X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fmpegvideo.h;h=5313edc62e04a82c551e254a05c58bdc5f9cf2d5;hb=ee3035f31e97ac095a9504c032c7320dcd9cfe4b;hp=998371802330576e45106e3cb70d5edced7e2186;hpb=160d679c075e9f575f96ea2c2c000545a49916a0;p=ffmpeg diff --git a/libavcodec/mpegvideo.h b/libavcodec/mpegvideo.h index 99837180233..5313edc62e0 100644 --- a/libavcodec/mpegvideo.h +++ b/libavcodec/mpegvideo.h @@ -3,31 +3,35 @@ * Copyright (c) 2000, 2001, 2002 Fabrice Bellard. * Copyright (c) 2002-2004 Michael Niedermayer * - * This library is free software; you can redistribute it and/or + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either - * version 2 of the License, or (at your option) any later version. + * version 2.1 of the License, or (at your option) any later version. * - * This library is distributed in the hope that it will be useful, + * FFmpeg is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public - * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ /** * @file mpegvideo.h * mpegvideo header. */ - + #ifndef AVCODEC_MPEGVIDEO_H #define AVCODEC_MPEGVIDEO_H #include "dsputil.h" #include "bitstream.h" +#include "ratecontrol.h" +#include "parser.h" #define FRAME_SKIPPED 100 ///< return value for header parsers if frame is not coded @@ -35,7 +39,7 @@ enum OutputFormat { FMT_MPEG1, FMT_H261, FMT_H263, - FMT_MJPEG, + FMT_MJPEG, FMT_H264, }; @@ -70,53 +74,17 @@ enum OutputFormat { #define MAX_MB_BYTES (30*16*16*3/8 + 120) -typedef struct Predictor{ - double coeff; - double count; - double decay; -} Predictor; - -typedef struct RateControlEntry{ - int pict_type; - float qscale; - int mv_bits; - int i_tex_bits; - int p_tex_bits; - int misc_bits; - uint64_t expected_bits; - int new_pict_type; - float new_qscale; - int mc_mb_var_sum; - int mb_var_sum; - int i_count; - int f_code; - int b_code; -}RateControlEntry; +#define INPLACE_OFFSET 16 -/** - * rate control context. - */ -typedef struct RateControlContext{ - FILE *stats_file; - int num_entries; ///< number of RateControlEntries - RateControlEntry *entry; - double buffer_index; ///< amount of bits in the video/audio buffer - Predictor pred[5]; - double short_term_qsum; ///< sum of recent qscales - double short_term_qcount; ///< count of recent qscales - double pass1_rc_eq_output_sum;///< sum of the output of the rc equation, this is used for normalization - double pass1_wanted_bits; ///< bits which should have been outputed by the pass1 code (including complexity init) - double last_qscale; - double last_qscale_for[5]; ///< last qscale for a specific pict type, used for max_diff & ipb factor stuff - int last_mc_mb_var_sum; - int last_mb_var_sum; - uint64_t i_cplx_sum[5]; - uint64_t p_cplx_sum[5]; - uint64_t mv_bits_sum[5]; - uint64_t qscale_sum[5]; - int frame_count[5]; - int last_non_b_pict_type; -}RateControlContext; +/* Start codes. */ +#define SEQ_END_CODE 0x000001b7 +#define SEQ_START_CODE 0x000001b3 +#define GOP_START_CODE 0x000001b8 +#define PICTURE_START_CODE 0x00000100 +#define SLICE_MIN_START_CODE 0x00000101 +#define SLICE_MAX_START_CODE 0x000001af +#define EXT_START_CODE 0x000001b5 +#define USER_START_CODE 0x000001b2 /** * Scantable. @@ -126,8 +94,8 @@ typedef struct ScanTable{ uint8_t permutated[64]; uint8_t raster_end[64]; #ifdef ARCH_POWERPC - /** Used by dct_quantise_alitvec to find last-non-zero */ - uint8_t __align8 inverse[64]; + /** Used by dct_quantise_alitvec to find last-non-zero */ + DECLARE_ALIGNED_8(uint8_t, inverse[64]); #endif } ScanTable; @@ -176,26 +144,15 @@ typedef struct Picture{ int ref_poc[2][16]; ///< h264 POCs of the frames used as reference int ref_count[2]; ///< number of entries in ref_poc - int mb_var_sum; ///< sum of MB variance for current frame - int mc_mb_var_sum; ///< motion compensated MB variance for current frame - uint16_t *mb_var; ///< Table for MB variances - uint16_t *mc_mb_var; ///< Table for motion compensated MB variances - uint8_t *mb_mean; ///< Table for MB luminance - int32_t *mb_cmp_score; ///< Table for MB cmp scores, for mb decission FIXME remove + int mb_var_sum; ///< sum of MB variance for current frame + int mc_mb_var_sum; ///< motion compensated MB variance for current frame + uint16_t *mb_var; ///< Table for MB variances + uint16_t *mc_mb_var; ///< Table for motion compensated MB variances + uint8_t *mb_mean; ///< Table for MB luminance + int32_t *mb_cmp_score; ///< Table for MB cmp scores, for mb decision FIXME remove int b_frame_score; /* */ } Picture; -typedef struct ParseContext{ - uint8_t *buffer; - int index; - int last_index; - int buffer_size; - uint32_t state; ///< contains the last few bytes in MSB order - int frame_start_found; - int overread; ///< the number of bytes which where irreversibly read from the next frame - int overread_index; ///< the index into ParseContext.buffer of the overreaded bytes -} ParseContext; - struct MpegEncContext; /** @@ -203,25 +160,29 @@ struct MpegEncContext; */ typedef struct MotionEstContext{ AVCodecContext *avctx; - int skip; ///< set if ME is skipped for the current MB - int co_located_mv[4][2]; ///< mv from last p frame for direct mode ME + int skip; ///< set if ME is skipped for the current MB + int co_located_mv[4][2]; ///< mv from last p frame for direct mode ME int direct_basis_mv[4][2]; - uint8_t *scratchpad; ///< data area for the me algo, so that the ME doesnt need to malloc/free + uint8_t *scratchpad; ///< data area for the me algo, so that the ME doesnt need to malloc/free uint8_t *best_mb; uint8_t *temp_mb[2]; uint8_t *temp; int best_bits; - uint32_t *map; ///< map to avoid duplicate evaluations - uint32_t *score_map; ///< map to store the scores - int map_generation; + uint32_t *map; ///< map to avoid duplicate evaluations + uint32_t *score_map; ///< map to store the scores + int map_generation; int pre_penalty_factor; - int penalty_factor; + int penalty_factor; /*!< an estimate of the bits required to + code a given mv value, e.g. (1,0) takes + more bits than (0,0). We have to + estimate whether any reduction in + residual is worth the extra bits. */ int sub_penalty_factor; int mb_penalty_factor; int flags; int sub_flags; int mb_flags; - int pre_pass; ///< = 1 for the pre pass + int pre_pass; ///< = 1 for the pre pass int dia_size; int xmin; int xmax; @@ -242,10 +203,10 @@ typedef struct MotionEstContext{ op_pixels_func (*hpel_avg)[4]; qpel_mc_func (*qpel_put)[16]; qpel_mc_func (*qpel_avg)[16]; - uint8_t (*mv_penalty)[MAX_MV*2+1]; ///< amount of bits needed to encode a MV + uint8_t (*mv_penalty)[MAX_MV*2+1]; ///< amount of bits needed to encode a MV uint8_t *current_mv_penalty; int (*sub_motion_search)(struct MpegEncContext * s, - int *mx_ptr, int *my_ptr, int dmin, + int *mx_ptr, int *my_ptr, int dmin, int src_index, int ref_index, int size, int h); }MotionEstContext; @@ -256,28 +217,30 @@ typedef struct MotionEstContext{ typedef struct MpegEncContext { struct AVCodecContext *avctx; /* the following parameters must be initialized before encoding */ - int width, height;///< picture size. must be a multiple of 16 + int width, height;///< picture size. must be a multiple of 16 int gop_size; - int intra_only; ///< if true, only intra pictures are generated - int bit_rate; ///< wanted bit rate - enum OutputFormat out_format; ///< output format - int h263_pred; ///< use mpeg4/h263 ac/dc predictions + int intra_only; ///< if true, only intra pictures are generated + int bit_rate; ///< wanted bit rate + enum OutputFormat out_format; ///< output format + int h263_pred; ///< use mpeg4/h263 ac/dc predictions /* the following codec id fields are deprecated in favor of codec_id */ - int h263_plus; ///< h263 plus headers + int h263_plus; ///< h263 plus headers int h263_msmpeg4; ///< generate MSMPEG4 compatible stream (deprecated, use msmpeg4_version instead) - int h263_flv; ///< use flv h263 header - + int h263_flv; ///< use flv h263 header + enum CodecID codec_id; /* see CODEC_ID_xxx */ - int fixed_qscale; ///< fixed qscale if non zero - int encoding; ///< true if we are encoding (vs decoding) - int flags; ///< AVCodecContext.flags (HQ, MV4, ...) + int fixed_qscale; ///< fixed qscale if non zero + int encoding; ///< true if we are encoding (vs decoding) + int flags; ///< AVCodecContext.flags (HQ, MV4, ...) int flags2; ///< AVCodecContext.flags2 - int max_b_frames; ///< max number of b-frames for encoding + int max_b_frames; ///< max number of b-frames for encoding int luma_elim_threshold; int chroma_elim_threshold; - int strict_std_compliance; ///< strictly follow the std (MPEG4, ...) - int workaround_bugs; ///< workaround bugs in encoders which cannot be detected automatically + int strict_std_compliance; ///< strictly follow the std (MPEG4, ...) + int workaround_bugs; ///< workaround bugs in encoders which cannot be detected automatically + int codec_tag; ///< internal codec_tag upper case converted from avctx codec_tag + int stream_codec_tag; ///< internal stream_codec_tag upper case converted from avctx stream_codec_tag /* the following fields are managed internally by the encoder */ /** bit output */ @@ -288,148 +251,151 @@ typedef struct MpegEncContext { int input_picture_number; ///< used to set pic->display_picture_number, shouldnt be used for/by anything else int coded_picture_number; ///< used to set pic->coded_picture_number, shouldnt be used for/by anything else int picture_number; //FIXME remove, unclear definition - int picture_in_gop_number; ///< 0-> first pic in gop, ... - int b_frames_since_non_b; ///< used for encoding, relative to not yet reordered input + int picture_in_gop_number; ///< 0-> first pic in gop, ... + int b_frames_since_non_b; ///< used for encoding, relative to not yet reordered input int64_t user_specified_pts;///< last non zero pts from AVFrame which was passed into avcodec_encode_video() - int mb_width, mb_height; ///< number of MBs horizontally & vertically - int mb_stride; ///< mb_width+1 used for some arrays to allow simple addressng of left & top MBs withoutt sig11 - int b8_stride; ///< 2*mb_width+1 used for some 8x8 block arrays to allow simple addressng - int b4_stride; ///< 4*mb_width+1 used for some 4x4 block arrays to allow simple addressng - int h_edge_pos, v_edge_pos;///< horizontal / vertical position of the right/bottom edge (pixel replicateion) - int mb_num; ///< number of MBs of a picture - int linesize; ///< line size, in bytes, may be different from width - int uvlinesize; ///< line size, for chroma in bytes, may be different from width - Picture *picture; ///< main picture buffer + int mb_width, mb_height; ///< number of MBs horizontally & vertically + int mb_stride; ///< mb_width+1 used for some arrays to allow simple addressing of left & top MBs without sig11 + int b8_stride; ///< 2*mb_width+1 used for some 8x8 block arrays to allow simple addressing + int b4_stride; ///< 4*mb_width+1 used for some 4x4 block arrays to allow simple addressing + int h_edge_pos, v_edge_pos;///< horizontal / vertical position of the right/bottom edge (pixel replication) + int mb_num; ///< number of MBs of a picture + int linesize; ///< line size, in bytes, may be different from width + int uvlinesize; ///< line size, for chroma in bytes, may be different from width + Picture *picture; ///< main picture buffer Picture **input_picture; ///< next pictures on display order for encoding Picture **reordered_input_picture; ///< pointer to the next pictures in codedorder for encoding - + int start_mb_y; ///< start mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y) int end_mb_y; ///< end mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y) struct MpegEncContext *thread_context[MAX_THREADS]; - - /** + + /** * copy of the previous picture structure. * note, linesize & data, might not match the previous picture (for field pictures) */ - Picture last_picture; - - /** + Picture last_picture; + + /** * copy of the next picture structure. * note, linesize & data, might not match the next picture (for field pictures) */ Picture next_picture; - - /** + + /** * copy of the source picture structure for encoding. * note, linesize & data, might not match the source picture (for field pictures) */ Picture new_picture; - - /** + + /** * copy of the current picture structure. * note, linesize & data, might not match the current picture (for field pictures) */ - Picture current_picture; ///< buffer to store the decompressed current picture - + Picture current_picture; ///< buffer to store the decompressed current picture + Picture *last_picture_ptr; ///< pointer to the previous picture. - Picture *next_picture_ptr; ///< pointer to the next picture (for bidir pred) + Picture *next_picture_ptr; ///< pointer to the next picture (for bidir pred) Picture *current_picture_ptr; ///< pointer to the current picture uint8_t *visualization_buffer[3]; //< temporary buffer vor MV visualization - int last_dc[3]; ///< last DC values for MPEG1 + int last_dc[3]; ///< last DC values for MPEG1 int16_t *dc_val_base; - int16_t *dc_val[3]; ///< used for mpeg4 DC prediction, all 3 arrays must be continuous + int16_t *dc_val[3]; ///< used for mpeg4 DC prediction, all 3 arrays must be continuous int16_t dc_cache[4*5]; int y_dc_scale, c_dc_scale; - uint8_t *y_dc_scale_table; ///< qscale -> y_dc_scale table - uint8_t *c_dc_scale_table; ///< qscale -> c_dc_scale table + const uint8_t *y_dc_scale_table; ///< qscale -> y_dc_scale table + const uint8_t *c_dc_scale_table; ///< qscale -> c_dc_scale table const uint8_t *chroma_qscale_table; ///< qscale -> chroma_qscale (h263) uint8_t *coded_block_base; uint8_t *coded_block; ///< used for coded block pattern prediction (msmpeg4v3, wmv1) int16_t (*ac_val_base)[16]; - int16_t (*ac_val[3])[16]; ///< used for for mpeg4 AC prediction, all 3 arrays must be continuous + int16_t (*ac_val[3])[16]; ///< used for for mpeg4 AC prediction, all 3 arrays must be continuous int ac_pred; - uint8_t *prev_pict_types; ///< previous picture types in bitstream order, used for mb skip + uint8_t *prev_pict_types; ///< previous picture types in bitstream order, used for mb skip #define PREV_PICT_TYPES_BUFFER_SIZE 256 - int mb_skipped; ///< MUST BE SET only during DECODING - uint8_t *mbskip_table; /**< used to avoid copy if macroblock skipped (for black regions for example) + int mb_skipped; ///< MUST BE SET only during DECODING + uint8_t *mbskip_table; /**< used to avoid copy if macroblock skipped (for black regions for example) and used for b-frame encoding & decoding (contains skip table of next P Frame) */ - uint8_t *mbintra_table; ///< used to avoid setting {ac, dc, cbp}-pred stuff to zero on inter MB decoding - uint8_t *cbp_table; ///< used to store cbp, ac_pred for partitioned decoding - uint8_t *pred_dir_table; ///< used to store pred_dir for partitioned decoding + uint8_t *mbintra_table; ///< used to avoid setting {ac, dc, cbp}-pred stuff to zero on inter MB decoding + uint8_t *cbp_table; ///< used to store cbp, ac_pred for partitioned decoding + uint8_t *pred_dir_table; ///< used to store pred_dir for partitioned decoding uint8_t *allocated_edge_emu_buffer; uint8_t *edge_emu_buffer; ///< points into the middle of allocated_edge_emu_buffer - uint8_t *rd_scratchpad; ///< scartchpad for rate distortion mb decission + uint8_t *rd_scratchpad; ///< scratchpad for rate distortion mb decision uint8_t *obmc_scratchpad; uint8_t *b_scratchpad; ///< scratchpad used for writing into write only buffers - int qscale; ///< QP - int chroma_qscale; ///< chroma QP - int lambda; ///< lagrange multipler used in rate distortion - int lambda2; ///< (lambda*lambda) >> FF_LAMBDA_SHIFT + int qscale; ///< QP + int chroma_qscale; ///< chroma QP + unsigned int lambda; ///< lagrange multipler used in rate distortion + unsigned int lambda2; ///< (lambda*lambda) >> FF_LAMBDA_SHIFT int *lambda_table; - int adaptive_quant; ///< use adaptive quantization - int dquant; ///< qscale difference to prev qscale - int pict_type; ///< I_TYPE, P_TYPE, B_TYPE, ... + int adaptive_quant; ///< use adaptive quantization + int dquant; ///< qscale difference to prev qscale + int pict_type; ///< I_TYPE, P_TYPE, B_TYPE, ... int last_pict_type; //FIXME removes - int last_non_b_pict_type; ///< used for mpeg4 gmc b-frames & ratecontrol + int last_non_b_pict_type; ///< used for mpeg4 gmc b-frames & ratecontrol int dropable; int frame_rate_index; + int last_lambda_for[5]; ///< last lambda for a specific pict type + int skipdct; ///< skip dct and code zero residual /* motion compensation */ - int unrestricted_mv; ///< mv can point outside of the coded picture - int h263_long_vectors; ///< use horrible h263v1 long vector mode + int unrestricted_mv; ///< mv can point outside of the coded picture + int h263_long_vectors; ///< use horrible h263v1 long vector mode int decode; ///< if 0 then decoding will be skipped (for encoding b frames for example) - DSPContext dsp; ///< pointers for accelerated dsp fucntions - int f_code; ///< forward MV resolution - int b_code; ///< backward MV resolution for B Frames (mpeg4) + DSPContext dsp; ///< pointers for accelerated dsp functions + int f_code; ///< forward MV resolution + int b_code; ///< backward MV resolution for B Frames (mpeg4) int16_t (*p_mv_table_base)[2]; int16_t (*b_forw_mv_table_base)[2]; int16_t (*b_back_mv_table_base)[2]; - int16_t (*b_bidir_forw_mv_table_base)[2]; - int16_t (*b_bidir_back_mv_table_base)[2]; + int16_t (*b_bidir_forw_mv_table_base)[2]; + int16_t (*b_bidir_back_mv_table_base)[2]; int16_t (*b_direct_mv_table_base)[2]; int16_t (*p_field_mv_table_base[2][2])[2]; int16_t (*b_field_mv_table_base[2][2][2])[2]; - int16_t (*p_mv_table)[2]; ///< MV table (1MV per MB) p-frame encoding - int16_t (*b_forw_mv_table)[2]; ///< MV table (1MV per MB) forward mode b-frame encoding - int16_t (*b_back_mv_table)[2]; ///< MV table (1MV per MB) backward mode b-frame encoding - int16_t (*b_bidir_forw_mv_table)[2]; ///< MV table (1MV per MB) bidir mode b-frame encoding - int16_t (*b_bidir_back_mv_table)[2]; ///< MV table (1MV per MB) bidir mode b-frame encoding - int16_t (*b_direct_mv_table)[2]; ///< MV table (1MV per MB) direct mode b-frame encoding + int16_t (*p_mv_table)[2]; ///< MV table (1MV per MB) p-frame encoding + int16_t (*b_forw_mv_table)[2]; ///< MV table (1MV per MB) forward mode b-frame encoding + int16_t (*b_back_mv_table)[2]; ///< MV table (1MV per MB) backward mode b-frame encoding + int16_t (*b_bidir_forw_mv_table)[2]; ///< MV table (1MV per MB) bidir mode b-frame encoding + int16_t (*b_bidir_back_mv_table)[2]; ///< MV table (1MV per MB) bidir mode b-frame encoding + int16_t (*b_direct_mv_table)[2]; ///< MV table (1MV per MB) direct mode b-frame encoding int16_t (*p_field_mv_table[2][2])[2]; ///< MV table (2MV per MB) interlaced p-frame encoding int16_t (*b_field_mv_table[2][2][2])[2];///< MV table (4MV per MB) interlaced b-frame encoding uint8_t (*p_field_select_table[2]); uint8_t (*b_field_select_table[2][2]); - int me_method; ///< ME algorithm + int me_method; ///< ME algorithm int mv_dir; #define MV_DIR_BACKWARD 1 #define MV_DIR_FORWARD 2 #define MV_DIRECT 4 ///< bidirectional mode where the difference equals the MV of the last P/S/I-Frame (mpeg4) int mv_type; -#define MV_TYPE_16X16 0 ///< 1 vector for the whole mb -#define MV_TYPE_8X8 1 ///< 4 vectors (h263, mpeg4 4MV) -#define MV_TYPE_16X8 2 ///< 2 vectors, one per 16x8 block -#define MV_TYPE_FIELD 3 ///< 2 vectors, one per field -#define MV_TYPE_DMV 4 ///< 2 vectors, special mpeg2 Dual Prime Vectors - /**motion vectors for a macroblock +#define MV_TYPE_16X16 0 ///< 1 vector for the whole mb +#define MV_TYPE_8X8 1 ///< 4 vectors (h263, mpeg4 4MV) +#define MV_TYPE_16X8 2 ///< 2 vectors, one per 16x8 block +#define MV_TYPE_FIELD 3 ///< 2 vectors, one per field +#define MV_TYPE_DMV 4 ///< 2 vectors, special mpeg2 Dual Prime Vectors + /**motion vectors for a macroblock first coordinate : 0 = forward 1 = backward second " : depend on type third " : 0 = x, 1 = y */ int mv[2][4][2]; int field_select[2][2]; - int last_mv[2][2][2]; ///< last MV, used for MV prediction in MPEG1 & B-frame MPEG4 - uint8_t *fcode_tab; ///< smallest fcode needed for each MV - + int last_mv[2][2][2]; ///< last MV, used for MV prediction in MPEG1 & B-frame MPEG4 + uint8_t *fcode_tab; ///< smallest fcode needed for each MV + int16_t direct_scale_mv[2][64]; ///< precomputed to avoid divisions in ff_mpeg4_set_direct_mv + MotionEstContext me; - int no_rounding; /**< apply no rounding to motion compensation (MPEG4, msmpeg4, ...) - for b-frames rounding mode is allways 0 */ + int no_rounding; /**< apply no rounding to motion compensation (MPEG4, msmpeg4, ...) + for b-frames rounding mode is always 0 */ int hurry_up; /**< when set to 1 during decoding, b frames will be skipped when set to 2 idct/dequant will be skipped too */ - + /* macroblock layer */ int mb_x, mb_y; int mb_skip_run; @@ -451,10 +417,12 @@ typedef struct MpegEncContext { #define CANDIDATE_MB_TYPE_BACKWARD_I 0x400 #define CANDIDATE_MB_TYPE_BIDIR_I 0x800 +#define CANDIDATE_MB_TYPE_DIRECT0 0x1000 + int block_index[6]; ///< index to current MB in block based arrays with edges int block_wrap[6]; uint8_t *dest[3]; - + int *mb_index2xy; ///< mb_index -> mb_x + mb_y*mb_stride /** matrix transmitted in the bitstream */ @@ -463,11 +431,11 @@ typedef struct MpegEncContext { uint16_t inter_matrix[64]; uint16_t chroma_inter_matrix[64]; #define QUANT_BIAS_SHIFT 8 - int intra_quant_bias; ///< bias for the quantizer - int inter_quant_bias; ///< bias for the quantizer - int min_qcoeff; ///< minimum encodable coefficient - int max_qcoeff; ///< maximum encodable coefficient - int ac_esc_length; ///< num of bits needed to encode the longest esc + int intra_quant_bias; ///< bias for the quantizer + int inter_quant_bias; ///< bias for the quantizer + int min_qcoeff; ///< minimum encodable coefficient + int max_qcoeff; ///< maximum encodable coefficient + int ac_esc_length; ///< num of bits needed to encode the longest esc uint8_t *intra_ac_vlc_length; uint8_t *intra_ac_vlc_last_length; uint8_t *inter_ac_vlc_length; @@ -476,7 +444,7 @@ typedef struct MpegEncContext { uint8_t *chroma_dc_vlc_length; #define UNI_AC_ENC_INDEX(run,level) ((run)*128 + (level)) - int coded_score[6]; + int coded_score[8]; /** precomputed matrix (combine qscale and DCT renorm) */ int (*q_intra_matrix)[64]; @@ -486,11 +454,11 @@ typedef struct MpegEncContext { uint16_t (*q_inter_matrix16)[2][64]; int block_last_index[12]; ///< last non zero coefficient in block /* scantables */ - ScanTable __align8 intra_scantable; + DECLARE_ALIGNED_8(ScanTable, intra_scantable); ScanTable intra_h_scantable; ScanTable intra_v_scantable; ScanTable inter_scantable; ///< if inter == intra then intra should be used to reduce tha cache usage - + /* noise reduction */ int (*dct_error_sum)[64]; int dct_count[2]; @@ -501,7 +469,8 @@ typedef struct MpegEncContext { /* bit rate control */ int64_t wanted_bits; int64_t total_bits; - int frame_bits; ///< bits used for the current frame + int frame_bits; ///< bits used for the current frame + int next_lambda; ///< next lambda used for retrying to encode a frame RateControlContext rc_context; ///< contains stuff only accessed in ratecontrol.c /* statistics, used for 2-pass encoding */ @@ -515,11 +484,11 @@ typedef struct MpegEncContext { int skip_count; int misc_bits; ///< cbp, mb_type int last_bits; ///< temp var used for calculating the above vars - + /* error concealment / resync */ int error_count; - uint8_t *error_status_table; ///< table of the error status of each MB -#define VP_START 1 ///< current MB is the first after a resync marker + uint8_t *error_status_table; ///< table of the error status of each MB +#define VP_START 1 ///< current MB is the first after a resync marker #define AC_ERROR 2 #define DC_ERROR 4 #define MV_ERROR 8 @@ -527,41 +496,40 @@ typedef struct MpegEncContext { #define DC_END 32 #define MV_END 64 //FIXME some prefix? - - int resync_mb_x; ///< x position of last resync marker - int resync_mb_y; ///< y position of last resync marker - GetBitContext last_resync_gb; ///< used to search for the next resync marker + + int resync_mb_x; ///< x position of last resync marker + int resync_mb_y; ///< y position of last resync marker + GetBitContext last_resync_gb; ///< used to search for the next resync marker int mb_num_left; ///< number of MBs left in this video packet (for partitioned Slices only) - int next_p_frame_damaged; ///< set if the next p frame is damaged, to avoid showing trashed b frames + int next_p_frame_damaged; ///< set if the next p frame is damaged, to avoid showing trashed b frames int error_resilience; - + ParseContext parse_context; /* H.263 specific */ int gob_index; int obmc; ///< overlapped block motion compensation - + /* H.263+ specific */ - int umvplus; ///< == H263+ && unrestricted_mv - int h263_aic; ///< Advanded INTRA Coding (AIC) + int umvplus; ///< == H263+ && unrestricted_mv + int h263_aic; ///< Advanded INTRA Coding (AIC) int h263_aic_dir; ///< AIC direction: 0 = left, 1 = top int h263_slice_structured; int alt_inter_vlc; ///< alternative inter vlc int modified_quant; - int loop_filter; + int loop_filter; int custom_pcf; - + /* mpeg4 specific */ - int time_increment_resolution; - int time_increment_bits; ///< number of bits to represent the fractional part of time + int time_increment_bits; ///< number of bits to represent the fractional part of time int last_time_base; - int time_base; ///< time in seconds of last I,P,S Frame - int64_t time; ///< time of current frame + int time_base; ///< time in seconds of last I,P,S Frame + int64_t time; ///< time of current frame int64_t last_non_b_time; - uint16_t pp_time; ///< time distance between the last 2 p,s,i frames - uint16_t pb_time; ///< time distance between the last b and p,s,i frame + uint16_t pp_time; ///< time distance between the last 2 p,s,i frames + uint16_t pb_time; ///< time distance between the last b and p,s,i frame uint16_t pp_field_time; - uint16_t pb_field_time; ///< like above, just for interlaced + uint16_t pb_field_time; ///< like above, just for interlaced int shape; int vol_sprite_usage; int sprite_width; @@ -571,12 +539,12 @@ typedef struct MpegEncContext { int sprite_brightness_change; int num_sprite_warping_points; int real_sprite_warping_points; - int sprite_offset[2][2]; ///< sprite offset[isChroma][isMVY] - int sprite_delta[2][2]; ///< sprite_delta [isY][isMVY] - int sprite_shift[2]; ///< sprite shift [isChroma] + int sprite_offset[2][2]; ///< sprite offset[isChroma][isMVY] + int sprite_delta[2][2]; ///< sprite_delta [isY][isMVY] + int sprite_shift[2]; ///< sprite shift [isChroma] int mcsel; int quant_precision; - int quarter_sample; ///< 1->qpel, 0->half pel ME/MC + int quarter_sample; ///< 1->qpel, 0->half pel ME/MC int scalability; int hierachy_type; int enhancement_type; @@ -585,19 +553,20 @@ typedef struct MpegEncContext { int aspect_ratio_info; //FIXME remove int sprite_warping_accuracy; int low_latency_sprite; - int data_partitioning; ///< data partitioning flag from header - int partitioned_frame; ///< is current frame partitioned - int rvlc; ///< reversible vlc + int data_partitioning; ///< data partitioning flag from header + int partitioned_frame; ///< is current frame partitioned + int rvlc; ///< reversible vlc int resync_marker; ///< could this stream contain resync markers - int low_delay; ///< no reordering needed / has no b-frames + int low_delay; ///< no reordering needed / has no b-frames int vo_type; - int vol_control_parameters; ///< does the stream contain the low_delay flag, used to workaround buggy encoders - int intra_dc_threshold; ///< QP above whch the ac VLC should be used for intra dc - PutBitContext tex_pb; ///< used for data partitioned VOPs - PutBitContext pb2; ///< used for data partitioned VOPs + int vol_control_parameters; ///< does the stream contain the low_delay flag, used to workaround buggy encoders + int intra_dc_threshold; ///< QP above whch the ac VLC should be used for intra dc + int use_intra_dc_vlc; + PutBitContext tex_pb; ///< used for data partitioned VOPs + PutBitContext pb2; ///< used for data partitioned VOPs int mpeg_quant; - int t_frame; ///< time distance of first I -> B, used for interlaced b frames - int padding_bug_score; ///< used to detect the VERY common padding bug in MPEG4 + int t_frame; ///< time distance of first I -> B, used for interlaced b frames + int padding_bug_score; ///< used to detect the VERY common padding bug in MPEG4 /* divx specific, used to workaround (many) bugs in divx5 */ int divx_version; @@ -605,24 +574,21 @@ typedef struct MpegEncContext { int divx_packed; uint8_t *bitstream_buffer; //Divx 5.01 puts several frames in a single one, this is used to reorder them int bitstream_buffer_size; - int allocated_bitstream_buffer_size; - + unsigned int allocated_bitstream_buffer_size; + int xvid_build; - + /* lavc specific stuff, used to workaround bugs in libavcodec */ - int ffmpeg_version; int lavc_build; - + /* RV10 specific */ - int rv10_version; ///< RV10 version: 0 or 3 + int rv10_version; ///< RV10 version: 0 or 3 int rv10_first_dc_coded[3]; - + /* MJPEG specific */ struct MJpegContext *mjpeg_ctx; - int mjpeg_vsample[3]; ///< vertical sampling factors, default = {2, 1, 1} - int mjpeg_hsample[3]; ///< horizontal sampling factors, default = {2, 1, 1} - int mjpeg_write_tables; ///< do we want to have quantisation- and huffmantables in the jpeg file ? - int mjpeg_data_only_frames; ///< frames only with SOI, SOS and EOI markers + int mjpeg_vsample[3]; ///< vertical sampling factors, default = {2, 1, 1} + int mjpeg_hsample[3]; ///< horizontal sampling factors, default = {2, 1, 1} /* MSMPEG4 specific */ int mv_table_index; @@ -630,8 +596,8 @@ typedef struct MpegEncContext { int rl_chroma_table_index; int dc_table_index; int use_skip_mb_code; - int slice_height; ///< in macroblocks - int first_slice_line; ///< used in mpeg4 too to handle resync markers + int slice_height; ///< in macroblocks + int first_slice_line; ///< used in mpeg4 too to handle resync markers int flipflop_rounding; int msmpeg4_version; ///< 0=not msmpeg4, 1=mp41, 2=mp42, 3=mp43/divx3 4=wmv1/7 5=wmv2/8 int per_mb_rl_table; @@ -646,11 +612,11 @@ typedef struct MpegEncContext { GetBitContext gb; /* Mpeg1 specific */ - int gop_picture_number; ///< index of the first picture of a GOP based on fake_pic_num & mpeg1 specific - int last_mv_dir; ///< last mv_dir, used for b frame encoding + int gop_picture_number; ///< index of the first picture of a GOP based on fake_pic_num & mpeg1 specific + int last_mv_dir; ///< last mv_dir, used for b frame encoding int broken_link; ///< no_output_of_prior_pics_flag - uint8_t *vbv_delay_ptr; ///< pointer to vbv_delay in the bitstream - + uint8_t *vbv_delay_ptr; ///< pointer to vbv_delay in the bitstream + /* MPEG2 specific - I wish I had not to support this mess. */ int progressive_sequence; int mpeg_f_code[2][2]; @@ -684,34 +650,34 @@ typedef struct MpegEncContext { /* RTP specific */ int rtp_mode; - + uint8_t *ptr_lastgob; int swap_uv;//vcr2 codec is mpeg2 varint with UV swaped short * pblocks[12]; - - DCTELEM (*block)[64]; ///< points to one of the following blocks - DCTELEM (*blocks)[6][64]; // for HQ mode we need to keep the best block + + DCTELEM (*block)[64]; ///< points to one of the following blocks + DCTELEM (*blocks)[8][64]; // for HQ mode we need to keep the best block int (*decode_mb)(struct MpegEncContext *s, DCTELEM block[6][64]); // used by some codecs to avoid a switch() #define SLICE_OK 0 #define SLICE_ERROR -1 #define SLICE_END -2 ///index_run[last][run]; - if (index >= rl->n) - return rl->n; - if (level > rl->max_level[last][run]) - return rl->n; - return index + level - 1; -} - -extern uint8_t ff_mpeg4_y_dc_scale_table[32]; -extern uint8_t ff_mpeg4_c_dc_scale_table[32]; -extern uint8_t ff_aic_dc_scale_table[32]; +extern const uint8_t ff_mpeg4_y_dc_scale_table[32]; +extern const uint8_t ff_mpeg4_c_dc_scale_table[32]; +extern const uint8_t ff_aic_dc_scale_table[32]; extern const int16_t ff_mpeg4_default_intra_matrix[64]; extern const int16_t ff_mpeg4_default_non_intra_matrix[64]; extern const uint8_t ff_h263_chroma_qscale_table[32]; @@ -876,18 +814,19 @@ void ff_h261_encode_mb(MpegEncContext *s, int motion_x, int motion_y); void ff_h261_encode_picture_header(MpegEncContext * s, int picture_number); void ff_h261_encode_init(MpegEncContext *s); +int ff_h261_get_picture_format(int width, int height); /* h263.c, h263dec.c */ int ff_h263_decode_init(AVCodecContext *avctx); -int ff_h263_decode_frame(AVCodecContext *avctx, +int ff_h263_decode_frame(AVCodecContext *avctx, void *data, int *data_size, uint8_t *buf, int buf_size); int ff_h263_decode_end(AVCodecContext *avctx); -void h263_encode_mb(MpegEncContext *s, +void h263_encode_mb(MpegEncContext *s, DCTELEM block[6][64], int motion_x, int motion_y); -void mpeg4_encode_mb(MpegEncContext *s, +void mpeg4_encode_mb(MpegEncContext *s, DCTELEM block[6][64], int motion_x, int motion_y); void h263_encode_picture_header(MpegEncContext *s, int picture_number); @@ -895,11 +834,15 @@ void ff_flv_encode_picture_header(MpegEncContext *s, int picture_number); void h263_encode_gob_header(MpegEncContext * s, int mb_line); int16_t *h263_pred_motion(MpegEncContext * s, int block, int dir, int *px, int *py); -void mpeg4_pred_ac(MpegEncContext * s, DCTELEM *block, int n, +void mpeg4_pred_ac(MpegEncContext * s, DCTELEM *block, int n, int dir); void ff_set_mpeg4_time(MpegEncContext * s, int picture_number); void mpeg4_encode_picture_header(MpegEncContext *s, int picture_number); +#ifdef CONFIG_ENCODERS void h263_encode_init(MpegEncContext *s); +#else +static void h263_encode_init(MpegEncContext *s) {assert(0);} +#endif void h263_decode_init_vlc(MpegEncContext *s); int h263_decode_picture_header(MpegEncContext *s); int ff_h263_decode_gob_header(MpegEncContext *s); @@ -928,10 +871,10 @@ int ff_mpeg4_decode_partitions(MpegEncContext *s); int ff_mpeg4_get_video_packet_prefix_length(MpegEncContext *s); int ff_h263_resync(MpegEncContext *s); int ff_h263_get_gob_height(MpegEncContext *s); +void ff_mpeg4_init_direct_mv(MpegEncContext *s); int ff_mpeg4_set_direct_mv(MpegEncContext *s, int mx, int my); int ff_h263_round_chroma(int x); void ff_h263_encode_motion(MpegEncContext * s, int val, int f_code); -int ff_mpeg4_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_size); /* rv10.c */ @@ -943,7 +886,7 @@ void rv20_encode_picture_header(MpegEncContext *s, int picture_number); /* msmpeg4.c */ void msmpeg4_encode_picture_header(MpegEncContext * s, int picture_number); void msmpeg4_encode_ext_header(MpegEncContext * s); -void msmpeg4_encode_mb(MpegEncContext * s, +void msmpeg4_encode_mb(MpegEncContext * s, DCTELEM block[6][64], int motion_x, int motion_y); int msmpeg4_decode_picture_header(MpegEncContext * s); @@ -958,30 +901,9 @@ void ff_mspel_motion(MpegEncContext *s, uint8_t **ref_picture, op_pixels_func (*pix_op)[4], int motion_x, int motion_y, int h); int ff_wmv2_encode_picture_header(MpegEncContext * s, int picture_number); -void ff_wmv2_encode_mb(MpegEncContext * s, +void ff_wmv2_encode_mb(MpegEncContext * s, DCTELEM block[6][64], int motion_x, int motion_y); -/* mjpeg.c */ -int mjpeg_init(MpegEncContext *s); -void mjpeg_close(MpegEncContext *s); -void mjpeg_encode_mb(MpegEncContext *s, - DCTELEM block[6][64]); -void mjpeg_picture_header(MpegEncContext *s); -void mjpeg_picture_trailer(MpegEncContext *s); -void ff_mjpeg_stuffing(PutBitContext * pbc); - - -/* rate control */ -int ff_rate_control_init(MpegEncContext *s); -float ff_rate_estimate_qscale(MpegEncContext *s); -void ff_write_pass1_stats(MpegEncContext *s); -void ff_rate_control_uninit(MpegEncContext *s); -double ff_eval(char *s, double *const_value, const char **const_name, - double (**func1)(void *, double), const char **func1_name, - double (**func2)(void *, double, double), char **func2_name, - void *opaque); -int ff_vbv_update(MpegEncContext *s, int frame_size); - - #endif /* AVCODEC_MPEGVIDEO_H */ +