X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=modules%2Fvideo_filter%2Fdeinterlace.c;h=ad13c4e90d1a972e9bc02f58d9fc2b6e97303674;hb=7cfd41b20bcc4ac3fba2874630187b0d9c2faea8;hp=e6ea85e7011d9401c5f2cba8a6bb51f582feee56;hpb=a5f23e848b79bada1c8c128089bd024d85b6dd8f;p=vlc diff --git a/modules/video_filter/deinterlace.c b/modules/video_filter/deinterlace.c index e6ea85e701..ad13c4e90d 100644 --- a/modules/video_filter/deinterlace.c +++ b/modules/video_filter/deinterlace.c @@ -18,19 +18,17 @@ * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA. + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA. *****************************************************************************/ /***************************************************************************** * Preamble *****************************************************************************/ #include -#include /* malloc(), free() */ -#include #include -#include -#include +#include +#include #include "vlc_filter.h" #ifdef HAVE_ALTIVEC_H @@ -72,7 +70,10 @@ static void MergeGeneric ( void *, const void *, const void *, size_t ); static void MergeAltivec ( void *, const void *, const void *, size_t ); #endif #if defined(CAN_COMPILE_MMXEXT) -static void MergeMMX ( void *, const void *, const void *, size_t ); +static void MergeMMXEXT ( void *, const void *, const void *, size_t ); +#endif +#if defined(CAN_COMPILE_3DNOW) +static void Merge3DNow ( void *, const void *, const void *, size_t ); #endif #if defined(CAN_COMPILE_SSE) static void MergeSSE2 ( void *, const void *, const void *, size_t ); @@ -80,6 +81,9 @@ static void MergeSSE2 ( void *, const void *, const void *, size_t ); #if defined(CAN_COMPILE_MMXEXT) || defined(CAN_COMPILE_SSE) static void EndMMX ( void ); #endif +#if defined(CAN_COMPILE_3DNOW) +static void End3DNow ( void ); +#endif static int SendEvents ( vlc_object_t *, char const *, vlc_value_t, vlc_value_t, void * ); @@ -100,20 +104,20 @@ static int FilterCallback ( vlc_object_t *, char const *, * Module descriptor *****************************************************************************/ #define MODE_TEXT N_("Deinterlace mode") -#define MODE_LONGTEXT N_("Default deinterlace method to use for local playback") +#define MODE_LONGTEXT N_("Deinterlace method to use for local playback.") -#define SOUT_MODE_TEXT N_("Deinterlace mode") -#define SOUT_MODE_LONGTEXT N_("Default deinterlace methode to use for streaming") +#define SOUT_MODE_TEXT N_("Streaming deinterlace mode") +#define SOUT_MODE_LONGTEXT N_("Deinterlace method to use for streaming.") #define FILTER_CFG_PREFIX "sout-deinterlace-" -static char *mode_list[] = { "discard", "blend", "mean", "bob", "linear", "x" }; -static char *mode_list_text[] = { N_("Discard"), N_("Blend"), N_("Mean"), +static const char *mode_list[] = { "discard", "blend", "mean", "bob", "linear", "x" }; +static const char *mode_list_text[] = { N_("Discard"), N_("Blend"), N_("Mean"), N_("Bob"), N_("Linear"), "X" }; vlc_module_begin(); set_description( _("Deinterlacing video filter") ); - set_shortname( N_("Deinterlace" )); + set_shortname( _("Deinterlace" )); set_capability( "video filter", 0 ); set_category( CAT_VIDEO ); set_subcategory( SUBCAT_VIDEO_VFILTER ); @@ -201,7 +205,7 @@ static int Create( vlc_object_t *p_this ) vlc_mutex_init( p_vout, &p_vout->p_sys->filter_lock ); #if defined(CAN_COMPILE_C_ALTIVEC) - if( p_vout->p_libvlc->i_cpu & CPU_CAPABILITY_ALTIVEC ) + if( vlc_CPU() & CPU_CAPABILITY_ALTIVEC ) { p_vout->p_sys->pf_merge = MergeAltivec; p_vout->p_sys->pf_end_merge = NULL; @@ -209,7 +213,7 @@ static int Create( vlc_object_t *p_this ) else #endif #if defined(CAN_COMPILE_SSE) - if( p_vout->p_libvlc->i_cpu & CPU_CAPABILITY_SSE2 ) + if( vlc_CPU() & CPU_CAPABILITY_SSE2 ) { p_vout->p_sys->pf_merge = MergeSSE2; p_vout->p_sys->pf_end_merge = EndMMX; @@ -217,12 +221,20 @@ static int Create( vlc_object_t *p_this ) else #endif #if defined(CAN_COMPILE_MMXEXT) - if( p_vout->p_libvlc->i_cpu & CPU_CAPABILITY_MMX ) + if( vlc_CPU() & CPU_CAPABILITY_MMXEXT ) { - p_vout->p_sys->pf_merge = MergeMMX; + p_vout->p_sys->pf_merge = MergeMMXEXT; p_vout->p_sys->pf_end_merge = EndMMX; } else +#endif +#if defined(CAN_COMPILE_3DNOW) + if( vlc_CPU() & CPU_CAPABILITY_3DNOW ) + { + p_vout->p_sys->pf_merge = Merge3DNow; + p_vout->p_sys->pf_end_merge = End3DNow; + } + else #endif { p_vout->p_sys->pf_merge = MergeGeneric; @@ -355,7 +367,8 @@ static int Init( vout_thread_t *p_vout ) static vout_thread_t *SpawnRealVout( vout_thread_t *p_vout ) { vout_thread_t *p_real_vout = NULL; - video_format_t fmt = {0}; + video_format_t fmt; + memset( &fmt, 0, sizeof( video_format_t ) ); msg_Dbg( p_vout, "spawning the real video output" ); @@ -371,7 +384,7 @@ static vout_thread_t *SpawnRealVout( vout_thread_t *p_vout ) case DEINTERLACE_MEAN: case DEINTERLACE_DISCARD: fmt.i_height /= 2; fmt.i_visible_height /= 2; fmt.i_y_offset /= 2; - fmt.i_aspect /= 2; fmt.i_sar_den *= 2; + fmt.i_sar_den *= 2; p_real_vout = vout_Create( p_vout, &fmt ); break; @@ -441,8 +454,25 @@ static void Destroy( vlc_object_t *p_this ) *****************************************************************************/ static void Render ( vout_thread_t *p_vout, picture_t *p_pic ) { + vout_sys_t *p_sys = p_vout->p_sys; picture_t *pp_outpic[2]; + p_vout->fmt_out.i_x_offset = p_sys->p_vout->fmt_in.i_x_offset = + p_vout->fmt_in.i_x_offset; + p_vout->fmt_out.i_y_offset = p_sys->p_vout->fmt_in.i_y_offset = + p_vout->fmt_in.i_y_offset; + p_vout->fmt_out.i_visible_width = p_sys->p_vout->fmt_in.i_visible_width = + p_vout->fmt_in.i_visible_width; + p_vout->fmt_out.i_visible_height = p_sys->p_vout->fmt_in.i_visible_height = + p_vout->fmt_in.i_visible_height; + if( p_vout->p_sys->i_mode == DEINTERLACE_MEAN || + p_vout->p_sys->i_mode == DEINTERLACE_DISCARD ) + { + p_vout->fmt_out.i_y_offset /= 2; p_sys->p_vout->fmt_in.i_y_offset /= 2; + p_vout->fmt_out.i_visible_height /= 2; + p_sys->p_vout->fmt_in.i_visible_height /= 2; + } + pp_outpic[0] = pp_outpic[1] = NULL; vlc_mutex_lock( &p_vout->p_sys->filter_lock ); @@ -500,16 +530,16 @@ static void Render ( vout_thread_t *p_vout, picture_t *p_pic ) break; case DEINTERLACE_BOB: - RenderBob( p_vout, pp_outpic[0], p_pic, 0 ); + RenderBob( p_vout, pp_outpic[0], p_pic, p_pic->b_top_field_first ? 0 : 1 ); vout_DisplayPicture( p_vout->p_sys->p_vout, pp_outpic[0] ); - RenderBob( p_vout, pp_outpic[1], p_pic, 1 ); + RenderBob( p_vout, pp_outpic[1], p_pic, p_pic->b_top_field_first ? 1 : 0 ); vout_DisplayPicture( p_vout->p_sys->p_vout, pp_outpic[1] ); break; case DEINTERLACE_LINEAR: - RenderLinear( p_vout, pp_outpic[0], p_pic, 0 ); + RenderLinear( p_vout, pp_outpic[0], p_pic, p_pic->b_top_field_first ? 0 : 1 ); vout_DisplayPicture( p_vout->p_sys->p_vout, pp_outpic[0] ); - RenderLinear( p_vout, pp_outpic[1], p_pic, 1 ); + RenderLinear( p_vout, pp_outpic[1], p_pic, p_pic->b_top_field_first ? 1 : 0 ); vout_DisplayPicture( p_vout->p_sys->p_vout, pp_outpic[1] ); break; @@ -560,10 +590,10 @@ static void RenderDiscard( vout_thread_t *p_vout, for( ; p_out < p_out_end ; ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += 2 * p_pic->p[i_plane].i_pitch; } break; @@ -576,12 +606,12 @@ static void RenderDiscard( vout_thread_t *p_vout, { for( ; p_out < p_out_end ; ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_out += p_outpic->p[i_plane].i_pitch; + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += i_increment; } } @@ -589,9 +619,9 @@ static void RenderDiscard( vout_thread_t *p_vout, { for( ; p_out < p_out_end ; ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += i_increment; } } @@ -629,37 +659,37 @@ static void RenderBob( vout_thread_t *p_vout, /* For BOTTOM field we need to add the first line */ if( i_field == 1 ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); p_in += p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; } p_out_end -= 2 * p_outpic->p[i_plane].i_pitch; for( ; p_out < p_out_end ; ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); p_in += 2 * p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; } - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); /* For TOP field we need to add the last line */ if( i_field == 0 ) { p_in += p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_out += p_outpic->p[i_plane].i_pitch; + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); } break; @@ -668,10 +698,10 @@ static void RenderBob( vout_thread_t *p_vout, /* For BOTTOM field we need to add the first line */ if( i_field == 1 ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); p_in += p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; } p_out_end -= 2 * p_outpic->p[i_plane].i_pitch; @@ -680,39 +710,39 @@ static void RenderBob( vout_thread_t *p_vout, { for( ; p_out < p_out_end ; ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); p_in += 2 * p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; } } else { for( ; p_out < p_out_end ; ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += 2 * p_pic->p[i_plane].i_pitch; } } - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); /* For TOP field we need to add the last line */ if( i_field == 0 ) { p_in += p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_out += p_outpic->p[i_plane].i_pitch; + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); } break; @@ -744,37 +774,37 @@ static void RenderLinear( vout_thread_t *p_vout, /* For BOTTOM field we need to add the first line */ if( i_field == 1 ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); p_in += p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; } p_out_end -= 2 * p_outpic->p[i_plane].i_pitch; for( ; p_out < p_out_end ; ) { - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; Merge( p_out, p_in, p_in + 2 * p_pic->p[i_plane].i_pitch, p_pic->p[i_plane].i_pitch ); p_in += 2 * p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; } - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); /* For TOP field we need to add the last line */ if( i_field == 0 ) { p_in += p_pic->p[i_plane].i_pitch; - p_out += p_pic->p[i_plane].i_pitch; - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_out += p_outpic->p[i_plane].i_pitch; + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); } } @@ -803,7 +833,7 @@ static void RenderMean( vout_thread_t *p_vout, Merge( p_out, p_in, p_in + p_pic->p[i_plane].i_pitch, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += 2 * p_pic->p[i_plane].i_pitch; } } @@ -832,9 +862,9 @@ static void RenderBlend( vout_thread_t *p_vout, case VLC_FOURCC('I','Y','U','V'): case VLC_FOURCC('Y','V','1','2'): /* First line: simple copy */ - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; /* Remaining lines: mean value */ for( ; p_out < p_out_end ; ) @@ -842,16 +872,16 @@ static void RenderBlend( vout_thread_t *p_vout, Merge( p_out, p_in, p_in + p_pic->p[i_plane].i_pitch, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += p_pic->p[i_plane].i_pitch; } break; case VLC_FOURCC('I','4','2','2'): /* First line: simple copy */ - p_vout->p_vlc->pf_memcpy( p_out, p_in, + p_vout->p_libvlc->pf_memcpy( p_out, p_in, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; /* Remaining lines: mean value */ if( i_plane == Y_PLANE ) @@ -861,7 +891,7 @@ static void RenderBlend( vout_thread_t *p_vout, Merge( p_out, p_in, p_in + p_pic->p[i_plane].i_pitch, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += p_pic->p[i_plane].i_pitch; } } @@ -873,7 +903,7 @@ static void RenderBlend( vout_thread_t *p_vout, Merge( p_out, p_in, p_in + p_pic->p[i_plane].i_pitch, p_pic->p[i_plane].i_pitch ); - p_out += p_pic->p[i_plane].i_pitch; + p_out += p_outpic->p[i_plane].i_pitch; p_in += 2*p_pic->p[i_plane].i_pitch; } } @@ -914,8 +944,8 @@ static void MergeGeneric( void *_p_dest, const void *_p_s1, } #if defined(CAN_COMPILE_MMXEXT) -static void MergeMMX( void *_p_dest, const void *_p_s1, const void *_p_s2, - size_t i_bytes ) +static void MergeMMXEXT( void *_p_dest, const void *_p_s1, const void *_p_s2, + size_t i_bytes ) { uint8_t* p_dest = (uint8_t*)_p_dest; const uint8_t *p_s1 = (const uint8_t *)_p_s1; @@ -942,6 +972,35 @@ static void MergeMMX( void *_p_dest, const void *_p_s1, const void *_p_s2, } #endif +#if defined(CAN_COMPILE_3DNOW) +static void Merge3DNow( void *_p_dest, const void *_p_s1, const void *_p_s2, + size_t i_bytes ) +{ + uint8_t* p_dest = (uint8_t*)_p_dest; + const uint8_t *p_s1 = (const uint8_t *)_p_s1; + const uint8_t *p_s2 = (const uint8_t *)_p_s2; + uint8_t* p_end = p_dest + i_bytes - 8; + while( p_dest < p_end ) + { + __asm__ __volatile__( "movq %2,%%mm1;" + "pavgusb %1, %%mm1;" + "movq %%mm1, %0" :"=m" (*p_dest): + "m" (*p_s1), + "m" (*p_s2) ); + p_dest += 8; + p_s1 += 8; + p_s2 += 8; + } + + p_end += 8; + + while( p_dest < p_end ) + { + *p_dest++ = ( (uint16_t)(*p_s1++) + (uint16_t)(*p_s2++) ) >> 1; + } +} +#endif + #if defined(CAN_COMPILE_SSE) static void MergeSSE2( void *_p_dest, const void *_p_s1, const void *_p_s2, size_t i_bytes ) @@ -950,10 +1009,10 @@ static void MergeSSE2( void *_p_dest, const void *_p_s1, const void *_p_s2, const uint8_t *p_s1 = (const uint8_t *)_p_s1; const uint8_t *p_s2 = (const uint8_t *)_p_s2; uint8_t* p_end; - while( (ptrdiff_t)p_s1 % 16 ) + while( (uintptr_t)p_s1 % 16 ) { *p_dest++ = ( (uint16_t)(*p_s1++) + (uint16_t)(*p_s2++) ) >> 1; - } + } p_end = p_dest + i_bytes - 16; while( p_dest < p_end ) { @@ -983,6 +1042,13 @@ static void EndMMX( void ) } #endif +#if defined(CAN_COMPILE_3DNOW) +static void End3DNow( void ) +{ + __asm__ __volatile__( "femms" :: ); +} +#endif + #ifdef CAN_COMPILE_C_ALTIVEC static void MergeAltivec( void *_p_dest, const void *_p_s1, const void *_p_s2, size_t i_bytes ) @@ -1651,7 +1717,7 @@ static inline void XDeint8x8FieldMotion( uint8_t *dst, int i_dst, #if 0 /* Kernel interpolation (1,-5,20,20,-5,1) - * Loose a bit more details+add aliasing than edge interpol but avoid + * Lose a bit more details+add aliasing than edge interpol but avoid * more artifacts */ static inline uint8_t clip1( int a ) @@ -1897,7 +1963,7 @@ static void RenderX( vout_thread_t *p_vout, uint8_t *src = &p_pic->p[i_plane].p_pixels[8*y*i_src]; #ifdef CAN_COMPILE_MMXEXT - if( p_vout->p_libvlc->i_cpu & CPU_CAPABILITY_MMXEXT ) + if( vlc_CPU() & CPU_CAPABILITY_MMXEXT ) XDeintBand8x8MMXEXT( dst, i_dst, src, i_src, i_mbx, i_modx ); else #endif @@ -1924,7 +1990,7 @@ static void RenderX( vout_thread_t *p_vout, } #ifdef CAN_COMPILE_MMXEXT - if( p_vout->p_libvlc->i_cpu & CPU_CAPABILITY_MMXEXT ) + if( vlc_CPU() & CPU_CAPABILITY_MMXEXT ) emms(); #endif } @@ -2138,7 +2204,7 @@ static int OpenFilter( vlc_object_t *p_this ) p_filter->p_sys = (filter_sys_t *)p_vout; p_vout->render.i_chroma = p_filter->fmt_in.video.i_chroma; - sout_CfgParse( p_filter, FILTER_CFG_PREFIX, ppsz_filter_options, + config_ChainParse( p_filter, FILTER_CFG_PREFIX, ppsz_filter_options, p_filter->p_cfg ); var_Get( p_filter, FILTER_CFG_PREFIX "mode", &val ); var_Create( p_filter, "deinterlace-mode", VLC_VAR_STRING );