git.sesse.net Git - vlc/blob - modules/video_chroma/i420_ymga.c

   1 /*****************************************************************************
   2  * i420_ymga.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
  28 #ifdef HAVE_CONFIG_H
  29 # include "config.h"
  30 #endif
  31
  32 #include <errno.h>                                                 /* ENOMEM */
  33
  34 #include <vlc_common.h>
  35 #include <vlc_plugin.h>
  36 #include <vlc_filter.h>
  37 #include <vlc_vout.h>
  38
  39 #define SRC_FOURCC  "I420,IYUV,YV12"
  40 #define DEST_FOURCC "YMGA"
  41
  42 /*****************************************************************************
  43  * Local and extern prototypes.
  44  *****************************************************************************/
  45 static int  Activate   ( vlc_object_t * );
  46 static void I420_YMGA  ( filter_t *, picture_t *, picture_t * );
  47 static picture_t *I420_YMGA_Filter( filter_t *, picture_t * );
  48
  49 /*****************************************************************************
  50  * Module descriptor
  51  *****************************************************************************/
  52 vlc_module_begin ()
  53 #if defined (MODULE_NAME_IS_i420_ymga)
  54     set_description( N_("Conversions from " SRC_FOURCC " to " DEST_FOURCC) )
  55     set_capability( "video filter2", 80 )
  56 #elif defined (MODULE_NAME_IS_i420_ymga_mmx)
  57     set_description( N_("MMX conversions from " SRC_FOURCC " to " DEST_FOURCC) )
  58     set_capability( "video filter2", 100 )
  59     add_requirement( MMX )
  60 #endif
  61     set_callbacks( Activate, NULL )
  62 vlc_module_end ()
  63
  64 /*****************************************************************************
  65  * Activate: allocate a chroma function
  66  *****************************************************************************
  67  * This function allocates and initializes a chroma function
  68  *****************************************************************************/
  69 static int Activate( vlc_object_t *p_this )
  70 {
  71     filter_t *p_filter = (filter_t *)p_this;
  72
  73     if( p_filter->fmt_in.video.i_width & 1
  74      || p_filter->fmt_in.video.i_height & 1 )
  75     {
  76         return -1;
  77     }
  78
  79     if( p_filter->fmt_in.video.i_width != p_filter->fmt_out.video.i_width
  80      || p_filter->fmt_in.video.i_height != p_filter->fmt_out.video.i_height )
  81         return -1;
  82
  83     switch( p_filter->fmt_in.video.i_chroma )
  84     {
  85         case VLC_FOURCC('Y','V','1','2'):
  86         case VLC_FOURCC('I','4','2','0'):
  87         case VLC_FOURCC('I','Y','U','V'):
  88             switch( p_filter->fmt_out.video.i_chroma )
  89             {
  90                 case VLC_FOURCC('Y','M','G','A'):
  91                     p_filter->pf_video_filter = I420_YMGA_Filter;
  92                     break;
  93
  94                 default:
  95                     return -1;
  96             }
  97             break;
  98
  99         default:
 100             return -1;
 101     }
 102
 103     return 0;
 104 }
 105
 106 /* Following functions are local */
 107
 108 VIDEO_FILTER_WRAPPER( I420_YMGA )
 109
 110 /*****************************************************************************
 111  * I420_YMGA: planar YUV 4:2:0 to Matrox's planar/packed YUV 4:2:0
 112  *****************************************************************************/
 113 static void I420_YMGA( filter_t *p_filter, picture_t *p_source,
 114                                            picture_t *p_dest )
 115 {
 116     uint8_t *p_uv = p_dest->U_PIXELS;
 117     uint8_t *p_u = p_source->U_PIXELS;
 118     uint8_t *p_v = p_source->V_PIXELS;
 119
 120     int i_x;
 121
 122     /* Copy the Y part */
 123     vlc_memcpy( p_dest->Y_PIXELS, p_source->Y_PIXELS,
 124                 p_dest->p[Y_PLANE].i_pitch * p_dest->p[Y_PLANE].i_visible_lines );
 125
 126     /* Copy the U:V part */
 127     for( i_x = p_dest->p[U_PLANE].i_pitch * p_dest->p[U_PLANE].i_visible_lines / 64;
 128          i_x--; )
 129     {
 130 #if defined (MODULE_NAME_IS_i420_ymga)
 131         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 132         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 133         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 134         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 135         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 136         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 137         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 138         *p_uv++ = *p_u++; *p_uv++ = *p_v++; *p_uv++ = *p_u++; *p_uv++ = *p_v++;
 139 #else
 140         __asm__( ".p2align 5 \n\
 141         movd       (%0), %%mm0  # Load 4 Cr   00 00 00 00 v3 v2 v1 v0     \n\
 142         movd      4(%0), %%mm2  # Load 4 Cr   00 00 00 00 v3 v2 v1 v0     \n\
 143         movd      8(%0), %%mm4  # Load 4 Cr   00 00 00 00 v3 v2 v1 v0     \n\
 144         movd     12(%0), %%mm6  # Load 4 Cr   00 00 00 00 v3 v2 v1 v0     \n\
 145         movd       (%1), %%mm1  # Load 4 Cb   00 00 00 00 u3 u2 u1 u0     \n\
 146         movd      4(%1), %%mm3  # Load 4 Cb   00 00 00 00 u3 u2 u1 u0     \n\
 147         movd      8(%1), %%mm5  # Load 4 Cb   00 00 00 00 u3 u2 u1 u0     \n\
 148         movd     12(%1), %%mm7  # Load 4 Cb   00 00 00 00 u3 u2 u1 u0     \n\
 149         punpcklbw %%mm1, %%mm0  #             u3 v3 u2 v2 u1 v1 u0 v0     \n\
 150         punpcklbw %%mm3, %%mm2  #             u3 v3 u2 v2 u1 v1 u0 v0     \n\
 151         punpcklbw %%mm5, %%mm4  #             u3 v3 u2 v2 u1 v1 u0 v0     \n\
 152         punpcklbw %%mm7, %%mm6  #             u3 v3 u2 v2 u1 v1 u0 v0     \n\
 153         movq      %%mm0, (%2)   # Store CrCb                              \n\
 154         movq      %%mm2, 8(%2)  # Store CrCb                              \n\
 155         movq      %%mm4, 16(%2) # Store CrCb                              \n\
 156         movq      %%mm6, 24(%2) # Store CrCb"
 157         : : "r" (p_v), "r" (p_u), "r" (p_uv) );
 158
 159         p_v += 16; p_u += 16; p_uv += 32;
 160 #endif
 161     }
 162 }
 163