch;
 459     const int i_dest_margin = p_dest->p->i_pitch
 460                                - p_dest->p->i_visible_pitch;
 461
 462     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 463     {
 464         p_line1 = p_line2;
 465         p_line2 += p_dest->p->i_pitch;
 466
 467         p_y1 = p_y2;
 468         p_y2 += p_source->p[Y_PLANE].i_pitch;
 469
 470         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 471         {
 472             C_YUV420_Y211( );
 473             C_YUV420_Y211( );
 474         }
 475
 476         p_y1 += i_source_margin;
 477         p_y2 += i_source_margin;
 478         p_line1 += i_dest_margin;
 479         p_line2 += i_dest_margin;
 480     }
 481 }
 482 #endif
 483