X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fvc1.c;h=f3d3cd4ed1bc43e787b1573143a4edae5347a1df;hb=feb993e5794b3dacb51085a9b77013495eb6dd1c;hp=5993f1573c108e0b004ae3bcaa52e7d4108c535f;hpb=9d1f80f2070349429e2f4d45a03fbd34a5c6bcf4;p=ffmpeg diff --git a/libavcodec/vc1.c b/libavcodec/vc1.c index 5993f1573c1..f3d3cd4ed1b 100644 --- a/libavcodec/vc1.c +++ b/libavcodec/vc1.c @@ -1,6 +1,6 @@ /* * VC-1 and WMV3 decoder - * Copyright (c) 2006 Konstantin Shishkov + * Copyright (c) 2006-2007 Konstantin Shishkov * Partly based on vc9.c (c) 2005 Anonymous, Alex Beregszaszi, Michael Niedermayer * * This file is part of FFmpeg. @@ -18,429 +18,34 @@ * You should have received a copy of the GNU Lesser General Public * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - * */ /** - * @file vc1.c + * @file libavcodec/vc1.c * VC-1 and WMV3 decoder * */ -#include "common.h" +#include "internal.h" #include "dsputil.h" #include "avcodec.h" #include "mpegvideo.h" +#include "vc1.h" #include "vc1data.h" #include "vc1acdata.h" +#include "msmpeg4data.h" +#include "unary.h" +#include "simple_idct.h" +#include "mathops.h" +#include "vdpau_internal.h" #undef NDEBUG #include -extern const uint32_t ff_table0_dc_lum[120][2], ff_table1_dc_lum[120][2]; -extern const uint32_t ff_table0_dc_chroma[120][2], ff_table1_dc_chroma[120][2]; -extern VLC ff_msmp4_dc_luma_vlc[2], ff_msmp4_dc_chroma_vlc[2]; #define MB_INTRA_VLC_BITS 9 -extern VLC ff_msmp4_mb_i_vlc; -extern const uint16_t ff_msmp4_mb_i_table[64][2]; #define DC_VLC_BITS 9 #define AC_VLC_BITS 9 static const uint16_t table_mb_intra[64][2]; -/** Markers used if VC-1 AP frame data */ -//@{ -enum VC1Code{ - VC1_CODE_RES0 = 0x00000100, - VC1_CODE_ENDOFSEQ = 0x0000010A, - VC1_CODE_SLICE, - VC1_CODE_FIELD, - VC1_CODE_FRAME, - VC1_CODE_ENTRYPOINT, - VC1_CODE_SEQHDR, -}; -//@} - -/** Available Profiles */ -//@{ -enum Profile { - PROFILE_SIMPLE, - PROFILE_MAIN, - PROFILE_COMPLEX, ///< TODO: WMV9 specific - PROFILE_ADVANCED -}; -//@} - -/** Sequence quantizer mode */ -//@{ -enum QuantMode { - QUANT_FRAME_IMPLICIT, ///< Implicitly specified at frame level - QUANT_FRAME_EXPLICIT, ///< Explicitly specified at frame level - QUANT_NON_UNIFORM, ///< Non-uniform quant used for all frames - QUANT_UNIFORM ///< Uniform quant used for all frames -}; -//@} - -/** Where quant can be changed */ -//@{ -enum DQProfile { - DQPROFILE_FOUR_EDGES, - DQPROFILE_DOUBLE_EDGES, - DQPROFILE_SINGLE_EDGE, - DQPROFILE_ALL_MBS -}; -//@} - -/** @name Where quant can be changed - */ -//@{ -enum DQSingleEdge { - DQSINGLE_BEDGE_LEFT, - DQSINGLE_BEDGE_TOP, - DQSINGLE_BEDGE_RIGHT, - DQSINGLE_BEDGE_BOTTOM -}; -//@} - -/** Which pair of edges is quantized with ALTPQUANT */ -//@{ -enum DQDoubleEdge { - DQDOUBLE_BEDGE_TOPLEFT, - DQDOUBLE_BEDGE_TOPRIGHT, - DQDOUBLE_BEDGE_BOTTOMRIGHT, - DQDOUBLE_BEDGE_BOTTOMLEFT -}; -//@} - -/** MV modes for P frames */ -//@{ -enum MVModes { - MV_PMODE_1MV_HPEL_BILIN, - MV_PMODE_1MV, - MV_PMODE_1MV_HPEL, - MV_PMODE_MIXED_MV, - MV_PMODE_INTENSITY_COMP -}; -//@} - -/** @name MV types for B frames */ -//@{ -enum BMVTypes { - BMV_TYPE_BACKWARD, - BMV_TYPE_FORWARD, - BMV_TYPE_INTERPOLATED -}; -//@} - -/** @name Block types for P/B frames */ -//@{ -enum TransformTypes { - TT_8X8, - TT_8X4_BOTTOM, - TT_8X4_TOP, - TT_8X4, //Both halves - TT_4X8_RIGHT, - TT_4X8_LEFT, - TT_4X8, //Both halves - TT_4X4 -}; -//@} - -/** Table for conversion between TTBLK and TTMB */ -static const int ttblk_to_tt[3][8] = { - { TT_8X4, TT_4X8, TT_8X8, TT_4X4, TT_8X4_TOP, TT_8X4_BOTTOM, TT_4X8_RIGHT, TT_4X8_LEFT }, - { TT_8X8, TT_4X8_RIGHT, TT_4X8_LEFT, TT_4X4, TT_8X4, TT_4X8, TT_8X4_BOTTOM, TT_8X4_TOP }, - { TT_8X8, TT_4X8, TT_4X4, TT_8X4_BOTTOM, TT_4X8_RIGHT, TT_4X8_LEFT, TT_8X4, TT_8X4_TOP } -}; - -static const int ttfrm_to_tt[4] = { TT_8X8, TT_8X4, TT_4X8, TT_4X4 }; - -/** MV P mode - the 5th element is only used for mode 1 */ -static const uint8_t mv_pmode_table[2][5] = { - { MV_PMODE_1MV_HPEL_BILIN, MV_PMODE_1MV, MV_PMODE_1MV_HPEL, MV_PMODE_INTENSITY_COMP, MV_PMODE_MIXED_MV }, - { MV_PMODE_1MV, MV_PMODE_MIXED_MV, MV_PMODE_1MV_HPEL, MV_PMODE_INTENSITY_COMP, MV_PMODE_1MV_HPEL_BILIN } -}; -static const uint8_t mv_pmode_table2[2][4] = { - { MV_PMODE_1MV_HPEL_BILIN, MV_PMODE_1MV, MV_PMODE_1MV_HPEL, MV_PMODE_MIXED_MV }, - { MV_PMODE_1MV, MV_PMODE_MIXED_MV, MV_PMODE_1MV_HPEL, MV_PMODE_1MV_HPEL_BILIN } -}; - -/** One more frame type */ -#define BI_TYPE 7 - -static const int fps_nr[5] = { 24, 25, 30, 50, 60 }, - fps_dr[2] = { 1000, 1001 }; -static const uint8_t pquant_table[3][32] = { - { /* Implicit quantizer */ - 0, 1, 2, 3, 4, 5, 6, 7, 8, 6, 7, 8, 9, 10, 11, 12, - 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 27, 29, 31 - }, - { /* Explicit quantizer, pquantizer uniform */ - 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, - 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 - }, - { /* Explicit quantizer, pquantizer non-uniform */ - 0, 1, 1, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, - 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 29, 31 - } -}; - -/** @name VC-1 VLC tables and defines - * @todo TODO move this into the context - */ -//@{ -#define VC1_BFRACTION_VLC_BITS 7 -static VLC vc1_bfraction_vlc; -#define VC1_IMODE_VLC_BITS 4 -static VLC vc1_imode_vlc; -#define VC1_NORM2_VLC_BITS 3 -static VLC vc1_norm2_vlc; -#define VC1_NORM6_VLC_BITS 9 -static VLC vc1_norm6_vlc; -/* Could be optimized, one table only needs 8 bits */ -#define VC1_TTMB_VLC_BITS 9 //12 -static VLC vc1_ttmb_vlc[3]; -#define VC1_MV_DIFF_VLC_BITS 9 //15 -static VLC vc1_mv_diff_vlc[4]; -#define VC1_CBPCY_P_VLC_BITS 9 //14 -static VLC vc1_cbpcy_p_vlc[4]; -#define VC1_4MV_BLOCK_PATTERN_VLC_BITS 6 -static VLC vc1_4mv_block_pattern_vlc[4]; -#define VC1_TTBLK_VLC_BITS 5 -static VLC vc1_ttblk_vlc[3]; -#define VC1_SUBBLKPAT_VLC_BITS 6 -static VLC vc1_subblkpat_vlc[3]; - -static VLC vc1_ac_coeff_table[8]; -//@} - -enum CodingSet { - CS_HIGH_MOT_INTRA = 0, - CS_HIGH_MOT_INTER, - CS_LOW_MOT_INTRA, - CS_LOW_MOT_INTER, - CS_MID_RATE_INTRA, - CS_MID_RATE_INTER, - CS_HIGH_RATE_INTRA, - CS_HIGH_RATE_INTER -}; - -/** @name Overlap conditions for Advanced Profile */ -//@{ -enum COTypes { - CONDOVER_NONE = 0, - CONDOVER_ALL, - CONDOVER_SELECT -}; -//@} - - -/** The VC1 Context - * @fixme Change size wherever another size is more efficient - * Many members are only used for Advanced Profile - */ -typedef struct VC1Context{ - MpegEncContext s; - - int bits; - - /** Simple/Main Profile sequence header */ - //@{ - int res_sm; ///< reserved, 2b - int res_x8; ///< reserved - int multires; ///< frame-level RESPIC syntax element present - int res_fasttx; ///< reserved, always 1 - int res_transtab; ///< reserved, always 0 - int rangered; ///< RANGEREDFRM (range reduction) syntax element present - ///< at frame level - int res_rtm_flag; ///< reserved, set to 1 - int reserved; ///< reserved - //@} - - /** Advanced Profile */ - //@{ - int level; ///< 3bits, for Advanced/Simple Profile, provided by TS layer - int chromaformat; ///< 2bits, 2=4:2:0, only defined - int postprocflag; ///< Per-frame processing suggestion flag present - int broadcast; ///< TFF/RFF present - int interlace; ///< Progressive/interlaced (RPTFTM syntax element) - int tfcntrflag; ///< TFCNTR present - int panscanflag; ///< NUMPANSCANWIN, TOPLEFT{X,Y}, BOTRIGHT{X,Y} present - int extended_dmv; ///< Additional extended dmv range at P/B frame-level - int color_prim; ///< 8bits, chroma coordinates of the color primaries - int transfer_char; ///< 8bits, Opto-electronic transfer characteristics - int matrix_coef; ///< 8bits, Color primaries->YCbCr transform matrix - int hrd_param_flag; ///< Presence of Hypothetical Reference - ///< Decoder parameters - int psf; ///< Progressive Segmented Frame - //@} - - /** Sequence header data for all Profiles - * TODO: choose between ints, uint8_ts and monobit flags - */ - //@{ - int profile; ///< 2bits, Profile - int frmrtq_postproc; ///< 3bits, - int bitrtq_postproc; ///< 5bits, quantized framerate-based postprocessing strength - int fastuvmc; ///< Rounding of qpel vector to hpel ? (not in Simple) - int extended_mv; ///< Ext MV in P/B (not in Simple) - int dquant; ///< How qscale varies with MBs, 2bits (not in Simple) - int vstransform; ///< variable-size [48]x[48] transform type + info - int overlap; ///< overlapped transforms in use - int quantizer_mode; ///< 2bits, quantizer mode used for sequence, see QUANT_* - int finterpflag; ///< INTERPFRM present - //@} - - /** Frame decoding info for all profiles */ - //@{ - uint8_t mv_mode; ///< MV coding monde - uint8_t mv_mode2; ///< Secondary MV coding mode (B frames) - int k_x; ///< Number of bits for MVs (depends on MV range) - int k_y; ///< Number of bits for MVs (depends on MV range) - int range_x, range_y; ///< MV range - uint8_t pq, altpq; ///< Current/alternate frame quantizer scale - /** pquant parameters */ - //@{ - uint8_t dquantfrm; - uint8_t dqprofile; - uint8_t dqsbedge; - uint8_t dqbilevel; - //@} - /** AC coding set indexes - * @see 8.1.1.10, p(1)10 - */ - //@{ - int c_ac_table_index; ///< Chroma index from ACFRM element - int y_ac_table_index; ///< Luma index from AC2FRM element - //@} - int ttfrm; ///< Transform type info present at frame level - uint8_t ttmbf; ///< Transform type flag - uint8_t ttblk4x4; ///< Value of ttblk which indicates a 4x4 transform - int codingset; ///< index of current table set from 11.8 to use for luma block decoding - int codingset2; ///< index of current table set from 11.8 to use for chroma block decoding - int pqindex; ///< raw pqindex used in coding set selection - int a_avail, c_avail; - uint8_t *mb_type_base, *mb_type[3]; - - - /** Luma compensation parameters */ - //@{ - uint8_t lumscale; - uint8_t lumshift; - //@} - int16_t bfraction; ///< Relative position % anchors=> how to scale MVs - uint8_t halfpq; ///< Uniform quant over image and qp+.5 - uint8_t respic; ///< Frame-level flag for resized images - int buffer_fullness; ///< HRD info - /** Ranges: - * -# 0 -> [-64n 63.f] x [-32, 31.f] - * -# 1 -> [-128, 127.f] x [-64, 63.f] - * -# 2 -> [-512, 511.f] x [-128, 127.f] - * -# 3 -> [-1024, 1023.f] x [-256, 255.f] - */ - uint8_t mvrange; - uint8_t pquantizer; ///< Uniform (over sequence) quantizer in use - VLC *cbpcy_vlc; ///< CBPCY VLC table - int tt_index; ///< Index for Transform Type tables - uint8_t* mv_type_mb_plane; ///< bitplane for mv_type == (4MV) - uint8_t* direct_mb_plane; ///< bitplane for "direct" MBs - int mv_type_is_raw; ///< mv type mb plane is not coded - int dmb_is_raw; ///< direct mb plane is raw - int skip_is_raw; ///< skip mb plane is not coded - uint8_t luty[256], lutuv[256]; // lookup tables used for intensity compensation - int use_ic; ///< use intensity compensation in B-frames - int rnd; ///< rounding control - - /** Frame decoding info for S/M profiles only */ - //@{ - uint8_t rangeredfrm; ///< out_sample = CLIP((in_sample-128)*2+128) - uint8_t interpfrm; - //@} - - /** Frame decoding info for Advanced profile */ - //@{ - uint8_t fcm; ///< 0->Progressive, 2->Frame-Interlace, 3->Field-Interlace - uint8_t numpanscanwin; - uint8_t tfcntr; - uint8_t rptfrm, tff, rff; - uint16_t topleftx; - uint16_t toplefty; - uint16_t bottomrightx; - uint16_t bottomrighty; - uint8_t uvsamp; - uint8_t postproc; - int hrd_num_leaky_buckets; - uint8_t bit_rate_exponent; - uint8_t buffer_size_exponent; - uint8_t* acpred_plane; ///< AC prediction flags bitplane - int acpred_is_raw; - uint8_t* over_flags_plane; ///< Overflags bitplane - int overflg_is_raw; - uint8_t condover; - uint16_t *hrd_rate, *hrd_buffer; - uint8_t *hrd_fullness; - uint8_t range_mapy_flag; - uint8_t range_mapuv_flag; - uint8_t range_mapy; - uint8_t range_mapuv; - //@} - - int p_frame_skipped; - int bi_type; -} VC1Context; - -/** - * Get unary code of limited length - * @fixme FIXME Slow and ugly - * @param gb GetBitContext - * @param[in] stop The bitstop value (unary code of 1's or 0's) - * @param[in] len Maximum length - * @return Unary length/index - */ -static int get_prefix(GetBitContext *gb, int stop, int len) -{ -#if 1 - int i; - - for(i = 0; i < len && get_bits1(gb) != stop; i++); - return i; -/* int i = 0, tmp = !stop; - - while (i != len && tmp != stop) - { - tmp = get_bits(gb, 1); - i++; - } - if (i == len && tmp != stop) return len+1; - return i;*/ -#else - unsigned int buf; - int log; - - OPEN_READER(re, gb); - UPDATE_CACHE(re, gb); - buf=GET_CACHE(re, gb); //Still not sure - if (stop) buf = ~buf; - - log= av_log2(-buf); //FIXME: -? - if (log < limit){ - LAST_SKIP_BITS(re, gb, log+1); - CLOSE_READER(re, gb); - return log; - } - - LAST_SKIP_BITS(re, gb, limit); - CLOSE_READER(re, gb); - return limit; -#endif -} - -static inline int decode210(GetBitContext *gb){ - int n; - n = get_bits1(gb); - if (n == 1) - return 0; - else - return 2 - get_bits1(gb); -} /** * Init VC-1 specific tables and VC1Context members @@ -458,44 +63,44 @@ static int vc1_init_common(VC1Context *v) if(!done) { done = 1; - init_vlc(&vc1_bfraction_vlc, VC1_BFRACTION_VLC_BITS, 23, - vc1_bfraction_bits, 1, 1, - vc1_bfraction_codes, 1, 1, 1); - init_vlc(&vc1_norm2_vlc, VC1_NORM2_VLC_BITS, 4, - vc1_norm2_bits, 1, 1, - vc1_norm2_codes, 1, 1, 1); - init_vlc(&vc1_norm6_vlc, VC1_NORM6_VLC_BITS, 64, - vc1_norm6_bits, 1, 1, - vc1_norm6_codes, 2, 2, 1); - init_vlc(&vc1_imode_vlc, VC1_IMODE_VLC_BITS, 7, - vc1_imode_bits, 1, 1, - vc1_imode_codes, 1, 1, 1); + init_vlc(&ff_vc1_bfraction_vlc, VC1_BFRACTION_VLC_BITS, 23, + ff_vc1_bfraction_bits, 1, 1, + ff_vc1_bfraction_codes, 1, 1, 1); + init_vlc(&ff_vc1_norm2_vlc, VC1_NORM2_VLC_BITS, 4, + ff_vc1_norm2_bits, 1, 1, + ff_vc1_norm2_codes, 1, 1, 1); + init_vlc(&ff_vc1_norm6_vlc, VC1_NORM6_VLC_BITS, 64, + ff_vc1_norm6_bits, 1, 1, + ff_vc1_norm6_codes, 2, 2, 1); + init_vlc(&ff_vc1_imode_vlc, VC1_IMODE_VLC_BITS, 7, + ff_vc1_imode_bits, 1, 1, + ff_vc1_imode_codes, 1, 1, 1); for (i=0; i<3; i++) { - init_vlc(&vc1_ttmb_vlc[i], VC1_TTMB_VLC_BITS, 16, - vc1_ttmb_bits[i], 1, 1, - vc1_ttmb_codes[i], 2, 2, 1); - init_vlc(&vc1_ttblk_vlc[i], VC1_TTBLK_VLC_BITS, 8, - vc1_ttblk_bits[i], 1, 1, - vc1_ttblk_codes[i], 1, 1, 1); - init_vlc(&vc1_subblkpat_vlc[i], VC1_SUBBLKPAT_VLC_BITS, 15, - vc1_subblkpat_bits[i], 1, 1, - vc1_subblkpat_codes[i], 1, 1, 1); + init_vlc(&ff_vc1_ttmb_vlc[i], VC1_TTMB_VLC_BITS, 16, + ff_vc1_ttmb_bits[i], 1, 1, + ff_vc1_ttmb_codes[i], 2, 2, 1); + init_vlc(&ff_vc1_ttblk_vlc[i], VC1_TTBLK_VLC_BITS, 8, + ff_vc1_ttblk_bits[i], 1, 1, + ff_vc1_ttblk_codes[i], 1, 1, 1); + init_vlc(&ff_vc1_subblkpat_vlc[i], VC1_SUBBLKPAT_VLC_BITS, 15, + ff_vc1_subblkpat_bits[i], 1, 1, + ff_vc1_subblkpat_codes[i], 1, 1, 1); } for(i=0; i<4; i++) { - init_vlc(&vc1_4mv_block_pattern_vlc[i], VC1_4MV_BLOCK_PATTERN_VLC_BITS, 16, - vc1_4mv_block_pattern_bits[i], 1, 1, - vc1_4mv_block_pattern_codes[i], 1, 1, 1); - init_vlc(&vc1_cbpcy_p_vlc[i], VC1_CBPCY_P_VLC_BITS, 64, - vc1_cbpcy_p_bits[i], 1, 1, - vc1_cbpcy_p_codes[i], 2, 2, 1); - init_vlc(&vc1_mv_diff_vlc[i], VC1_MV_DIFF_VLC_BITS, 73, - vc1_mv_diff_bits[i], 1, 1, - vc1_mv_diff_codes[i], 2, 2, 1); + init_vlc(&ff_vc1_4mv_block_pattern_vlc[i], VC1_4MV_BLOCK_PATTERN_VLC_BITS, 16, + ff_vc1_4mv_block_pattern_bits[i], 1, 1, + ff_vc1_4mv_block_pattern_codes[i], 1, 1, 1); + init_vlc(&ff_vc1_cbpcy_p_vlc[i], VC1_CBPCY_P_VLC_BITS, 64, + ff_vc1_cbpcy_p_bits[i], 1, 1, + ff_vc1_cbpcy_p_codes[i], 2, 2, 1); + init_vlc(&ff_vc1_mv_diff_vlc[i], VC1_MV_DIFF_VLC_BITS, 73, + ff_vc1_mv_diff_bits[i], 1, 1, + ff_vc1_mv_diff_codes[i], 2, 2, 1); } for(i=0; i<8; i++) - init_vlc(&vc1_ac_coeff_table[i], AC_VLC_BITS, vc1_ac_sizes[i], + init_vlc(&ff_vc1_ac_coeff_table[i], AC_VLC_BITS, vc1_ac_sizes[i], &vc1_ac_tables[i][0][1], 8, 4, &vc1_ac_tables[i][0][0], 8, 4, 1); init_vlc(&ff_msmp4_mb_i_vlc, MB_INTRA_VLC_BITS, 64, @@ -512,12 +117,12 @@ static int vc1_init_common(VC1Context *v) /***********************************************************************/ /** - * @defgroup bitplane VC9 Bitplane decoding + * @defgroup vc1bitplane VC-1 Bitplane decoding * @see 8.7, p56 * @{ */ -/** @addtogroup bitplane +/** * Imode types * @{ */ @@ -542,11 +147,11 @@ static void decode_rowskip(uint8_t* plane, int width, int height, int stride, Ge int x, y; for (y=0; ys.mb_width; height = v->s.mb_height; stride = v->s.mb_stride; - invert = get_bits(gb, 1); - imode = get_vlc2(gb, vc1_imode_vlc.table, VC1_IMODE_VLC_BITS, 1); + invert = get_bits1(gb); + imode = get_vlc2(gb, ff_vc1_imode_vlc.table, VC1_IMODE_VLC_BITS, 1); *raw_flag = 0; switch (imode) @@ -603,13 +209,13 @@ static int bitplane_decoding(uint8_t* data, int *raw_flag, VC1Context *v) case IMODE_NORM2: if ((height * width) & 1) { - *planep++ = get_bits(gb, 1); + *planep++ = get_bits1(gb); offset = 1; } else offset = 0; // decode bitplane as one long line for (y = offset; y < height * width; y += 2) { - code = get_vlc2(gb, vc1_norm2_vlc.table, VC1_NORM2_VLC_BITS, 1); + code = get_vlc2(gb, ff_vc1_norm2_vlc.table, VC1_NORM2_VLC_BITS, 1); *planep++ = code & 1; offset++; if(offset == width) { @@ -629,7 +235,7 @@ static int bitplane_decoding(uint8_t* data, int *raw_flag, VC1Context *v) if(!(height % 3) && (width % 3)) { // use 2x3 decoding for(y = 0; y < height; y+= 3) { for(x = width & 1; x < width; x += 2) { - code = get_vlc2(gb, vc1_norm6_vlc.table, VC1_NORM6_VLC_BITS, 2); + code = get_vlc2(gb, ff_vc1_norm6_vlc.table, VC1_NORM6_VLC_BITS, 2); if(code < 0){ av_log(v->s.avctx, AV_LOG_DEBUG, "invalid NORM-6 VLC\n"); return -1; @@ -648,7 +254,7 @@ static int bitplane_decoding(uint8_t* data, int *raw_flag, VC1Context *v) planep += (height & 1) * stride; for(y = height & 1; y < height; y += 2) { for(x = width % 3; x < width; x += 3) { - code = get_vlc2(gb, vc1_norm6_vlc.table, VC1_NORM6_VLC_BITS, 2); + code = get_vlc2(gb, ff_vc1_norm6_vlc.table, VC1_NORM6_VLC_BITS, 2); if(code < 0){ av_log(v->s.avctx, AV_LOG_DEBUG, "invalid NORM-6 VLC\n"); return -1; @@ -704,6 +310,90 @@ static int bitplane_decoding(uint8_t* data, int *raw_flag, VC1Context *v) /** @} */ //Bitplane group +/** + * VC-1 in-loop deblocking filter for one line + * @param src source block type + * @param stride block stride + * @param pq block quantizer + * @return whether other 3 pairs should be filtered or not + * @see 8.6 + */ +static av_always_inline int vc1_filter_line(uint8_t* src, int stride, int pq){ + uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; + + int a0 = (2*(src[-2*stride] - src[ 1*stride]) - 5*(src[-1*stride] - src[ 0*stride]) + 4) >> 3; + int a0_sign = a0 >> 31; /* Store sign */ + a0 = (a0 ^ a0_sign) - a0_sign; /* a0 = FFABS(a0); */ + if(a0 < pq){ + int a1 = FFABS((2*(src[-4*stride] - src[-1*stride]) - 5*(src[-3*stride] - src[-2*stride]) + 4) >> 3); + int a2 = FFABS((2*(src[ 0*stride] - src[ 3*stride]) - 5*(src[ 1*stride] - src[ 2*stride]) + 4) >> 3); + if(a1 < a0 || a2 < a0){ + int clip = src[-1*stride] - src[ 0*stride]; + int clip_sign = clip >> 31; + clip = ((clip ^ clip_sign) - clip_sign)>>1; + if(clip){ + int a3 = FFMIN(a1, a2); + int d = 5 * (a3 - a0); + int d_sign = (d >> 31); + d = ((d ^ d_sign) - d_sign) >> 3; + d_sign ^= a0_sign; + + if( d_sign ^ clip_sign ) + d = 0; + else{ + d = FFMIN(d, clip); + d = (d ^ d_sign) - d_sign; /* Restore sign */ + src[-1*stride] = cm[src[-1*stride] - d]; + src[ 0*stride] = cm[src[ 0*stride] + d]; + } + return 1; + } + } + } + return 0; +} + +/** + * VC-1 in-loop deblocking filter + * @param src source block type + * @param step distance between horizontally adjacent elements + * @param stride distance between vertically adjacent elements + * @param len edge length to filter (4 or 8 pixels) + * @param pq block quantizer + * @see 8.6 + */ +static void vc1_loop_filter(uint8_t* src, int step, int stride, int len, int pq) +{ + int i; + int filt3; + + for(i = 0; i < len; i += 4){ + filt3 = vc1_filter_line(src + 2*step, stride, pq); + if(filt3){ + vc1_filter_line(src + 0*step, stride, pq); + vc1_filter_line(src + 1*step, stride, pq); + vc1_filter_line(src + 3*step, stride, pq); + } + src += step * 4; + } +} + +static void vc1_loop_filter_iblk(MpegEncContext *s, int pq) +{ + int i, j; + if(!s->first_slice_line) + vc1_loop_filter(s->dest[0], 1, s->linesize, 16, pq); + vc1_loop_filter(s->dest[0] + 8*s->linesize, 1, s->linesize, 16, pq); + for(i = !s->mb_x*8; i < 16; i += 8) + vc1_loop_filter(s->dest[0] + i, s->linesize, 1, 16, pq); + for(j = 0; j < 2; j++){ + if(!s->first_slice_line) + vc1_loop_filter(s->dest[j+1], 1, s->uvlinesize, 8, pq); + if(s->mb_x) + vc1_loop_filter(s->dest[j+1], s->uvlinesize, 1, 8, pq); + } +} + /***********************************************************************/ /** VOP Dquant decoding * @param v VC-1 Context @@ -722,7 +412,7 @@ static int vop_dquant_decoding(VC1Context *v) } else { - v->dquantfrm = get_bits(gb, 1); + v->dquantfrm = get_bits1(gb); if ( v->dquantfrm ) { v->dqprofile = get_bits(gb, 2); @@ -733,7 +423,9 @@ static int vop_dquant_decoding(VC1Context *v) v->dqsbedge = get_bits(gb, 2); break; case DQPROFILE_ALL_MBS: - v->dqbilevel = get_bits(gb, 1); + v->dqbilevel = get_bits1(gb); + if(!v->dqbilevel) + v->halfpq = 0; default: break; //Forbidden ? } if (v->dqbilevel || v->dqprofile != DQPROFILE_ALL_MBS) @@ -796,7 +488,7 @@ static void vc1_mc_1mv(VC1Context *v, int dir) my = s->mv[dir][0][1]; // store motion vectors for further use in B frames - if(s->pict_type == P_TYPE) { + if(s->pict_type == FF_P_TYPE) { s->current_picture.motion_val[1][s->block_index[0]][0] = mx; s->current_picture.motion_val[1][s->block_index[0]][1] = my; } @@ -821,10 +513,17 @@ static void vc1_mc_1mv(VC1Context *v, int dir) uvsrc_x = s->mb_x * 8 + (uvmx >> 2); uvsrc_y = s->mb_y * 8 + (uvmy >> 2); - src_x = av_clip( src_x, -16, s->mb_width * 16); - src_y = av_clip( src_y, -16, s->mb_height * 16); - uvsrc_x = av_clip(uvsrc_x, -8, s->mb_width * 8); - uvsrc_y = av_clip(uvsrc_y, -8, s->mb_height * 8); + if(v->profile != PROFILE_ADVANCED){ + src_x = av_clip( src_x, -16, s->mb_width * 16); + src_y = av_clip( src_y, -16, s->mb_height * 16); + uvsrc_x = av_clip(uvsrc_x, -8, s->mb_width * 8); + uvsrc_y = av_clip(uvsrc_y, -8, s->mb_height * 8); + }else{ + src_x = av_clip( src_x, -17, s->avctx->coded_width); + src_y = av_clip( src_y, -18, s->avctx->coded_height + 1); + uvsrc_x = av_clip(uvsrc_x, -8, s->avctx->coded_width >> 1); + uvsrc_y = av_clip(uvsrc_y, -8, s->avctx->coded_height >> 1); + } srcY += src_y * s->linesize + src_x; srcU += uvsrc_y * s->uvlinesize + uvsrc_x; @@ -944,8 +643,13 @@ static void vc1_mc_4mv_luma(VC1Context *v, int n) src_x = s->mb_x * 16 + (n&1) * 8 + (mx >> 2); src_y = s->mb_y * 16 + (n&2) * 4 + (my >> 2); - src_x = av_clip( src_x, -16, s->mb_width * 16); - src_y = av_clip( src_y, -16, s->mb_height * 16); + if(v->profile != PROFILE_ADVANCED){ + src_x = av_clip( src_x, -16, s->mb_width * 16); + src_y = av_clip( src_y, -16, s->mb_height * 16); + }else{ + src_x = av_clip( src_x, -17, s->avctx->coded_width); + src_y = av_clip( src_y, -18, s->avctx->coded_height + 1); + } srcY += src_y * s->linesize + src_x; @@ -1056,8 +760,11 @@ static void vc1_mc_4mv_chroma(VC1Context *v) for(i= t1+1; i<4; i++)if(!intra[i]) {t2 = i; break;} tx = (mvx[t1] + mvx[t2]) / 2; ty = (mvy[t1] + mvy[t2]) / 2; - } else + } else { + s->current_picture.motion_val[1][s->block_index[0]][0] = 0; + s->current_picture.motion_val[1][s->block_index[0]][1] = 0; return; //no need to do MC for inter blocks + } s->current_picture.motion_val[1][s->block_index[0]][0] = tx; s->current_picture.motion_val[1][s->block_index[0]][1] = ty; @@ -1071,8 +778,14 @@ static void vc1_mc_4mv_chroma(VC1Context *v) uvsrc_x = s->mb_x * 8 + (uvmx >> 2); uvsrc_y = s->mb_y * 8 + (uvmy >> 2); - uvsrc_x = av_clip(uvsrc_x, -8, s->mb_width * 8); - uvsrc_y = av_clip(uvsrc_y, -8, s->mb_height * 8); + if(v->profile != PROFILE_ADVANCED){ + uvsrc_x = av_clip(uvsrc_x, -8, s->mb_width * 8); + uvsrc_y = av_clip(uvsrc_y, -8, s->mb_height * 8); + }else{ + uvsrc_x = av_clip(uvsrc_x, -8, s->avctx->coded_width >> 1); + uvsrc_y = av_clip(uvsrc_y, -8, s->avctx->coded_height >> 1); + } + srcU = s->last_picture.data[1] + uvsrc_y * s->uvlinesize + uvsrc_x; srcV = s->last_picture.data[2] + uvsrc_y * s->uvlinesize + uvsrc_x; if(v->rangeredfrm || (v->mv_mode == MV_PMODE_INTENSITY_COMP) @@ -1152,10 +865,14 @@ static int decode_sequence_header(AVCodecContext *avctx, GetBitContext *gb) if (v->profile == PROFILE_ADVANCED) { + v->zz_8x4 = ff_vc1_adv_progressive_8x4_zz; + v->zz_4x8 = ff_vc1_adv_progressive_4x8_zz; return decode_sequence_header_adv(v, gb); } else { + v->zz_8x4 = wmv2_scantableA; + v->zz_4x8 = wmv2_scantableB; v->res_sm = get_bits(gb, 2); //reserved if (v->res_sm) { @@ -1169,37 +886,34 @@ static int decode_sequence_header(AVCodecContext *avctx, GetBitContext *gb) v->frmrtq_postproc = get_bits(gb, 3); //common // (bitrate-32kbps)/64kbps v->bitrtq_postproc = get_bits(gb, 5); //common - v->s.loop_filter = get_bits(gb, 1); //common + v->s.loop_filter = get_bits1(gb); //common if(v->s.loop_filter == 1 && v->profile == PROFILE_SIMPLE) { av_log(avctx, AV_LOG_ERROR, "LOOPFILTER shell not be enabled in simple profile\n"); } + if(v->s.avctx->skip_loop_filter >= AVDISCARD_ALL) + v->s.loop_filter = 0; - v->res_x8 = get_bits(gb, 1); //reserved - if (v->res_x8) - { - av_log(avctx, AV_LOG_ERROR, - "1 for reserved RES_X8 is forbidden\n"); - //return -1; - } - v->multires = get_bits(gb, 1); - v->res_fasttx = get_bits(gb, 1); + v->res_x8 = get_bits1(gb); //reserved + v->multires = get_bits1(gb); + v->res_fasttx = get_bits1(gb); if (!v->res_fasttx) { - av_log(avctx, AV_LOG_ERROR, - "0 for reserved RES_FASTTX is forbidden\n"); - //return -1; + v->s.dsp.vc1_inv_trans_8x8 = ff_simple_idct; + v->s.dsp.vc1_inv_trans_8x4 = ff_simple_idct84_add; + v->s.dsp.vc1_inv_trans_4x8 = ff_simple_idct48_add; + v->s.dsp.vc1_inv_trans_4x4 = ff_simple_idct44_add; } - v->fastuvmc = get_bits(gb, 1); //common + v->fastuvmc = get_bits1(gb); //common if (!v->profile && !v->fastuvmc) { av_log(avctx, AV_LOG_ERROR, "FASTUVMC unavailable in Simple Profile\n"); return -1; } - v->extended_mv = get_bits(gb, 1); //common + v->extended_mv = get_bits1(gb); //common if (!v->profile && v->extended_mv) { av_log(avctx, AV_LOG_ERROR, @@ -1207,9 +921,9 @@ static int decode_sequence_header(AVCodecContext *avctx, GetBitContext *gb) return -1; } v->dquant = get_bits(gb, 2); //common - v->vstransform = get_bits(gb, 1); //common + v->vstransform = get_bits1(gb); //common - v->res_transtab = get_bits(gb, 1); + v->res_transtab = get_bits1(gb); if (v->res_transtab) { av_log(avctx, AV_LOG_ERROR, @@ -1217,10 +931,10 @@ static int decode_sequence_header(AVCodecContext *avctx, GetBitContext *gb) return -1; } - v->overlap = get_bits(gb, 1); //common + v->overlap = get_bits1(gb); //common - v->s.resync_marker = get_bits(gb, 1); - v->rangered = get_bits(gb, 1); + v->s.resync_marker = get_bits1(gb); + v->rangered = get_bits1(gb); if (v->rangered && v->profile == PROFILE_SIMPLE) { av_log(avctx, AV_LOG_INFO, @@ -1230,8 +944,8 @@ static int decode_sequence_header(AVCodecContext *avctx, GetBitContext *gb) v->s.max_b_frames = avctx->max_b_frames = get_bits(gb, 3); //common v->quantizer_mode = get_bits(gb, 2); //common - v->finterpflag = get_bits(gb, 1); //common - v->res_rtm_flag = get_bits(gb, 1); //reserved + v->finterpflag = get_bits1(gb); //common + v->res_rtm_flag = get_bits1(gb); //reserved if (!v->res_rtm_flag) { // av_log(avctx, AV_LOG_ERROR, @@ -1240,6 +954,8 @@ static int decode_sequence_header(AVCodecContext *avctx, GetBitContext *gb) "Old WMV3 version detected, only I-frames will be decoded\n"); //return -1; } + //TODO: figure out what they mean (always 0x402F) + if(!v->res_fasttx) skip_bits(gb, 16); av_log(avctx, AV_LOG_DEBUG, "Profile %i:\nfrmrtq_postproc=%i, bitrtq_postproc=%i\n" "LoopFilter=%i, MultiRes=%i, FastUVMC=%i, Extended MV=%i\n" @@ -1273,7 +989,7 @@ static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb) v->frmrtq_postproc = get_bits(gb, 3); //common // (bitrate-32kbps)/64kbps v->bitrtq_postproc = get_bits(gb, 5); //common - v->postprocflag = get_bits(gb, 1); //common + v->postprocflag = get_bits1(gb); //common v->s.avctx->coded_width = (get_bits(gb, 12) + 1) << 1; v->s.avctx->coded_height = (get_bits(gb, 12) + 1) << 1; @@ -1283,7 +999,10 @@ static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb) v->interlace = get_bits1(gb); v->tfcntrflag = get_bits1(gb); v->finterpflag = get_bits1(gb); - get_bits1(gb); // reserved + skip_bits1(gb); // reserved + + v->s.h_edge_pos = v->s.avctx->coded_width; + v->s.v_edge_pos = v->s.avctx->coded_height; av_log(v->s.avctx, AV_LOG_DEBUG, "Advanced Profile level %i:\nfrmrtq_postproc=%i, bitrtq_postproc=%i\n" @@ -1303,18 +1022,19 @@ static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb) if(get_bits1(gb)) { //Display Info - decoding is not affected by it int w, h, ar = 0; av_log(v->s.avctx, AV_LOG_DEBUG, "Display extended info:\n"); - v->s.avctx->width = v->s.width = w = get_bits(gb, 14) + 1; - v->s.avctx->height = v->s.height = h = get_bits(gb, 14) + 1; + v->s.avctx->coded_width = w = get_bits(gb, 14) + 1; + v->s.avctx->coded_height = h = get_bits(gb, 14) + 1; av_log(v->s.avctx, AV_LOG_DEBUG, "Display dimensions: %ix%i\n", w, h); if(get_bits1(gb)) ar = get_bits(gb, 4); if(ar && ar < 14){ - v->s.avctx->sample_aspect_ratio = vc1_pixel_aspect[ar]; + v->s.avctx->sample_aspect_ratio = ff_vc1_pixel_aspect[ar]; }else if(ar == 15){ w = get_bits(gb, 8); h = get_bits(gb, 8); v->s.avctx->sample_aspect_ratio = (AVRational){w, h}; } + av_log(v->s.avctx, AV_LOG_DEBUG, "Aspect: %i:%i\n", v->s.avctx->sample_aspect_ratio.num, v->s.avctx->sample_aspect_ratio.den); if(get_bits1(gb)){ //framerate stuff if(get_bits1(gb)) { @@ -1325,8 +1045,8 @@ static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb) nr = get_bits(gb, 8); dr = get_bits(gb, 4); if(nr && nr < 8 && dr && dr < 3){ - v->s.avctx->time_base.num = fps_dr[dr - 1]; - v->s.avctx->time_base.den = fps_nr[nr - 1] * 1000; + v->s.avctx->time_base.num = ff_vc1_fps_dr[dr - 1]; + v->s.avctx->time_base.den = ff_vc1_fps_nr[nr - 1] * 1000; } } } @@ -1342,11 +1062,11 @@ static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb) if(v->hrd_param_flag) { int i; v->hrd_num_leaky_buckets = get_bits(gb, 5); - get_bits(gb, 4); //bitrate exponent - get_bits(gb, 4); //buffer size exponent + skip_bits(gb, 4); //bitrate exponent + skip_bits(gb, 4); //buffer size exponent for(i = 0; i < v->hrd_num_leaky_buckets; i++) { - get_bits(gb, 16); //hrd_rate[n] - get_bits(gb, 16); //hrd_buffer[n] + skip_bits(gb, 16); //hrd_rate[n] + skip_bits(gb, 16); //hrd_buffer[n] } } return 0; @@ -1355,13 +1075,13 @@ static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb) static int decode_entry_point(AVCodecContext *avctx, GetBitContext *gb) { VC1Context *v = avctx->priv_data; - int i, blink, clentry, refdist; + int i; av_log(avctx, AV_LOG_DEBUG, "Entry point: %08X\n", show_bits_long(gb, 32)); - blink = get_bits1(gb); // broken link - clentry = get_bits1(gb); // closed entry + v->broken_link = get_bits1(gb); + v->closed_entry = get_bits1(gb); v->panscanflag = get_bits1(gb); - refdist = get_bits1(gb); // refdist flag + v->refdist_flag = get_bits1(gb); v->s.loop_filter = get_bits1(gb); v->fastuvmc = get_bits1(gb); v->extended_mv = get_bits1(gb); @@ -1372,7 +1092,7 @@ static int decode_entry_point(AVCodecContext *avctx, GetBitContext *gb) if(v->hrd_param_flag){ for(i = 0; i < v->hrd_num_leaky_buckets; i++) { - get_bits(gb, 8); //hrd_full[n] + skip_bits(gb, 8); //hrd_full[n] } } @@ -1382,20 +1102,20 @@ static int decode_entry_point(AVCodecContext *avctx, GetBitContext *gb) } if(v->extended_mv) v->extended_dmv = get_bits1(gb); - if(get_bits1(gb)) { + if((v->range_mapy_flag = get_bits1(gb))) { av_log(avctx, AV_LOG_ERROR, "Luma scaling is not supported, expect wrong picture\n"); - skip_bits(gb, 3); // Y range, ignored for now + v->range_mapy = get_bits(gb, 3); } - if(get_bits1(gb)) { + if((v->range_mapuv_flag = get_bits1(gb))) { av_log(avctx, AV_LOG_ERROR, "Chroma scaling is not supported, expect wrong picture\n"); - skip_bits(gb, 3); // UV range, ignored for now + v->range_mapuv = get_bits(gb, 3); } av_log(avctx, AV_LOG_DEBUG, "Entry point info:\n" "BrokenLink=%i, ClosedEntry=%i, PanscanFlag=%i\n" "RefDist=%i, Postproc=%i, FastUVMC=%i, ExtMV=%i\n" "DQuant=%i, VSTransform=%i, Overlap=%i, Qmode=%i\n", - blink, clentry, v->panscanflag, refdist, v->s.loop_filter, + v->broken_link, v->closed_entry, v->panscanflag, v->refdist_flag, v->s.loop_filter, v->fastuvmc, v->extended_mv, v->dquant, v->vstransform, v->overlap, v->quantizer_mode); return 0; @@ -1405,41 +1125,42 @@ static int vc1_parse_frame_header(VC1Context *v, GetBitContext* gb) { int pqindex, lowquant, status; - if(v->finterpflag) v->interpfrm = get_bits(gb, 1); + if(v->finterpflag) v->interpfrm = get_bits1(gb); skip_bits(gb, 2); //framecnt unused v->rangeredfrm = 0; - if (v->rangered) v->rangeredfrm = get_bits(gb, 1); - v->s.pict_type = get_bits(gb, 1); + if (v->rangered) v->rangeredfrm = get_bits1(gb); + v->s.pict_type = get_bits1(gb); if (v->s.avctx->max_b_frames) { if (!v->s.pict_type) { - if (get_bits(gb, 1)) v->s.pict_type = I_TYPE; - else v->s.pict_type = B_TYPE; - } else v->s.pict_type = P_TYPE; - } else v->s.pict_type = v->s.pict_type ? P_TYPE : I_TYPE; + if (get_bits1(gb)) v->s.pict_type = FF_I_TYPE; + else v->s.pict_type = FF_B_TYPE; + } else v->s.pict_type = FF_P_TYPE; + } else v->s.pict_type = v->s.pict_type ? FF_P_TYPE : FF_I_TYPE; v->bi_type = 0; - if(v->s.pict_type == B_TYPE) { - v->bfraction = get_vlc2(gb, vc1_bfraction_vlc.table, VC1_BFRACTION_VLC_BITS, 1); - v->bfraction = vc1_bfraction_lut[v->bfraction]; + if(v->s.pict_type == FF_B_TYPE) { + v->bfraction_lut_index = get_vlc2(gb, ff_vc1_bfraction_vlc.table, VC1_BFRACTION_VLC_BITS, 1); + v->bfraction = ff_vc1_bfraction_lut[v->bfraction_lut_index]; if(v->bfraction == 0) { - v->s.pict_type = BI_TYPE; + v->s.pict_type = FF_BI_TYPE; } } - if(v->s.pict_type == I_TYPE || v->s.pict_type == BI_TYPE) - get_bits(gb, 7); // skip buffer fullness + if(v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_BI_TYPE) + skip_bits(gb, 7); // skip buffer fullness /* calculate RND */ - if(v->s.pict_type == I_TYPE || v->s.pict_type == BI_TYPE) + if(v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_BI_TYPE) v->rnd = 1; - if(v->s.pict_type == P_TYPE) + if(v->s.pict_type == FF_P_TYPE) v->rnd ^= 1; /* Quantizer stuff */ pqindex = get_bits(gb, 5); + if(!pqindex) return -1; if (v->quantizer_mode == QUANT_FRAME_IMPLICIT) - v->pq = pquant_table[0][pqindex]; + v->pq = ff_vc1_pquant_table[0][pqindex]; else - v->pq = pquant_table[1][pqindex]; + v->pq = ff_vc1_pquant_table[1][pqindex]; v->pquantizer = 1; if (v->quantizer_mode == QUANT_FRAME_IMPLICIT) @@ -1447,43 +1168,38 @@ static int vc1_parse_frame_header(VC1Context *v, GetBitContext* gb) if (v->quantizer_mode == QUANT_NON_UNIFORM) v->pquantizer = 0; v->pqindex = pqindex; - if (pqindex < 9) v->halfpq = get_bits(gb, 1); + if (pqindex < 9) v->halfpq = get_bits1(gb); else v->halfpq = 0; if (v->quantizer_mode == QUANT_FRAME_EXPLICIT) - v->pquantizer = get_bits(gb, 1); + v->pquantizer = get_bits1(gb); v->dquantfrm = 0; - if (v->extended_mv == 1) v->mvrange = get_prefix(gb, 0, 3); + if (v->extended_mv == 1) v->mvrange = get_unary(gb, 0, 3); v->k_x = v->mvrange + 9 + (v->mvrange >> 1); //k_x can be 9 10 12 13 v->k_y = v->mvrange + 8; //k_y can be 8 9 10 11 v->range_x = 1 << (v->k_x - 1); v->range_y = 1 << (v->k_y - 1); - if (v->profile == PROFILE_ADVANCED) - { - if (v->postprocflag) v->postproc = get_bits(gb, 1); - } - else - if (v->multires && v->s.pict_type != B_TYPE) v->respic = get_bits(gb, 2); + if (v->multires && v->s.pict_type != FF_B_TYPE) v->respic = get_bits(gb, 2); - if(v->res_x8 && (v->s.pict_type == I_TYPE || v->s.pict_type == BI_TYPE)){ - if(get_bits1(gb))return -1; - } + if(v->res_x8 && (v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_BI_TYPE)){ + v->x8_type = get_bits1(gb); + }else v->x8_type = 0; //av_log(v->s.avctx, AV_LOG_INFO, "%c Frame: QP=[%i]%i (+%i/2) %i\n", -// (v->s.pict_type == P_TYPE) ? 'P' : ((v->s.pict_type == I_TYPE) ? 'I' : 'B'), pqindex, v->pq, v->halfpq, v->rangeredfrm); +// (v->s.pict_type == FF_P_TYPE) ? 'P' : ((v->s.pict_type == FF_I_TYPE) ? 'I' : 'B'), pqindex, v->pq, v->halfpq, v->rangeredfrm); - if(v->s.pict_type == I_TYPE || v->s.pict_type == P_TYPE) v->use_ic = 0; + if(v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_P_TYPE) v->use_ic = 0; switch(v->s.pict_type) { - case P_TYPE: + case FF_P_TYPE: if (v->pq < 5) v->tt_index = 0; else if(v->pq < 13) v->tt_index = 1; else v->tt_index = 2; lowquant = (v->pq > 12) ? 0 : 1; - v->mv_mode = mv_pmode_table[lowquant][get_prefix(gb, 1, 4)]; + v->mv_mode = ff_vc1_mv_pmode_table[lowquant][get_unary(gb, 1, 4)]; if (v->mv_mode == MV_PMODE_INTENSITY_COMP) { int scale, shift, i; - v->mv_mode2 = mv_pmode_table2[lowquant][get_prefix(gb, 1, 3)]; + v->mv_mode2 = ff_vc1_mv_pmode_table2[lowquant][get_unary(gb, 1, 3)]; v->lumscale = get_bits(gb, 6); v->lumshift = get_bits(gb, 6); v->use_ic = 1; @@ -1534,8 +1250,8 @@ static int vc1_parse_frame_header(VC1Context *v, GetBitContext* gb) "Imode: %i, Invert: %i\n", status>>1, status&1); /* Hopefully this is correct for P frames */ - v->s.mv_table_index = get_bits(gb, 2); //but using vc1_ tables - v->cbpcy_vlc = &vc1_cbpcy_p_vlc[get_bits(gb, 2)]; + v->s.mv_table_index = get_bits(gb, 2); //but using ff_vc1_ tables + v->cbpcy_vlc = &ff_vc1_cbpcy_p_vlc[get_bits(gb, 2)]; if (v->dquant) { @@ -1546,17 +1262,17 @@ static int vc1_parse_frame_header(VC1Context *v, GetBitContext* gb) v->ttfrm = 0; //FIXME Is that so ? if (v->vstransform) { - v->ttmbf = get_bits(gb, 1); + v->ttmbf = get_bits1(gb); if (v->ttmbf) { - v->ttfrm = ttfrm_to_tt[get_bits(gb, 2)]; + v->ttfrm = ff_vc1_ttfrm_to_tt[get_bits(gb, 2)]; } } else { v->ttmbf = 1; v->ttfrm = TT_8X8; } break; - case B_TYPE: + case FF_B_TYPE: if (v->pq < 5) v->tt_index = 0; else if(v->pq < 13) v->tt_index = 1; else v->tt_index = 2; @@ -1576,7 +1292,7 @@ static int vc1_parse_frame_header(VC1Context *v, GetBitContext* gb) "Imode: %i, Invert: %i\n", status>>1, status&1); v->s.mv_table_index = get_bits(gb, 2); - v->cbpcy_vlc = &vc1_cbpcy_p_vlc[get_bits(gb, 2)]; + v->cbpcy_vlc = &ff_vc1_cbpcy_p_vlc[get_bits(gb, 2)]; if (v->dquant) { @@ -1587,10 +1303,10 @@ static int vc1_parse_frame_header(VC1Context *v, GetBitContext* gb) v->ttfrm = 0; if (v->vstransform) { - v->ttmbf = get_bits(gb, 1); + v->ttmbf = get_bits1(gb); if (v->ttmbf) { - v->ttfrm = ttfrm_to_tt[get_bits(gb, 2)]; + v->ttfrm = ff_vc1_ttfrm_to_tt[get_bits(gb, 2)]; } } else { v->ttmbf = 1; @@ -1599,17 +1315,20 @@ static int vc1_parse_frame_header(VC1Context *v, GetBitContext* gb) break; } - /* AC Syntax */ - v->c_ac_table_index = decode012(gb); - if (v->s.pict_type == I_TYPE || v->s.pict_type == BI_TYPE) + if(!v->x8_type) { - v->y_ac_table_index = decode012(gb); + /* AC Syntax */ + v->c_ac_table_index = decode012(gb); + if (v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_BI_TYPE) + { + v->y_ac_table_index = decode012(gb); + } + /* DC Syntax */ + v->s.dc_table_index = get_bits1(gb); } - /* DC Syntax */ - v->s.dc_table_index = get_bits(gb, 1); - if(v->s.pict_type == BI_TYPE) { - v->s.pict_type = B_TYPE; + if(v->s.pict_type == FF_BI_TYPE) { + v->s.pict_type = FF_B_TYPE; v->bi_type = 1; } return 0; @@ -1626,26 +1345,26 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) v->fcm = decode012(gb); if(v->fcm) return -1; // interlaced frames/fields are not implemented } - switch(get_prefix(gb, 0, 4)) { + switch(get_unary(gb, 0, 4)) { case 0: - v->s.pict_type = P_TYPE; + v->s.pict_type = FF_P_TYPE; break; case 1: - v->s.pict_type = B_TYPE; + v->s.pict_type = FF_B_TYPE; break; case 2: - v->s.pict_type = I_TYPE; + v->s.pict_type = FF_I_TYPE; break; case 3: - v->s.pict_type = BI_TYPE; + v->s.pict_type = FF_BI_TYPE; break; case 4: - v->s.pict_type = P_TYPE; // skipped pic + v->s.pict_type = FF_P_TYPE; // skipped pic v->p_frame_skipped = 1; return 0; } if(v->tfcntrflag) - get_bits(gb, 8); + skip_bits(gb, 8); if(v->broadcast) { if(!v->interlace || v->psf) { v->rptfrm = get_bits(gb, 2); @@ -1660,20 +1379,21 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) v->rnd = get_bits1(gb); if(v->interlace) v->uvsamp = get_bits1(gb); - if(v->finterpflag) v->interpfrm = get_bits(gb, 1); - if(v->s.pict_type == B_TYPE) { - v->bfraction = get_vlc2(gb, vc1_bfraction_vlc.table, VC1_BFRACTION_VLC_BITS, 1); - v->bfraction = vc1_bfraction_lut[v->bfraction]; + if(v->finterpflag) v->interpfrm = get_bits1(gb); + if(v->s.pict_type == FF_B_TYPE) { + v->bfraction_lut_index = get_vlc2(gb, ff_vc1_bfraction_vlc.table, VC1_BFRACTION_VLC_BITS, 1); + v->bfraction = ff_vc1_bfraction_lut[v->bfraction_lut_index]; if(v->bfraction == 0) { - v->s.pict_type = BI_TYPE; /* XXX: should not happen here */ + v->s.pict_type = FF_BI_TYPE; /* XXX: should not happen here */ } } pqindex = get_bits(gb, 5); + if(!pqindex) return -1; v->pqindex = pqindex; if (v->quantizer_mode == QUANT_FRAME_IMPLICIT) - v->pq = pquant_table[0][pqindex]; + v->pq = ff_vc1_pquant_table[0][pqindex]; else - v->pq = pquant_table[1][pqindex]; + v->pq = ff_vc1_pquant_table[1][pqindex]; v->pquantizer = 1; if (v->quantizer_mode == QUANT_FRAME_IMPLICIT) @@ -1681,16 +1401,18 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) if (v->quantizer_mode == QUANT_NON_UNIFORM) v->pquantizer = 0; v->pqindex = pqindex; - if (pqindex < 9) v->halfpq = get_bits(gb, 1); + if (pqindex < 9) v->halfpq = get_bits1(gb); else v->halfpq = 0; if (v->quantizer_mode == QUANT_FRAME_EXPLICIT) - v->pquantizer = get_bits(gb, 1); + v->pquantizer = get_bits1(gb); + if(v->postprocflag) + v->postproc = get_bits(gb, 2); - if(v->s.pict_type == I_TYPE || v->s.pict_type == P_TYPE) v->use_ic = 0; + if(v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_P_TYPE) v->use_ic = 0; switch(v->s.pict_type) { - case I_TYPE: - case BI_TYPE: + case FF_I_TYPE: + case FF_BI_TYPE: status = bitplane_decoding(v->acpred_plane, &v->acpred_is_raw, v); if (status < 0) return -1; av_log(v->s.avctx, AV_LOG_DEBUG, "ACPRED plane encoding: " @@ -1706,10 +1428,8 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) } } break; - case P_TYPE: - if(v->postprocflag) - v->postproc = get_bits1(gb); - if (v->extended_mv) v->mvrange = get_prefix(gb, 0, 3); + case FF_P_TYPE: + if (v->extended_mv) v->mvrange = get_unary(gb, 0, 3); else v->mvrange = 0; v->k_x = v->mvrange + 9 + (v->mvrange >> 1); //k_x can be 9 10 12 13 v->k_y = v->mvrange + 8; //k_y can be 8 9 10 11 @@ -1721,11 +1441,11 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) else v->tt_index = 2; lowquant = (v->pq > 12) ? 0 : 1; - v->mv_mode = mv_pmode_table[lowquant][get_prefix(gb, 1, 4)]; + v->mv_mode = ff_vc1_mv_pmode_table[lowquant][get_unary(gb, 1, 4)]; if (v->mv_mode == MV_PMODE_INTENSITY_COMP) { int scale, shift, i; - v->mv_mode2 = mv_pmode_table2[lowquant][get_prefix(gb, 1, 3)]; + v->mv_mode2 = ff_vc1_mv_pmode_table2[lowquant][get_unary(gb, 1, 3)]; v->lumscale = get_bits(gb, 6); v->lumshift = get_bits(gb, 6); /* fill lookup tables for intensity compensation */ @@ -1776,8 +1496,8 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) "Imode: %i, Invert: %i\n", status>>1, status&1); /* Hopefully this is correct for P frames */ - v->s.mv_table_index = get_bits(gb, 2); //but using vc1_ tables - v->cbpcy_vlc = &vc1_cbpcy_p_vlc[get_bits(gb, 2)]; + v->s.mv_table_index = get_bits(gb, 2); //but using ff_vc1_ tables + v->cbpcy_vlc = &ff_vc1_cbpcy_p_vlc[get_bits(gb, 2)]; if (v->dquant) { av_log(v->s.avctx, AV_LOG_DEBUG, "VOP DQuant info\n"); @@ -1787,20 +1507,18 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) v->ttfrm = 0; //FIXME Is that so ? if (v->vstransform) { - v->ttmbf = get_bits(gb, 1); + v->ttmbf = get_bits1(gb); if (v->ttmbf) { - v->ttfrm = ttfrm_to_tt[get_bits(gb, 2)]; + v->ttfrm = ff_vc1_ttfrm_to_tt[get_bits(gb, 2)]; } } else { v->ttmbf = 1; v->ttfrm = TT_8X8; } break; - case B_TYPE: - if(v->postprocflag) - v->postproc = get_bits1(gb); - if (v->extended_mv) v->mvrange = get_prefix(gb, 0, 3); + case FF_B_TYPE: + if (v->extended_mv) v->mvrange = get_unary(gb, 0, 3); else v->mvrange = 0; v->k_x = v->mvrange + 9 + (v->mvrange >> 1); //k_x can be 9 10 12 13 v->k_y = v->mvrange + 8; //k_y can be 8 9 10 11 @@ -1826,7 +1544,7 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) "Imode: %i, Invert: %i\n", status>>1, status&1); v->s.mv_table_index = get_bits(gb, 2); - v->cbpcy_vlc = &vc1_cbpcy_p_vlc[get_bits(gb, 2)]; + v->cbpcy_vlc = &ff_vc1_cbpcy_p_vlc[get_bits(gb, 2)]; if (v->dquant) { @@ -1837,10 +1555,10 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) v->ttfrm = 0; if (v->vstransform) { - v->ttmbf = get_bits(gb, 1); + v->ttmbf = get_bits1(gb); if (v->ttmbf) { - v->ttfrm = ttfrm_to_tt[get_bits(gb, 2)]; + v->ttfrm = ff_vc1_ttfrm_to_tt[get_bits(gb, 2)]; } } else { v->ttmbf = 1; @@ -1851,20 +1569,20 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) /* AC Syntax */ v->c_ac_table_index = decode012(gb); - if (v->s.pict_type == I_TYPE || v->s.pict_type == BI_TYPE) + if (v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_BI_TYPE) { v->y_ac_table_index = decode012(gb); } /* DC Syntax */ - v->s.dc_table_index = get_bits(gb, 1); - if ((v->s.pict_type == I_TYPE || v->s.pict_type == BI_TYPE) && v->dquant) { + v->s.dc_table_index = get_bits1(gb); + if ((v->s.pict_type == FF_I_TYPE || v->s.pict_type == FF_BI_TYPE) && v->dquant) { av_log(v->s.avctx, AV_LOG_DEBUG, "VOP DQuant info\n"); vop_dquant_decoding(v); } v->bi_type = 0; - if(v->s.pict_type == BI_TYPE) { - v->s.pict_type = B_TYPE; + if(v->s.pict_type == FF_BI_TYPE) { + v->s.pict_type = FF_B_TYPE; v->bi_type = 1; } return 0; @@ -1872,7 +1590,7 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) /***********************************************************************/ /** - * @defgroup block VC-1 Block-level functions + * @defgroup vc1block VC-1 Block-level functions * @see 7.1.4, p91 and 8.1.1.7, p(1)04 * @{ */ @@ -1889,7 +1607,7 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) { \ if (v->dqbilevel) \ { \ - mquant = (get_bits(gb, 1)) ? v->altpq : v->pq; \ + mquant = (get_bits1(gb)) ? v->altpq : v->pq; \ } \ else \ { \ @@ -1922,7 +1640,7 @@ static int vc1_parse_frame_header_adv(VC1Context *v, GetBitContext* gb) * @param _dmv_y Vertical differential for decoded MV */ #define GET_MVDATA(_dmv_x, _dmv_y) \ - index = 1 + get_vlc2(gb, vc1_mv_diff_vlc[s->mv_table_index].table,\ + index = 1 + get_vlc2(gb, ff_vc1_mv_diff_vlc[s->mv_table_index].table,\ VC1_MV_DIFF_VLC_BITS, 2); \ if (index > 36) \ { \ @@ -1983,6 +1701,8 @@ static inline void vc1_pred_mv(MpegEncContext *s, int n, int dmv_x, int dmv_y, i if(s->mb_intra){ s->mv[0][n][0] = s->current_picture.motion_val[0][xy][0] = 0; s->mv[0][n][1] = s->current_picture.motion_val[0][xy][1] = 0; + s->current_picture.motion_val[1][xy][0] = 0; + s->current_picture.motion_val[1][xy][1] = 0; if(mv1) { /* duplicate motion data for 1-MV block */ s->current_picture.motion_val[0][xy + 1][0] = 0; s->current_picture.motion_val[0][xy + 1][1] = 0; @@ -1990,6 +1710,12 @@ static inline void vc1_pred_mv(MpegEncContext *s, int n, int dmv_x, int dmv_y, i s->current_picture.motion_val[0][xy + wrap][1] = 0; s->current_picture.motion_val[0][xy + wrap + 1][0] = 0; s->current_picture.motion_val[0][xy + wrap + 1][1] = 0; + s->current_picture.motion_val[1][xy + 1][0] = 0; + s->current_picture.motion_val[1][xy + 1][1] = 0; + s->current_picture.motion_val[1][xy + wrap][0] = 0; + s->current_picture.motion_val[1][xy + wrap][1] = 0; + s->current_picture.motion_val[1][xy + wrap + 1][0] = 0; + s->current_picture.motion_val[1][xy + wrap + 1][1] = 0; } return; } @@ -2118,10 +1844,17 @@ static void vc1_interp_mc(VC1Context *v) uvsrc_x = s->mb_x * 8 + (uvmx >> 2); uvsrc_y = s->mb_y * 8 + (uvmy >> 2); - src_x = av_clip( src_x, -16, s->mb_width * 16); - src_y = av_clip( src_y, -16, s->mb_height * 16); - uvsrc_x = av_clip(uvsrc_x, -8, s->mb_width * 8); - uvsrc_y = av_clip(uvsrc_y, -8, s->mb_height * 8); + if(v->profile != PROFILE_ADVANCED){ + src_x = av_clip( src_x, -16, s->mb_width * 16); + src_y = av_clip( src_y, -16, s->mb_height * 16); + uvsrc_x = av_clip(uvsrc_x, -8, s->mb_width * 8); + uvsrc_y = av_clip(uvsrc_y, -8, s->mb_height * 8); + }else{ + src_x = av_clip( src_x, -17, s->avctx->coded_width); + src_y = av_clip( src_y, -18, s->avctx->coded_height + 1); + uvsrc_x = av_clip(uvsrc_x, -8, s->avctx->coded_width >> 1); + uvsrc_y = av_clip(uvsrc_y, -8, s->avctx->coded_height >> 1); + } srcY += src_y * s->linesize + src_x; srcU += uvsrc_y * s->uvlinesize + uvsrc_x; @@ -2263,6 +1996,12 @@ static inline void vc1_pred_b_mv(VC1Context *v, int dmv_x[2], int dmv_y[2], int s->mv[0][0][1] = scale_mv(s->next_picture.motion_val[1][xy][1], v->bfraction, 0, s->quarter_sample); s->mv[1][0][0] = scale_mv(s->next_picture.motion_val[1][xy][0], v->bfraction, 1, s->quarter_sample); s->mv[1][0][1] = scale_mv(s->next_picture.motion_val[1][xy][1], v->bfraction, 1, s->quarter_sample); + + /* Pullback predicted motion vectors as specified in 8.4.5.4 */ + s->mv[0][0][0] = av_clip(s->mv[0][0][0], -60 - (s->mb_x << 6), (s->mb_width << 6) - 4 - (s->mb_x << 6)); + s->mv[0][0][1] = av_clip(s->mv[0][0][1], -60 - (s->mb_y << 6), (s->mb_height << 6) - 4 - (s->mb_y << 6)); + s->mv[1][0][0] = av_clip(s->mv[1][0][0], -60 - (s->mb_x << 6), (s->mb_width << 6) - 4 - (s->mb_x << 6)); + s->mv[1][0][1] = av_clip(s->mv[1][0][1], -60 - (s->mb_y << 6), (s->mb_height << 6) - 4 - (s->mb_y << 6)); if(direct) { s->current_picture.motion_val[0][xy][0] = s->mv[0][0][0]; s->current_picture.motion_val[0][xy][1] = s->mv[0][0][1]; @@ -2277,6 +2016,7 @@ static inline void vc1_pred_b_mv(VC1Context *v, int dmv_x[2], int dmv_y[2], int off = (s->mb_x == (s->mb_width - 1)) ? -2 : 2; B = s->current_picture.motion_val[0][xy - wrap*2 + off]; + if(!s->mb_x) C[0] = C[1] = 0; if(!s->first_slice_line) { // predictor A is not out of bounds if(s->mb_width == 1) { px = A[0]; @@ -2354,6 +2094,7 @@ static inline void vc1_pred_b_mv(VC1Context *v, int dmv_x[2], int dmv_y[2], int off = (s->mb_x == (s->mb_width - 1)) ? -2 : 2; B = s->current_picture.motion_val[1][xy - wrap*2 + off]; + if(!s->mb_x) C[0] = C[1] = 0; if(!s->first_slice_line) { // predictor A is not out of bounds if(s->mb_width == 1) { px = A[0]; @@ -2435,6 +2176,8 @@ static inline void vc1_pred_b_mv(VC1Context *v, int dmv_x[2], int dmv_y[2], int /** Get predicted DC value for I-frames only * prediction dir: left=0, top=1 * @param s MpegEncContext + * @param overlap flag indicating that overlap filtering is used + * @param pq integer part of picture quantizer * @param[in] n block index in the current MB * @param dc_val_ptr Pointer to DC predictor * @param dir_ptr Prediction direction for use in AC prediction @@ -2495,7 +2238,11 @@ static inline int vc1_i_pred_dc(MpegEncContext *s, int overlap, int pq, int n, /** Get predicted DC value * prediction dir: left=0, top=1 * @param s MpegEncContext + * @param overlap flag indicating that overlap filtering is used + * @param pq integer part of picture quantizer * @param[in] n block index in the current MB + * @param a_avail flag indicating top block availability + * @param c_avail flag indicating left block availability * @param dc_val_ptr Pointer to DC predictor * @param dir_ptr Prediction direction for use in AC prediction */ @@ -2526,12 +2273,12 @@ static inline int vc1_pred_dc(MpegEncContext *s, int overlap, int pq, int n, if(c_avail && (n!= 1 && n!=3)) { q2 = s->current_picture.qscale_table[mb_pos - 1]; if(q2 && q2 != q1) - c = (c * s->y_dc_scale_table[q2] * vc1_dqscale[s->y_dc_scale_table[q1] - 1] + 0x20000) >> 18; + c = (c * s->y_dc_scale_table[q2] * ff_vc1_dqscale[s->y_dc_scale_table[q1] - 1] + 0x20000) >> 18; } if(a_avail && (n!= 2 && n!=3)) { q2 = s->current_picture.qscale_table[mb_pos - s->mb_stride]; if(q2 && q2 != q1) - a = (a * s->y_dc_scale_table[q2] * vc1_dqscale[s->y_dc_scale_table[q1] - 1] + 0x20000) >> 18; + a = (a * s->y_dc_scale_table[q2] * ff_vc1_dqscale[s->y_dc_scale_table[q1] - 1] + 0x20000) >> 18; } if(a_avail && c_avail && (n!=3)) { int off = mb_pos; @@ -2539,7 +2286,7 @@ static inline int vc1_pred_dc(MpegEncContext *s, int overlap, int pq, int n, if(n != 2) off -= s->mb_stride; q2 = s->current_picture.qscale_table[off]; if(q2 && q2 != q1) - b = (b * s->y_dc_scale_table[q2] * vc1_dqscale[s->y_dc_scale_table[q1] - 1] + 0x20000) >> 18; + b = (b * s->y_dc_scale_table[q2] * ff_vc1_dqscale[s->y_dc_scale_table[q1] - 1] + 0x20000) >> 18; } if(a_avail && c_avail) { @@ -2566,9 +2313,10 @@ static inline int vc1_pred_dc(MpegEncContext *s, int overlap, int pq, int n, return pred; } +/** @} */ // Block group /** - * @defgroup std_mb VC1 Macroblock-level functions in Simple/Main Profiles + * @defgroup vc1_std_mb VC1 Macroblock-level functions in Simple/Main Profiles * @see 7.1.4, p91 and 8.1.1.7, p(1)04 * @{ */ @@ -2605,6 +2353,7 @@ static inline int vc1_coded_block_pred(MpegEncContext * s, int n, uint8_t **code * @param last Last coefficient * @param skip How much zero coefficients to skip * @param value Decoded AC coefficient value + * @param codingset set of VLC to decode data * @see 8.1.3.4 */ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip, int *value, int codingset) @@ -2612,17 +2361,17 @@ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip, int *value, GetBitContext *gb = &v->s.gb; int index, escape, run = 0, level = 0, lst = 0; - index = get_vlc2(gb, vc1_ac_coeff_table[codingset].table, AC_VLC_BITS, 3); + index = get_vlc2(gb, ff_vc1_ac_coeff_table[codingset].table, AC_VLC_BITS, 3); if (index != vc1_ac_sizes[codingset] - 1) { run = vc1_index_decode_table[codingset][index][0]; level = vc1_index_decode_table[codingset][index][1]; lst = index >= vc1_last_decode_table[codingset]; - if(get_bits(gb, 1)) + if(get_bits1(gb)) level = -level; } else { escape = decode210(gb); if (escape != 2) { - index = get_vlc2(gb, vc1_ac_coeff_table[codingset].table, AC_VLC_BITS, 3); + index = get_vlc2(gb, ff_vc1_ac_coeff_table[codingset].table, AC_VLC_BITS, 3); run = vc1_index_decode_table[codingset][index][0]; level = vc1_index_decode_table[codingset][index][1]; lst = index >= vc1_last_decode_table[codingset]; @@ -2637,23 +2386,23 @@ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip, int *value, else run += vc1_delta_run_table[codingset][level] + 1; } - if(get_bits(gb, 1)) + if(get_bits1(gb)) level = -level; } else { int sign; - lst = get_bits(gb, 1); + lst = get_bits1(gb); if(v->s.esc3_level_length == 0) { if(v->pq < 8 || v->dquantfrm) { // table 59 v->s.esc3_level_length = get_bits(gb, 3); if(!v->s.esc3_level_length) v->s.esc3_level_length = get_bits(gb, 2) + 8; } else { //table 60 - v->s.esc3_level_length = get_prefix(gb, 1, 6) + 2; + v->s.esc3_level_length = get_unary(gb, 1, 6) + 2; } v->s.esc3_run_length = 3 + get_bits(gb, 2); } run = get_bits(gb, v->s.esc3_run_length); - sign = get_bits(gb, 1); + sign = get_bits1(gb); level = get_bits(gb, v->s.esc3_level_length); if(sign) level = -level; @@ -2668,6 +2417,7 @@ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip, int *value, /** Decode intra block in intra frames - should be faster than decode_intra_block * @param v VC1Context * @param block block to decode + * @param[in] n subblock index * @param coded are AC coeffs present or not * @param codingset set of VLC to decode data */ @@ -2705,9 +2455,9 @@ static int vc1_decode_i_block(VC1Context *v, DCTELEM block[64], int n, int coded if (v->pq == 1) dcdiff = (dcdiff<<2) + get_bits(gb, 2) - 3; else if (v->pq == 2) - dcdiff = (dcdiff<<1) + get_bits(gb, 1) - 1; + dcdiff = (dcdiff<<1) + get_bits1(gb) - 1; } - if (get_bits(gb, 1)) + if (get_bits1(gb)) dcdiff = -dcdiff; } @@ -2741,11 +2491,11 @@ static int vc1_decode_i_block(VC1Context *v, DCTELEM block[64], int n, int coded if(v->s.ac_pred) { if(!dc_pred_dir) - zz_table = vc1_horizontal_zz; + zz_table = wmv1_scantable[2]; else - zz_table = vc1_vertical_zz; + zz_table = wmv1_scantable[3]; } else - zz_table = vc1_normal_zz; + zz_table = wmv1_scantable[1]; ac_val = s->ac_val[0][0] + s->block_index[n] * 16; ac_val2 = ac_val; @@ -2833,8 +2583,10 @@ not_coded: /** Decode intra block in intra frames - should be faster than decode_intra_block * @param v VC1Context * @param block block to decode + * @param[in] n subblock number * @param coded are AC coeffs present or not * @param codingset set of VLC to decode data + * @param mquant quantizer value for this macroblock */ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int coded, int codingset, int mquant) { @@ -2875,9 +2627,9 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c if (mquant == 1) dcdiff = (dcdiff<<2) + get_bits(gb, 2) - 3; else if (mquant == 2) - dcdiff = (dcdiff<<1) + get_bits(gb, 1) - 1; + dcdiff = (dcdiff<<1) + get_bits1(gb) - 1; } - if (get_bits(gb, 1)) + if (get_bits1(gb)) dcdiff = -dcdiff; } @@ -2898,14 +2650,12 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c //AC Decoding i = 1; - /* check if AC is needed at all and adjust direction if needed */ - if(!a_avail) dc_pred_dir = 1; - if(!c_avail) dc_pred_dir = 0; + /* check if AC is needed at all */ if(!a_avail && !c_avail) use_pred = 0; ac_val = s->ac_val[0][0] + s->block_index[n] * 16; ac_val2 = ac_val; - scale = mquant * 2 + v->halfpq; + scale = mquant * 2 + ((mquant == v->pq) ? v->halfpq : 0); if(dc_pred_dir) //left ac_val -= 16; @@ -2915,7 +2665,9 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c q1 = s->current_picture.qscale_table[mb_pos]; if(dc_pred_dir && c_avail && mb_pos) q2 = s->current_picture.qscale_table[mb_pos - 1]; if(!dc_pred_dir && a_avail && mb_pos >= s->mb_stride) q2 = s->current_picture.qscale_table[mb_pos - s->mb_stride]; - if(n && n<4) q2 = q1; + if(dc_pred_dir && n==1) q2 = q1; + if(!dc_pred_dir && n==2) q2 = q1; + if(n==3) q2 = q1; if(coded) { int last = 0, skip, value; @@ -2924,11 +2676,11 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c if(v->s.ac_pred) { if(!dc_pred_dir) - zz_table = vc1_horizontal_zz; + zz_table = wmv1_scantable[2]; else - zz_table = vc1_vertical_zz; + zz_table = wmv1_scantable[3]; } else - zz_table = vc1_normal_zz; + zz_table = wmv1_scantable[1]; while (!last) { vc1_decode_ac_coeff(v, &last, &skip, &value, codingset); @@ -2947,10 +2699,10 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c if(dc_pred_dir) { //left for(k = 1; k < 8; k++) - block[k << 3] += (ac_val[k] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + block[k << 3] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } else { //top for(k = 1; k < 8; k++) - block[k] += (ac_val[k + 8] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + block[k] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } } else { if(dc_pred_dir) { //left @@ -2988,7 +2740,7 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; for(k = 1; k < 8; k++) - ac_val2[k] = (ac_val2[k] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + ac_val2[k] = (ac_val2[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } } } else {//top @@ -2998,7 +2750,7 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; for(k = 1; k < 8; k++) - ac_val2[k + 8] = (ac_val2[k + 8] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + ac_val2[k + 8] = (ac_val2[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } } } @@ -3029,6 +2781,7 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, int c /** Decode intra block in inter frames - more generic version than vc1_decode_i_block * @param v VC1Context * @param block block to decode + * @param[in] n subblock index * @param coded are AC coeffs present or not * @param mquant block quantizer * @param codingset set of VLC to decode data @@ -3079,9 +2832,9 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c if (mquant == 1) dcdiff = (dcdiff<<2) + get_bits(gb, 2) - 3; else if (mquant == 2) - dcdiff = (dcdiff<<1) + get_bits(gb, 1) - 1; + dcdiff = (dcdiff<<1) + get_bits1(gb) - 1; } - if (get_bits(gb, 1)) + if (get_bits1(gb)) dcdiff = -dcdiff; } @@ -3120,14 +2873,16 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c q1 = s->current_picture.qscale_table[mb_pos]; if(dc_pred_dir && c_avail && mb_pos) q2 = s->current_picture.qscale_table[mb_pos - 1]; if(!dc_pred_dir && a_avail && mb_pos >= s->mb_stride) q2 = s->current_picture.qscale_table[mb_pos - s->mb_stride]; - if(n && n<4) q2 = q1; + if(dc_pred_dir && n==1) q2 = q1; + if(!dc_pred_dir && n==2) q2 = q1; + if(n==3) q2 = q1; if(coded) { int last = 0, skip, value; const int8_t *zz_table; int k; - zz_table = vc1_simple_progressive_8x8_zz; + zz_table = wmv1_scantable[0]; while (!last) { vc1_decode_ac_coeff(v, &last, &skip, &value, codingset); @@ -3146,10 +2901,10 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c if(dc_pred_dir) { //left for(k = 1; k < 8; k++) - block[k << 3] += (ac_val[k] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + block[k << 3] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } else { //top for(k = 1; k < 8; k++) - block[k] += (ac_val[k + 8] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + block[k] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } } else { if(dc_pred_dir) { //left @@ -3187,7 +2942,7 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; for(k = 1; k < 8; k++) - ac_val2[k] = (ac_val2[k] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + ac_val2[k] = (ac_val2[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } } } else {//top @@ -3197,7 +2952,7 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; for(k = 1; k < 8; k++) - ac_val2[k + 8] = (ac_val2[k + 8] * q2 * vc1_dqscale[q1 - 1] + 0x20000) >> 18; + ac_val2[k + 8] = (ac_val2[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } } } @@ -3227,7 +2982,8 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, int c /** Decode P block */ -static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquant, int ttmb, int first_block) +static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquant, int ttmb, int first_block, + uint8_t *dst, int linesize, int skip_block, int apply_filter, int cbp_top, int cbp_left) { MpegEncContext *s = &v->s; GetBitContext *gb = &s->gb; @@ -3235,12 +2991,13 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan int subblkpat = 0; int scale, off, idx, last, skip, value; int ttblk = ttmb & 7; + int pat = 0; if(ttmb == -1) { - ttblk = ttblk_to_tt[v->tt_index][get_vlc2(gb, vc1_ttblk_vlc[v->tt_index].table, VC1_TTBLK_VLC_BITS, 1)]; + ttblk = ff_vc1_ttblk_to_tt[v->tt_index][get_vlc2(gb, ff_vc1_ttblk_vlc[v->tt_index].table, VC1_TTBLK_VLC_BITS, 1)]; } if(ttblk == TT_4X4) { - subblkpat = ~(get_vlc2(gb, vc1_subblkpat_vlc[v->tt_index].table, VC1_SUBBLKPAT_VLC_BITS, 1) + 1); + subblkpat = ~(get_vlc2(gb, ff_vc1_subblkpat_vlc[v->tt_index].table, VC1_SUBBLKPAT_VLC_BITS, 1) + 1); } if((ttblk != TT_8X8 && ttblk != TT_4X4) && (v->ttmbf || (ttmb != -1 && (ttmb & 8) && !first_block))) { subblkpat = decode012(gb); @@ -3248,7 +3005,7 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan if(ttblk == TT_8X4_TOP || ttblk == TT_8X4_BOTTOM) ttblk = TT_8X4; if(ttblk == TT_4X8_RIGHT || ttblk == TT_4X8_LEFT) ttblk = TT_4X8; } - scale = 2 * mquant + v->halfpq; + scale = 2 * mquant + ((v->pq == mquant) ? v->halfpq : 0); // convert transforms like 8X4_TOP to generic TT and SUBBLKPAT if(ttblk == TT_8X4_TOP || ttblk == TT_8X4_BOTTOM) { @@ -3261,6 +3018,7 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan } switch(ttblk) { case TT_8X8: + pat = 0xF; i = 0; last = 0; while (!last) { @@ -3268,14 +3026,22 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan i += skip; if(i > 63) break; - idx = vc1_simple_progressive_8x8_zz[i++]; + idx = wmv1_scantable[0][i++]; block[idx] = value * scale; if(!v->pquantizer) block[idx] += (block[idx] < 0) ? -mquant : mquant; } - s->dsp.vc1_inv_trans_8x8(block); + if(!skip_block){ + s->dsp.vc1_inv_trans_8x8(block); + s->dsp.add_pixels_clamped(block, dst, linesize); + if(apply_filter && cbp_top & 0xC) + vc1_loop_filter(dst, 1, linesize, 8, mquant); + if(apply_filter && cbp_left & 0xA) + vc1_loop_filter(dst, linesize, 1, 8, mquant); + } break; case TT_4X4: + pat = ~subblkpat & 0xF; for(j = 0; j < 4; j++) { last = subblkpat & (1 << (3 - j)); i = 0; @@ -3285,16 +3051,22 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan i += skip; if(i > 15) break; - idx = vc1_simple_progressive_4x4_zz[i++]; + idx = ff_vc1_simple_progressive_4x4_zz[i++]; block[idx + off] = value * scale; if(!v->pquantizer) block[idx + off] += (block[idx + off] < 0) ? -mquant : mquant; } - if(!(subblkpat & (1 << (3 - j)))) - s->dsp.vc1_inv_trans_4x4(block, j); + if(!(subblkpat & (1 << (3 - j))) && !skip_block){ + s->dsp.vc1_inv_trans_4x4(dst + (j&1)*4 + (j&2)*2*linesize, linesize, block + off); + if(apply_filter && (j&2 ? pat & (1<<(j-2)) : (cbp_top & (1 << (j + 2))))) + vc1_loop_filter(dst + (j&1)*4 + (j&2)*2*linesize, 1, linesize, 4, mquant); + if(apply_filter && (j&1 ? pat & (1<<(j-1)) : (cbp_left & (1 << (j + 1))))) + vc1_loop_filter(dst + (j&1)*4 + (j&2)*2*linesize, linesize, 1, 4, mquant); + } } break; case TT_8X4: + pat = ~((subblkpat & 2)*6 + (subblkpat & 1)*3) & 0xF; for(j = 0; j < 2; j++) { last = subblkpat & (1 << (1 - j)); i = 0; @@ -3304,19 +3076,22 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan i += skip; if(i > 31) break; - if(v->profile < PROFILE_ADVANCED) - idx = vc1_simple_progressive_8x4_zz[i++]; - else - idx = vc1_adv_progressive_8x4_zz[i++]; - block[idx + off] = value * scale; + idx = v->zz_8x4[i++]+off; + block[idx] = value * scale; if(!v->pquantizer) - block[idx + off] += (block[idx + off] < 0) ? -mquant : mquant; + block[idx] += (block[idx] < 0) ? -mquant : mquant; + } + if(!(subblkpat & (1 << (1 - j))) && !skip_block){ + s->dsp.vc1_inv_trans_8x4(dst + j*4*linesize, linesize, block + off); + if(apply_filter && j ? pat & 0x3 : (cbp_top & 0xC)) + vc1_loop_filter(dst + j*4*linesize, 1, linesize, 8, mquant); + if(apply_filter && cbp_left & (2 << j)) + vc1_loop_filter(dst + j*4*linesize, linesize, 1, 4, mquant); } - if(!(subblkpat & (1 << (1 - j)))) - s->dsp.vc1_inv_trans_8x4(block, j); } break; case TT_4X8: + pat = ~(subblkpat*5) & 0xF; for(j = 0; j < 2; j++) { last = subblkpat & (1 << (1 - j)); i = 0; @@ -3326,22 +3101,28 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan i += skip; if(i > 31) break; - if(v->profile < PROFILE_ADVANCED) - idx = vc1_simple_progressive_4x8_zz[i++]; - else - idx = vc1_adv_progressive_4x8_zz[i++]; - block[idx + off] = value * scale; + idx = v->zz_4x8[i++]+off; + block[idx] = value * scale; if(!v->pquantizer) - block[idx + off] += (block[idx + off] < 0) ? -mquant : mquant; + block[idx] += (block[idx] < 0) ? -mquant : mquant; + } + if(!(subblkpat & (1 << (1 - j))) && !skip_block){ + s->dsp.vc1_inv_trans_4x8(dst + j*4, linesize, block + off); + if(apply_filter && cbp_top & (2 << j)) + vc1_loop_filter(dst + j*4, 1, linesize, 4, mquant); + if(apply_filter && j ? pat & 0x5 : (cbp_left & 0xA)) + vc1_loop_filter(dst + j*4, linesize, 1, 8, mquant); } - if(!(subblkpat & (1 << (1 - j)))) - s->dsp.vc1_inv_trans_4x8(block, j); } break; } - return 0; + return pat; } +/** @} */ // Macroblock group + +static const int size_table [6] = { 0, 2, 3, 4, 5, 8 }; +static const int offset_table[6] = { 0, 1, 3, 7, 15, 31 }; /** Decode one P-frame MB (in Simple/Main profile) */ @@ -3354,17 +3135,16 @@ static int vc1_decode_p_mb(VC1Context *v) int cbp; /* cbp decoding stuff */ int mqdiff, mquant; /* MB quantization */ int ttmb = v->ttfrm; /* MB Transform type */ - int status; - static const int size_table[6] = { 0, 2, 3, 4, 5, 8 }, - offset_table[6] = { 0, 1, 3, 7, 15, 31 }; int mb_has_coeffs = 1; /* last_flag */ int dmv_x, dmv_y; /* Differential MV components */ - int index, index1; /* LUT indices */ + int index, index1; /* LUT indexes */ int val, sign; /* temp values */ int first_block = 1; int dst_idx, off; int skipped, fourmv; + int block_cbp = 0, pat; + int apply_loop_filter; mquant = v->pq; /* Loosy initialization */ @@ -3379,6 +3159,7 @@ static int vc1_decode_p_mb(VC1Context *v) s->dsp.clear_blocks(s->block[0]); + apply_loop_filter = s->loop_filter && !(s->avctx->skip_loop_filter >= AVDISCARD_NONKEY); if (!fourmv) /* 1MV mode */ { if (!skipped) @@ -3396,12 +3177,12 @@ static int vc1_decode_p_mb(VC1Context *v) if (s->mb_intra && !mb_has_coeffs) { GET_MQUANT(); - s->ac_pred = get_bits(gb, 1); + s->ac_pred = get_bits1(gb); cbp = 0; } else if (mb_has_coeffs) { - if (s->mb_intra) s->ac_pred = get_bits(gb, 1); + if (s->mb_intra) s->ac_pred = get_bits1(gb); cbp = get_vlc2(&v->s.gb, v->cbpcy_vlc->table, VC1_CBPCY_P_VLC_BITS, 2); GET_MQUANT(); } @@ -3413,7 +3194,7 @@ static int vc1_decode_p_mb(VC1Context *v) s->current_picture.qscale_table[mb_pos] = mquant; if (!v->ttmbf && !s->mb_intra && mb_has_coeffs) - ttmb = get_vlc2(gb, vc1_ttmb_vlc[v->tt_index].table, + ttmb = get_vlc2(gb, ff_vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2); if(!s->mb_intra) vc1_mc_1mv(v, 0); dst_idx = 0; @@ -3436,21 +3217,48 @@ static int vc1_decode_p_mb(VC1Context *v) if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; s->dsp.vc1_inv_trans_8x8(s->block[i]); if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1; - for(j = 0; j < 64; j++) s->block[i][j] += 128; - if(!v->res_fasttx && v->res_x8) for(j = 0; j < 64; j++) s->block[i][j] += 16; - s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); + s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); if(v->pq >= 9 && v->overlap) { if(v->c_avail) s->dsp.vc1_h_overlap(s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); if(v->a_avail) s->dsp.vc1_v_overlap(s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); } + if(apply_loop_filter && s->mb_x && s->mb_x != (s->mb_width - 1) && s->mb_y && s->mb_y != (s->mb_height - 1)){ + int left_cbp, top_cbp; + if(i & 4){ + left_cbp = v->cbp[s->mb_x - 1] >> (i * 4); + top_cbp = v->cbp[s->mb_x - s->mb_stride] >> (i * 4); + }else{ + left_cbp = (i & 1) ? (cbp >> ((i-1)*4)) : (v->cbp[s->mb_x - 1] >> ((i+1)*4)); + top_cbp = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4)); + } + if(left_cbp & 0xC) + vc1_loop_filter(s->dest[dst_idx] + off, 1, i & 4 ? s->uvlinesize : s->linesize, 8, mquant); + if(top_cbp & 0xA) + vc1_loop_filter(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, 1, 8, mquant); + } + block_cbp |= 0xF << (i << 2); } else if(val) { - vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block); + int left_cbp = 0, top_cbp = 0, filter = 0; + if(apply_loop_filter && s->mb_x && s->mb_x != (s->mb_width - 1) && s->mb_y && s->mb_y != (s->mb_height - 1)){ + filter = 1; + if(i & 4){ + left_cbp = v->cbp[s->mb_x - 1] >> (i * 4); + top_cbp = v->cbp[s->mb_x - s->mb_stride] >> (i * 4); + }else{ + left_cbp = (i & 1) ? (cbp >> ((i-1)*4)) : (v->cbp[s->mb_x - 1] >> ((i+1)*4)); + top_cbp = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4)); + } + if(left_cbp & 0xC) + vc1_loop_filter(s->dest[dst_idx] + off, 1, i & 4 ? s->uvlinesize : s->linesize, 8, mquant); + if(top_cbp & 0xA) + vc1_loop_filter(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, 1, 8, mquant); + } + pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize, (i&4) && (s->flags & CODEC_FLAG_GRAY), filter, left_cbp, top_cbp); + block_cbp |= pat << (i << 2); if(!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; - if((i<4) || !(s->flags & CODEC_FLAG_GRAY)) - s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); } } } @@ -3518,11 +3326,11 @@ static int vc1_decode_p_mb(VC1Context *v) break; } } - if(intrapred)s->ac_pred = get_bits(gb, 1); + if(intrapred)s->ac_pred = get_bits1(gb); else s->ac_pred = 0; } if (!v->ttmbf && coded_inter) - ttmb = get_vlc2(gb, vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2); + ttmb = get_vlc2(gb, ff_vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2); for (i=0; i<6; i++) { dst_idx += i >> 2; @@ -3540,24 +3348,51 @@ static int vc1_decode_p_mb(VC1Context *v) if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; s->dsp.vc1_inv_trans_8x8(s->block[i]); if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1; - for(j = 0; j < 64; j++) s->block[i][j] += 128; - if(!v->res_fasttx && v->res_x8) for(j = 0; j < 64; j++) s->block[i][j] += 16; - s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); + s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); if(v->pq >= 9 && v->overlap) { if(v->c_avail) s->dsp.vc1_h_overlap(s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); if(v->a_avail) s->dsp.vc1_v_overlap(s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); } + if(v->s.loop_filter && s->mb_x && s->mb_x != (s->mb_width - 1) && s->mb_y && s->mb_y != (s->mb_height - 1)){ + int left_cbp, top_cbp; + if(i & 4){ + left_cbp = v->cbp[s->mb_x - 1] >> (i * 4); + top_cbp = v->cbp[s->mb_x - s->mb_stride] >> (i * 4); + }else{ + left_cbp = (i & 1) ? (cbp >> ((i-1)*4)) : (v->cbp[s->mb_x - 1] >> ((i+1)*4)); + top_cbp = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4)); + } + if(left_cbp & 0xC) + vc1_loop_filter(s->dest[dst_idx] + off, 1, i & 4 ? s->uvlinesize : s->linesize, 8, mquant); + if(top_cbp & 0xA) + vc1_loop_filter(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, 1, 8, mquant); + } + block_cbp |= 0xF << (i << 2); } else if(is_coded[i]) { - status = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block); + int left_cbp = 0, top_cbp = 0, filter = 0; + if(v->s.loop_filter && s->mb_x && s->mb_x != (s->mb_width - 1) && s->mb_y && s->mb_y != (s->mb_height - 1)){ + filter = 1; + if(i & 4){ + left_cbp = v->cbp[s->mb_x - 1] >> (i * 4); + top_cbp = v->cbp[s->mb_x - s->mb_stride] >> (i * 4); + }else{ + left_cbp = (i & 1) ? (cbp >> ((i-1)*4)) : (v->cbp[s->mb_x - 1] >> ((i+1)*4)); + top_cbp = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4)); + } + if(left_cbp & 0xC) + vc1_loop_filter(s->dest[dst_idx] + off, 1, i & 4 ? s->uvlinesize : s->linesize, 8, mquant); + if(top_cbp & 0xA) + vc1_loop_filter(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, 1, 8, mquant); + } + pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize, (i&4) && (s->flags & CODEC_FLAG_GRAY), filter, left_cbp, top_cbp); + block_cbp |= pat << (i << 2); if(!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; - if((i<4) || !(s->flags & CODEC_FLAG_GRAY)) - s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); } } - return status; + return 0; } else //Skipped MB { @@ -3577,6 +3412,7 @@ static int vc1_decode_p_mb(VC1Context *v) return 0; } } + v->cbp[s->mb_x] = block_cbp; /* Should never happen */ return -1; @@ -3593,11 +3429,8 @@ static void vc1_decode_b_mb(VC1Context *v) int cbp = 0; /* cbp decoding stuff */ int mqdiff, mquant; /* MB quantization */ int ttmb = v->ttfrm; /* MB Transform type */ - - static const int size_table[6] = { 0, 2, 3, 4, 5, 8 }, - offset_table[6] = { 0, 1, 3, 7, 15, 31 }; int mb_has_coeffs = 0; /* last_flag */ - int index, index1; /* LUT indices */ + int index, index1; /* LUT indexes */ int val, sign; /* temp values */ int first_block = 1; int dst_idx, off; @@ -3662,7 +3495,7 @@ static void vc1_decode_b_mb(VC1Context *v) mb_has_coeffs = 0; s->current_picture.qscale_table[mb_pos] = mquant; if(!v->ttmbf) - ttmb = get_vlc2(gb, vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2); + ttmb = get_vlc2(gb, ff_vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2); dmv_x[0] = dmv_y[0] = dmv_x[1] = dmv_y[1] = 0; vc1_pred_b_mv(v, dmv_x, dmv_y, direct, bmvtype); vc1_b_mc(v, dmv_x, dmv_y, direct, bmvtype); @@ -3699,7 +3532,7 @@ static void vc1_decode_b_mb(VC1Context *v) GET_MQUANT(); s->current_picture.qscale_table[mb_pos] = mquant; if(!v->ttmbf && !s->mb_intra && mb_has_coeffs) - ttmb = get_vlc2(gb, vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2); + ttmb = get_vlc2(gb, ff_vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2); } } dst_idx = 0; @@ -3722,14 +3555,11 @@ static void vc1_decode_b_mb(VC1Context *v) if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue; s->dsp.vc1_inv_trans_8x8(s->block[i]); if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1; - for(j = 0; j < 64; j++) s->block[i][j] += 128; - s->dsp.put_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); + s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, s->linesize >> ((i & 4) >> 2)); } else if(val) { - vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block); + vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize, (i&4) && (s->flags & CODEC_FLAG_GRAY), 0, 0, 0); if(!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; - if((i<4) || !(s->flags & CODEC_FLAG_GRAY)) - s->dsp.add_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize); } } } @@ -3777,7 +3607,6 @@ static void vc1_decode_i_blocks(VC1Context *v) s->mb_x = s->mb_y = 0; s->mb_intra = 1; s->first_slice_line = 1; - ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); for(s->mb_y = 0; s->mb_y < s->mb_height; s->mb_y++) { for(s->mb_x = 0; s->mb_x < s->mb_width; s->mb_x++) { ff_init_block_index(s); @@ -3791,7 +3620,7 @@ static void vc1_decode_i_blocks(VC1Context *v) // do actual MB decoding and displaying cbp = get_vlc2(&v->s.gb, ff_msmp4_mb_i_vlc.table, MB_INTRA_VLC_BITS, 2); - v->s.ac_pred = get_bits(&v->s.gb, 1); + v->s.ac_pred = get_bits1(&v->s.gb); for(k = 0; k < 6; k++) { val = ((cbp >> (5 - k)) & 1); @@ -3834,8 +3663,10 @@ static void vc1_decode_i_blocks(VC1Context *v) s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize); s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize); } + if(v->s.loop_filter) vc1_loop_filter_iblk(s, s->current_picture.qscale_table[mb_pos]); if(get_bits_count(&s->gb) > v->bits) { + ff_er_add_slice(s, 0, 0, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END)); av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i\n", get_bits_count(&s->gb), v->bits); return; } @@ -3843,6 +3674,7 @@ static void vc1_decode_i_blocks(VC1Context *v) ff_draw_horiz_band(s, s->mb_y * 16, 16); s->first_slice_line = 0; } + ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); } /** Decode blocks of I-frame for advanced profile @@ -3888,7 +3720,6 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) s->mb_x = s->mb_y = 0; s->mb_intra = 1; s->first_slice_line = 1; - ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); for(s->mb_y = 0; s->mb_y < s->mb_height; s->mb_y++) { for(s->mb_x = 0; s->mb_x < s->mb_width; s->mb_x++) { ff_init_block_index(s); @@ -3902,13 +3733,13 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) // do actual MB decoding and displaying cbp = get_vlc2(&v->s.gb, ff_msmp4_mb_i_vlc.table, MB_INTRA_VLC_BITS, 2); if(v->acpred_is_raw) - v->s.ac_pred = get_bits(&v->s.gb, 1); + v->s.ac_pred = get_bits1(&v->s.gb); else v->s.ac_pred = v->acpred_plane[mb_pos]; if(v->condover == CONDOVER_SELECT) { if(v->overflg_is_raw) - overlap = get_bits(&v->s.gb, 1); + overlap = get_bits1(&v->s.gb); else overlap = v->over_flags_plane[mb_pos]; } else @@ -3963,8 +3794,10 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize); s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize); } + if(v->s.loop_filter) vc1_loop_filter_iblk(s, s->current_picture.qscale_table[mb_pos]); if(get_bits_count(&s->gb) > v->bits) { + ff_er_add_slice(s, 0, 0, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END)); av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i\n", get_bits_count(&s->gb), v->bits); return; } @@ -3972,6 +3805,7 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) ff_draw_horiz_band(s, s->mb_y * 16, 16); s->first_slice_line = 0; } + ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); } static void vc1_decode_p_blocks(VC1Context *v) @@ -4003,8 +3837,8 @@ static void vc1_decode_p_blocks(VC1Context *v) break; } - ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); s->first_slice_line = 1; + memset(v->cbp_base, 0, sizeof(v->cbp_base[0])*2*s->mb_stride); for(s->mb_y = 0; s->mb_y < s->mb_height; s->mb_y++) { for(s->mb_x = 0; s->mb_x < s->mb_width; s->mb_x++) { ff_init_block_index(s); @@ -4013,13 +3847,16 @@ static void vc1_decode_p_blocks(VC1Context *v) vc1_decode_p_mb(v); if(get_bits_count(&s->gb) > v->bits || get_bits_count(&s->gb) < 0) { + ff_er_add_slice(s, 0, 0, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END)); av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i at %ix%i\n", get_bits_count(&s->gb), v->bits,s->mb_x,s->mb_y); return; } } + memmove(v->cbp_base, v->cbp, sizeof(v->cbp_base[0])*s->mb_stride); ff_draw_horiz_band(s, s->mb_y * 16, 16); s->first_slice_line = 0; } + ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); } static void vc1_decode_b_blocks(VC1Context *v) @@ -4051,7 +3888,6 @@ static void vc1_decode_b_blocks(VC1Context *v) break; } - ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); s->first_slice_line = 1; for(s->mb_y = 0; s->mb_y < s->mb_height; s->mb_y++) { for(s->mb_x = 0; s->mb_x < s->mb_width; s->mb_x++) { @@ -4061,13 +3897,16 @@ static void vc1_decode_b_blocks(VC1Context *v) vc1_decode_b_mb(v); if(get_bits_count(&s->gb) > v->bits || get_bits_count(&s->gb) < 0) { + ff_er_add_slice(s, 0, 0, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END)); av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i at %ix%i\n", get_bits_count(&s->gb), v->bits,s->mb_x,s->mb_y); return; } + if(v->s.loop_filter) vc1_loop_filter_iblk(s, s->current_picture.qscale_table[s->mb_x + s->mb_y *s->mb_stride]); } ff_draw_horiz_band(s, s->mb_y * 16, 16); s->first_slice_line = 0; } + ff_er_add_slice(s, 0, 0, s->mb_width - 1, s->mb_height - 1, (AC_END|DC_END|MV_END)); } static void vc1_decode_skip_blocks(VC1Context *v) @@ -4086,45 +3925,47 @@ static void vc1_decode_skip_blocks(VC1Context *v) ff_draw_horiz_band(s, s->mb_y * 16, 16); s->first_slice_line = 0; } - s->pict_type = P_TYPE; + s->pict_type = FF_P_TYPE; } static void vc1_decode_blocks(VC1Context *v) { v->s.esc3_level_length = 0; + if(v->x8_type){ + ff_intrax8_decode_picture(&v->x8, 2*v->pq+v->halfpq, v->pq*(!v->pquantizer) ); + }else{ - switch(v->s.pict_type) { - case I_TYPE: - if(v->profile == PROFILE_ADVANCED) - vc1_decode_i_blocks_adv(v); - else - vc1_decode_i_blocks(v); - break; - case P_TYPE: - if(v->p_frame_skipped) - vc1_decode_skip_blocks(v); - else - vc1_decode_p_blocks(v); - break; - case B_TYPE: - if(v->bi_type){ + switch(v->s.pict_type) { + case FF_I_TYPE: if(v->profile == PROFILE_ADVANCED) vc1_decode_i_blocks_adv(v); else vc1_decode_i_blocks(v); - }else - vc1_decode_b_blocks(v); - break; + break; + case FF_P_TYPE: + if(v->p_frame_skipped) + vc1_decode_skip_blocks(v); + else + vc1_decode_p_blocks(v); + break; + case FF_B_TYPE: + if(v->bi_type){ + if(v->profile == PROFILE_ADVANCED) + vc1_decode_i_blocks_adv(v); + else + vc1_decode_i_blocks(v); + }else + vc1_decode_b_blocks(v); + break; + } } } -#define IS_MARKER(x) (((x) & ~0xFF) == VC1_CODE_RES0) - /** Find VC-1 marker in buffer * @return position where next marker starts or end of buffer if no marker found */ -static av_always_inline uint8_t* find_next_marker(uint8_t *src, uint8_t *end) +static av_always_inline const uint8_t* find_next_marker(const uint8_t *src, const uint8_t *end) { uint32_t mrk = 0xFFFFFFFF; @@ -4137,7 +3978,7 @@ static av_always_inline uint8_t* find_next_marker(uint8_t *src, uint8_t *end) return end; } -static av_always_inline int vc1_unescape_buffer(uint8_t *src, int size, uint8_t *dst) +static av_always_inline int vc1_unescape_buffer(const uint8_t *src, int size, uint8_t *dst) { int dsize = 0, i; @@ -4160,7 +4001,7 @@ static av_always_inline int vc1_unescape_buffer(uint8_t *src, int size, uint8_t * @todo TODO: Handle VC-1 IDUs (Transport level?) * @todo TODO: Decypher remaining bits in extra_data */ -static int vc1_decode_init(AVCodecContext *avctx) +static av_cold int vc1_decode_init(AVCodecContext *avctx) { VC1Context *v = avctx->priv_data; MpegEncContext *s = &v->s; @@ -4168,13 +4009,18 @@ static int vc1_decode_init(AVCodecContext *avctx) if (!avctx->extradata_size || !avctx->extradata) return -1; if (!(avctx->flags & CODEC_FLAG_GRAY)) - avctx->pix_fmt = PIX_FMT_YUV420P; + avctx->pix_fmt = avctx->get_format(avctx, avctx->codec->pix_fmts); else avctx->pix_fmt = PIX_FMT_GRAY8; + avctx->hwaccel = ff_find_hwaccel(avctx->codec->id, avctx->pix_fmt); v->s.avctx = avctx; avctx->flags |= CODEC_FLAG_EMU_EDGE; v->s.flags |= CODEC_FLAG_EMU_EDGE; + if(avctx->idct_algo==FF_IDCT_AUTO){ + avctx->idct_algo=FF_IDCT_WMV2; + } + if(ff_h263_decode_init(avctx) < 0) return -1; if (vc1_init_common(v) < 0) return -1; @@ -4206,10 +4052,12 @@ static int vc1_decode_init(AVCodecContext *avctx) av_log(avctx, AV_LOG_INFO, "Read %i bits in overflow\n", -count); } } else { // VC1/WVC1 - uint8_t *start = avctx->extradata, *end = avctx->extradata + avctx->extradata_size; - uint8_t *next; int size, buf2_size; + const uint8_t *start = avctx->extradata; + uint8_t *end = avctx->extradata + avctx->extradata_size; + const uint8_t *next; + int size, buf2_size; uint8_t *buf2 = NULL; - int seq_inited = 0, ep_inited = 0; + int seq_initialized = 0, ep_initialized = 0; if(avctx->extradata_size < 16) { av_log(avctx, AV_LOG_ERROR, "Extradata size too small: %i\n", avctx->extradata_size); @@ -4231,19 +4079,19 @@ static int vc1_decode_init(AVCodecContext *avctx) av_free(buf2); return -1; } - seq_inited = 1; + seq_initialized = 1; break; case VC1_CODE_ENTRYPOINT: if(decode_entry_point(avctx, &gb) < 0){ av_free(buf2); return -1; } - ep_inited = 1; + ep_initialized = 1; break; } } av_free(buf2); - if(!seq_inited || !ep_inited){ + if(!seq_initialized || !ep_initialized){ av_log(avctx, AV_LOG_ERROR, "Incomplete extradata\n"); return -1; } @@ -4260,6 +4108,9 @@ static int vc1_decode_init(AVCodecContext *avctx) v->acpred_plane = av_malloc(s->mb_stride * s->mb_height); v->over_flags_plane = av_malloc(s->mb_stride * s->mb_height); + v->cbp_base = av_malloc(sizeof(v->cbp_base[0]) * 2 * s->mb_stride); + v->cbp = v->cbp_base + s->mb_stride; + /* allocate block type info in that way so it could be used with s->block_index[] */ v->mb_type_base = av_malloc(s->b8_stride * (s->mb_height * 2 + 1) + s->mb_stride * (s->mb_height + 1) * 2); v->mb_type[0] = v->mb_type_base + s->b8_stride + 1; @@ -4275,6 +4126,7 @@ static int vc1_decode_init(AVCodecContext *avctx) // return -1; } + ff_intrax8_common_init(&v->x8,s); return 0; } @@ -4284,12 +4136,15 @@ static int vc1_decode_init(AVCodecContext *avctx) */ static int vc1_decode_frame(AVCodecContext *avctx, void *data, int *data_size, - uint8_t *buf, int buf_size) + AVPacket *avpkt) { + const uint8_t *buf = avpkt->data; + int buf_size = avpkt->size; VC1Context *v = avctx->priv_data; MpegEncContext *s = &v->s; AVFrame *pict = data; uint8_t *buf2 = NULL; + const uint8_t *buf_start = buf; /* no supplementary picture */ if (buf_size == 0) { @@ -4304,19 +4159,27 @@ static int vc1_decode_frame(AVCodecContext *avctx, return 0; } - //we need to set current_picture_ptr before reading the header, otherwise we cant store anyting im there + /* We need to set current_picture_ptr before reading the header, + * otherwise we cannot store anything in there. */ if(s->current_picture_ptr==NULL || s->current_picture_ptr->data[0]){ int i= ff_find_unused_picture(s, 0); s->current_picture_ptr= &s->picture[i]; } + if (s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU){ + if (v->profile < PROFILE_ADVANCED) + avctx->pix_fmt = PIX_FMT_VDPAU_WMV3; + else + avctx->pix_fmt = PIX_FMT_VDPAU_VC1; + } + //for advanced profile we may need to parse and unescape data if (avctx->codec_id == CODEC_ID_VC1) { int buf_size2 = 0; buf2 = av_mallocz(buf_size + FF_INPUT_BUFFER_PADDING_SIZE); if(IS_MARKER(AV_RB32(buf))){ /* frame starts with marker and needs to be parsed */ - uint8_t *dst = buf2, *start, *end, *next; + const uint8_t *start, *end, *next; int size; next = buf; @@ -4326,6 +4189,9 @@ static int vc1_decode_frame(AVCodecContext *avctx, if(size <= 0) continue; switch(AV_RB32(start)){ case VC1_CODE_FRAME: + if (avctx->hwaccel || + s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU) + buf_start = start; buf_size2 = vc1_unescape_buffer(start + 4, size, buf2); break; case VC1_CODE_ENTRYPOINT: /* it should be before frame data */ @@ -4340,11 +4206,12 @@ static int vc1_decode_frame(AVCodecContext *avctx, } } }else if(v->interlace && ((buf[0] & 0xC0) == 0xC0)){ /* WVC1 interlaced stores both fields divided by marker */ - uint8_t *divider; + const uint8_t *divider; divider = find_next_marker(buf, buf + buf_size); if((divider == (buf + buf_size)) || AV_RB32(divider) != VC1_CODE_FIELD){ av_log(avctx, AV_LOG_ERROR, "Error in WVC1 interlaced frame\n"); + av_free(buf2); return -1; } @@ -4370,24 +4237,24 @@ static int vc1_decode_frame(AVCodecContext *avctx, } } - if(s->pict_type != I_TYPE && !v->res_rtm_flag){ + if(s->pict_type != FF_I_TYPE && !v->res_rtm_flag){ av_free(buf2); return -1; } // for hurry_up==5 s->current_picture.pict_type= s->pict_type; - s->current_picture.key_frame= s->pict_type == I_TYPE; + s->current_picture.key_frame= s->pict_type == FF_I_TYPE; /* skip B-frames if we don't have reference frames */ - if(s->last_picture_ptr==NULL && (s->pict_type==B_TYPE || s->dropable)){ + if(s->last_picture_ptr==NULL && (s->pict_type==FF_B_TYPE || s->dropable)){ av_free(buf2); return -1;//buf_size; } /* skip b frames if we are in a hurry */ - if(avctx->hurry_up && s->pict_type==B_TYPE) return -1;//buf_size; - if( (avctx->skip_frame >= AVDISCARD_NONREF && s->pict_type==B_TYPE) - || (avctx->skip_frame >= AVDISCARD_NONKEY && s->pict_type!=I_TYPE) + if(avctx->hurry_up && s->pict_type==FF_B_TYPE) return -1;//buf_size; + if( (avctx->skip_frame >= AVDISCARD_NONREF && s->pict_type==FF_B_TYPE) + || (avctx->skip_frame >= AVDISCARD_NONKEY && s->pict_type!=FF_I_TYPE) || avctx->skip_frame >= AVDISCARD_ALL) { av_free(buf2); return buf_size; @@ -4399,7 +4266,7 @@ static int vc1_decode_frame(AVCodecContext *avctx, } if(s->next_p_frame_damaged){ - if(s->pict_type==B_TYPE) + if(s->pict_type==FF_B_TYPE) return buf_size; else s->next_p_frame_damaged=0; @@ -4410,20 +4277,35 @@ static int vc1_decode_frame(AVCodecContext *avctx, return -1; } - ff_er_frame_start(s); + s->me.qpel_put= s->dsp.put_qpel_pixels_tab; + s->me.qpel_avg= s->dsp.avg_qpel_pixels_tab; - v->bits = buf_size * 8; - vc1_decode_blocks(v); + if ((CONFIG_VC1_VDPAU_DECODER || CONFIG_WMV3_VDPAU_DECODER) + &&s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU) + ff_vdpau_vc1_decode_picture(s, buf_start, (buf + buf_size) - buf_start); + else if (avctx->hwaccel) { + if (avctx->hwaccel->start_frame(avctx, buf, buf_size) < 0) + return -1; + if (avctx->hwaccel->decode_slice(avctx, buf_start, (buf + buf_size) - buf_start) < 0) + return -1; + if (avctx->hwaccel->end_frame(avctx) < 0) + return -1; + } else { + ff_er_frame_start(s); + + v->bits = buf_size * 8; + vc1_decode_blocks(v); //av_log(s->avctx, AV_LOG_INFO, "Consumed %i/%i bits\n", get_bits_count(&s->gb), buf_size*8); // if(get_bits_count(&s->gb) > buf_size * 8) // return -1; - ff_er_frame_end(s); + ff_er_frame_end(s); + } MPV_frame_end(s); assert(s->current_picture.pict_type == s->current_picture_ptr->pict_type); assert(s->current_picture.pict_type == s->pict_type); - if (s->pict_type == B_TYPE || s->low_delay) { + if (s->pict_type == FF_B_TYPE || s->low_delay) { *pict= *(AVFrame*)s->current_picture_ptr; } else if (s->last_picture_ptr != NULL) { *pict= *(AVFrame*)s->last_picture_ptr; @@ -4435,7 +4317,7 @@ assert(s->current_picture.pict_type == s->pict_type); } /* Return the Picture timestamp as the frame number */ - /* we substract 1 because it is added on utils.c */ + /* we subtract 1 because it is added on utils.c */ avctx->frame_number = s->picture_number - 1; av_free(buf2); @@ -4446,7 +4328,7 @@ assert(s->current_picture.pict_type == s->pict_type); /** Close a VC1/WMV3 decoder * @warning Initial try at using MpegEncContext stuff */ -static int vc1_decode_end(AVCodecContext *avctx) +static av_cold int vc1_decode_end(AVCodecContext *avctx) { VC1Context *v = avctx->priv_data; @@ -4458,6 +4340,8 @@ static int vc1_decode_end(AVCodecContext *avctx) av_freep(&v->acpred_plane); av_freep(&v->over_flags_plane); av_freep(&v->mb_type_base); + av_freep(&v->cbp_base); + ff_intrax8_common_end(&v->x8); return 0; } @@ -4472,7 +4356,9 @@ AVCodec vc1_decoder = { vc1_decode_end, vc1_decode_frame, CODEC_CAP_DELAY, - NULL + NULL, + .long_name = NULL_IF_CONFIG_SMALL("SMPTE VC-1"), + .pix_fmts = ff_hwaccel_pixfmt_list_420 }; AVCodec wmv3_decoder = { @@ -4485,96 +4371,41 @@ AVCodec wmv3_decoder = { vc1_decode_end, vc1_decode_frame, CODEC_CAP_DELAY, - NULL + NULL, + .long_name = NULL_IF_CONFIG_SMALL("Windows Media Video 9"), + .pix_fmts = ff_hwaccel_pixfmt_list_420 }; -#ifdef CONFIG_VC1_PARSER -/** - * finds the end of the current frame in the bitstream. - * @return the position of the first byte of the next frame, or -1 - */ -static int vc1_find_frame_end(ParseContext *pc, const uint8_t *buf, - int buf_size) { - int pic_found, i; - uint32_t state; - - pic_found= pc->frame_start_found; - state= pc->state; - - i=0; - if(!pic_found){ - for(i=0; iframe_start_found=0; - pc->state=-1; - return i-3; - } - } - } - pc->frame_start_found= pic_found; - pc->state= state; - return END_NOT_FOUND; -} - -static int vc1_parse(AVCodecParserContext *s, - AVCodecContext *avctx, - uint8_t **poutbuf, int *poutbuf_size, - const uint8_t *buf, int buf_size) -{ - ParseContext *pc = s->priv_data; - int next; - - if(s->flags & PARSER_FLAG_COMPLETE_FRAMES){ - next= buf_size; - }else{ - next= vc1_find_frame_end(pc, buf, buf_size); - - if (ff_combine_frame(pc, next, (uint8_t **)&buf, &buf_size) < 0) { - *poutbuf = NULL; - *poutbuf_size = 0; - return buf_size; - } - } - *poutbuf = (uint8_t *)buf; - *poutbuf_size = buf_size; - return next; -} - -int vc1_split(AVCodecContext *avctx, - const uint8_t *buf, int buf_size) -{ - int i; - uint32_t state= -1; - - for(i=0; i