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 the VideoLAN team
   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., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, 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 #if defined (MODULE_NAME_IS_i420_yuy2_altivec) && defined(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[0].i_pitch
 253                                  - p_source->p[0].i_visible_pitch;
 254     const int i_source_margin_c = p_source->p[1].i_pitch
 255                                  - p_source->p[1].i_visible_pitch;
 256     const int i_dest_margin = p_dest->p->i_pitch
 257                                - p_dest->p->i_visible_pitch;
 258
 259     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 260     {
 261         p_line1 = p_line2;
 262         p_line2 += p_dest->p->i_pitch;
 263
 264         p_y1 = p_y2;
 265         p_y2 += p_source->p[Y_PLANE].i_pitch;
 266
 267 #if !defined (MODULE_NAME_IS_i420_yuy2_mmx)
 268         for( i_x = p_vout->render.i_width / 2 ; i_x-- ; )
 269         {
 270             C_YUV420_YUYV( );
 271         }
 272 #else
 273         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 274         {
 275             MMX_CALL( MMX_YUV420_YUYV );
 276         }
 277         for( i_x = ( p_vout->render.i_width % 8 ) / 2; i_x-- ; )
 278         {
 279             C_YUV420_YUYV( );
 280         }
 281 #endif
 282
 283         p_y1 += i_source_margin;
 284         p_y2 += i_source_margin;
 285         p_u += i_source_margin_c;
 286         p_v += i_source_margin_c;
 287         p_line1 += i_dest_margin;
 288         p_line2 += i_dest_margin;
 289     }
 290
 291 #if defined (MODULE_NAME_IS_i420_yuy2_altivec)
 292     }
 293 #endif
 294 }
 295
 296 /*****************************************************************************
 297  * I420_YVYU: planar YUV 4:2:0 to packed YVYU 4:2:2
 298  *****************************************************************************/
 299 #if !defined (MODULE_NAME_IS_i420_yuy2_altivec)
 300 static void I420_YVYU( vout_thread_t *p_vout, picture_t *p_source,
 301                                               picture_t *p_dest )
 302 {
 303     uint8_t *p_line1, *p_line2 = p_dest->p->p_pixels;
 304     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 305     uint8_t *p_u = p_source->U_PIXELS;
 306     uint8_t *p_v = p_source->V_PIXELS;
 307
 308     int i_x, i_y;
 309
 310     const int i_source_margin = p_source->p[0].i_pitch
 311                                  - p_source->p[0].i_visible_pitch;
 312     const int i_source_margin_c = p_source->p[1].i_pitch
 313                                  - p_source->p[1].i_visible_pitch;
 314     const int i_dest_margin = p_dest->p->i_pitch
 315                                - p_dest->p->i_visible_pitch;
 316
 317     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 318     {
 319         p_line1 = p_line2;
 320         p_line2 += p_dest->p->i_pitch;
 321
 322         p_y1 = p_y2;
 323         p_y2 += p_source->p[Y_PLANE].i_pitch;
 324
 325         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 326         {
 327 #if defined (MODULE_NAME_IS_i420_yuy2)
 328             C_YUV420_YVYU( );
 329             C_YUV420_YVYU( );
 330             C_YUV420_YVYU( );
 331             C_YUV420_YVYU( );
 332 #else
 333             MMX_CALL( MMX_YUV420_YVYU );
 334 #endif
 335         }
 336
 337         p_y1 += i_source_margin;
 338         p_y2 += i_source_margin;
 339         p_u += i_source_margin_c;
 340         p_v += i_source_margin_c;
 341         p_line1 += i_dest_margin;
 342         p_line2 += i_dest_margin;
 343     }
 344 }
 345
 346 /*****************************************************************************
 347  * I420_UYVY: planar YUV 4:2:0 to packed UYVY 4:2:2
 348  *****************************************************************************/
 349 static void I420_UYVY( vout_thread_t *p_vout, picture_t *p_source,
 350                                               picture_t *p_dest )
 351 {
 352     uint8_t *p_line1, *p_line2 = p_dest->p->p_pixels;
 353     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 354     uint8_t *p_u = p_source->U_PIXELS;
 355     uint8_t *p_v = p_source->V_PIXELS;
 356
 357     int i_x, i_y;
 358
 359     const int i_source_margin = p_source->p[0].i_pitch
 360                                  - p_source->p[0].i_visible_pitch;
 361     const int i_source_margin_c = p_source->p[1].i_pitch
 362                                  - p_source->p[1].i_visible_pitch;
 363     const int i_dest_margin = p_dest->p->i_pitch
 364                                - p_dest->p->i_visible_pitch;
 365
 366     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 367     {
 368         p_line1 = p_line2;
 369         p_line2 += p_dest->p->i_pitch;
 370
 371         p_y1 = p_y2;
 372         p_y2 += p_source->p[Y_PLANE].i_pitch;
 373
 374         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 375         {
 376 #if defined (MODULE_NAME_IS_i420_yuy2)
 377             C_YUV420_UYVY( );
 378             C_YUV420_UYVY( );
 379             C_YUV420_UYVY( );
 380             C_YUV420_UYVY( );
 381 #else
 382             MMX_CALL( MMX_YUV420_UYVY );
 383 #endif
 384         }
 385
 386         p_y1 += i_source_margin;
 387         p_y2 += i_source_margin;
 388         p_u += i_source_margin_c;
 389         p_v += i_source_margin_c;
 390         p_line1 += i_dest_margin;
 391         p_line2 += i_dest_margin;
 392     }
 393
 394 #if defined (MODULE_NAME_IS_i420_yuy2_mmx)
 395     __asm__ __volatile__("emms" :: );
 396 #endif
 397 }
 398
 399 /*****************************************************************************
 400  * I420_IUYV: planar YUV 4:2:0 to interleaved packed UYVY 4:2:2
 401  *****************************************************************************/
 402 static void I420_IUYV( vout_thread_t *p_vout, picture_t *p_source,
 403                                               picture_t *p_dest )
 404 {
 405     /* FIXME: TODO ! */
 406     msg_Err( p_vout, "I420_IUYV unimplemented, please harass <sam@zoy.org>" );
 407 }
 408
 409 /*****************************************************************************
 410  * I420_cyuv: planar YUV 4:2:0 to upside-down packed UYVY 4:2:2
 411  *****************************************************************************/
 412 static void I420_cyuv( vout_thread_t *p_vout, picture_t *p_source,
 413                                               picture_t *p_dest )
 414 {
 415     uint8_t *p_line1 = p_dest->p->p_pixels +
 416                        p_dest->p->i_visible_lines * p_dest->p->i_pitch
 417                        + p_dest->p->i_pitch;
 418     uint8_t *p_line2 = p_dest->p->p_pixels +
 419                        p_dest->p->i_visible_lines * p_dest->p->i_pitch;
 420     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 421     uint8_t *p_u = p_source->U_PIXELS;
 422     uint8_t *p_v = p_source->V_PIXELS;
 423
 424     int i_x, i_y;
 425
 426     const int i_source_margin = p_source->p[0].i_pitch
 427                                  - p_source->p[0].i_visible_pitch;
 428     const int i_source_margin_c = p_source->p[1].i_pitch
 429                                  - p_source->p[1].i_visible_pitch;
 430     const int i_dest_margin = p_dest->p->i_pitch
 431                                - p_dest->p->i_visible_pitch;
 432
 433     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 434     {
 435         p_line1 -= 3 * p_dest->p->i_pitch;
 436         p_line2 -= 3 * p_dest->p->i_pitch;
 437
 438         p_y1 = p_y2;
 439         p_y2 += p_source->p[Y_PLANE].i_pitch;
 440
 441         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 442         {
 443 #if defined (MODULE_NAME_IS_i420_yuy2)
 444             C_YUV420_UYVY( );
 445             C_YUV420_UYVY( );
 446             C_YUV420_UYVY( );
 447             C_YUV420_UYVY( );
 448 #else
 449             MMX_CALL( MMX_YUV420_UYVY );
 450 #endif
 451         }
 452
 453         p_y1 += i_source_margin;
 454         p_y2 += i_source_margin;
 455         p_u += i_source_margin_c;
 456         p_v += i_source_margin_c;
 457         p_line1 += i_dest_margin;
 458         p_line2 += i_dest_margin;
 459     }
 460 }
 461 #endif // !defined (MODULE_NAME_IS_i420_yuy2_altivec)
 462
 463 /*****************************************************************************
 464  * I420_Y211: planar YUV 4:2:0 to packed YUYV 2:1:1
 465  *****************************************************************************/
 466 #if defined (MODULE_NAME_IS_i420_yuy2)
 467 static void I420_Y211( vout_thread_t *p_vout, picture_t *p_source,
 468                                               picture_t *p_dest )
 469 {
 470     uint8_t *p_line1, *p_line2 = p_dest->p->p_pixels;
 471     uint8_t *p_y1, *p_y2 = p_source->Y_PIXELS;
 472     uint8_t *p_u = p_source->U_PIXELS;
 473     uint8_t *p_v = p_source->V_PIXELS;
 474
 475     int i_x, i_y;
 476
 477     const int i_source_margin = p_source->p[0].i_pitch
 478                                  - p_source->p[0].i_visible_pitch;
 479     const int i_source_margin_c = p_source->p[1].i_pitch
 480                                  - p_source->p[1].i_visible_pitch;
 481     const int i_dest_margin = p_dest->p->i_pitch
 482                                - p_dest->p->i_visible_pitch;
 483
 484     for( i_y = p_vout->render.i_height / 2 ; i_y-- ; )
 485     {
 486         p_line1 = p_line2;
 487         p_line2 += p_dest->p->i_pitch;
 488
 489         p_y1 = p_y2;
 490         p_y2 += p_source->p[Y_PLANE].i_pitch;
 491
 492         for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
 493         {
 494             C_YUV420_Y211( );
 495             C_YUV420_Y211( );
 496         }
 497
 498         p_y1 += i_source_margin;
 499         p_y2 += i_source_margin;
 500         p_u += i_source_margin_c;
 501         p_v += i_source_margin_c;
 502         p_line1 += i_dest_margin;
 503         p_line2 += i_dest_margin;
 504     }
 505 }
 506 #endif
 507