1 /*****************************************************************************
2 * common.h: h264 encoder
3 *****************************************************************************
4 * Copyright (C) 2003 Laurent Aimar
5 * $Id: common.h,v 1.1 2004/06/03 19:27:06 fenrir Exp $
7 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA.
22 *****************************************************************************/
37 #define snprintf _snprintf
38 #define X264_VERSION "" // no configure script for msvc
41 /* alloca: force 16byte alignment */
43 #define x264_alloca(x) (void*)(((intptr_t)_alloca((x)+15)+15)&~15)
45 #define x264_alloca(x) (void*)(((intptr_t) alloca((x)+15)+15)&~15)
51 #define pthread_t HANDLE
52 #define pthread_create(t,u,f,d) *(t)=CreateThread(NULL,0,f,d,0,NULL)
53 #define pthread_join(t,s) { WaitForSingleObject(t,INFINITE); \
55 #define HAVE_PTHREAD 1
57 #elif defined(SYS_BEOS)
58 #include <kernel/OS.h>
59 #define pthread_t thread_id
60 #define pthread_create(t,u,f,d) { *(t)=spawn_thread(f,"",10,d); \
61 resume_thread(*(t)); }
62 #define pthread_join(t,s) { long tmp; \
63 wait_for_thread(t,(s)?(long*)(s):&tmp); }
64 #define HAVE_PTHREAD 1
66 #elif defined(HAVE_PTHREAD)
70 /****************************************************************************
72 ****************************************************************************/
73 #define X264_MIN(a,b) ( (a)<(b) ? (a) : (b) )
74 #define X264_MAX(a,b) ( (a)>(b) ? (a) : (b) )
75 #define X264_MIN3(a,b,c) X264_MIN((a),X264_MIN((b),(c)))
76 #define X264_MAX3(a,b,c) X264_MAX((a),X264_MAX((b),(c)))
77 #define X264_MIN4(a,b,c,d) X264_MIN((a),X264_MIN3((b),(c),(d)))
78 #define X264_MAX4(a,b,c,d) X264_MAX((a),X264_MAX3((b),(c),(d)))
79 #define XCHG(type,a,b) { type t = a; a = b; b = t; }
80 #define FIX8(f) ((int)(f*(1<<8)+.5))
82 #if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 0)
83 #define UNUSED __attribute__((unused))
88 #define CHECKED_MALLOC( var, size )\
90 var = x264_malloc( size );\
93 x264_log( h, X264_LOG_ERROR, "malloc failed\n" );\
98 #define X264_BFRAME_MAX 16
99 #define X264_SLICE_MAX 4
100 #define X264_NAL_MAX (4 + X264_SLICE_MAX)
102 /****************************************************************************
104 ****************************************************************************/
117 /****************************************************************************
119 ****************************************************************************/
120 /* x264_malloc : will do or emulate a memalign
121 * XXX you HAVE TO use x264_free for buffer allocated
124 void *x264_malloc( int );
125 void *x264_realloc( void *p, int i_size );
126 void x264_free( void * );
128 /* x264_slurp_file: malloc space for the whole file and read it */
129 char *x264_slurp_file( const char *filename );
131 /* mdate: return the current date in microsecond */
132 int64_t x264_mdate( void );
134 /* x264_param2string: return a (malloced) string containing most of
135 * the encoding options */
136 char *x264_param2string( x264_param_t *p, int b_res );
139 void x264_log( x264_t *h, int i_level, const char *psz_fmt, ... );
141 void x264_reduce_fraction( int *n, int *d );
143 static inline int x264_clip3( int v, int i_min, int i_max )
145 return ( (v < i_min) ? i_min : (v > i_max) ? i_max : v );
148 static inline float x264_clip3f( float v, float f_min, float f_max )
150 return ( (v < f_min) ? f_min : (v > f_max) ? f_max : v );
153 static inline int x264_median( int a, int b, int c )
159 max = b; /* no need to do 'b > max' (more consuming than always doing affectation) */
166 return a + b + c - min - max;
170 /****************************************************************************
172 ****************************************************************************/
182 static const char slice_type_to_char[] = { 'P', 'B', 'I', 'S', 'S' };
201 int i_idr_pic_id; /* -1 if nal_type != 5 */
204 int i_delta_poc_bottom;
207 int i_redundant_pic_cnt;
209 int b_direct_spatial_mv_pred;
211 int b_num_ref_idx_override;
212 int i_num_ref_idx_l0_active;
213 int i_num_ref_idx_l1_active;
215 int b_ref_pic_list_reordering_l0;
216 int b_ref_pic_list_reordering_l1;
220 } ref_pic_list_order[2][16];
222 int i_cabac_init_idc;
229 /* deblocking filter */
230 int i_disable_deblocking_filter_idc;
231 int i_alpha_c0_offset;
234 } x264_slice_header_t;
238 #define X264_SCAN8_SIZE (6*8)
239 #define X264_SCAN8_0 (4+1*8)
241 static const int x264_scan8[16+2*4] =
244 4+1*8, 5+1*8, 4+2*8, 5+2*8,
245 6+1*8, 7+1*8, 6+2*8, 7+2*8,
246 4+3*8, 5+3*8, 4+4*8, 5+4*8,
247 6+3*8, 7+3*8, 6+4*8, 7+4*8,
267 typedef struct x264_ratecontrol_t x264_ratecontrol_t;
268 typedef struct x264_vlc_table_t x264_vlc_table_t;
272 /* encoder parameters */
275 x264_t *thread[X264_SLICE_MAX];
277 /* bitstream output */
281 x264_nal_t nal[X264_NAL_MAX];
282 int i_bitstream; /* size of p_bitstream */
283 uint8_t *p_bitstream; /* will hold data for all nal */
287 /* frame number/poc */
290 int i_frame_offset; /* decoding only */
291 int i_frame_num; /* decoding only */
292 int i_poc_msb; /* decoding only */
293 int i_poc_lsb; /* decoding only */
294 int i_poc; /* decoding only */
296 int i_thread_num; /* threads only */
297 int i_nal_type; /* threads only */
298 int i_nal_ref_idc; /* threads only */
300 /* We use only one SPS and one PPS */
301 x264_sps_t sps_array[1];
303 x264_pps_t pps_array[1];
307 int (*dequant4_mf[4])[4][4]; /* [4][6][4][4] */
308 int (*dequant8_mf[2])[8][8]; /* [2][6][8][8] */
309 int (*quant4_mf[4])[4][4]; /* [4][6][4][4] */
310 int (*quant8_mf[2])[8][8]; /* [2][6][8][8] */
311 int (*unquant4_mf[4])[16]; /* [4][52][16] */
312 int (*unquant8_mf[2])[64]; /* [2][52][64] */
314 uint32_t nr_residual_sum[2][64];
315 uint32_t nr_offset[2][64];
316 uint32_t nr_count[2];
319 x264_slice_header_t sh;
326 /* Frames to be encoded (whose types have been decided) */
327 x264_frame_t *current[X264_BFRAME_MAX+3];
328 /* Temporary buffer (frames types not yet decided) */
329 x264_frame_t *next[X264_BFRAME_MAX+3];
331 x264_frame_t *unused[X264_BFRAME_MAX+3];
332 /* For adaptive B decision */
333 x264_frame_t *last_nonb;
335 /* frames used for reference +1 for decoding + sentinels */
336 x264_frame_t *reference[16+2+1+2];
338 int i_last_idr; /* Frame number of the last IDR */
340 int i_input; /* Number of input frames already accepted */
342 int i_max_dpb; /* Number of frames allocated in the decoded picture buffer */
345 int i_delay; /* Number of frames buffered for B reordering */
346 int b_have_lowres; /* Whether 1/2 resolution luma planes are being used */
349 /* current frame being encoded */
352 /* frame being reconstructed */
355 /* references lists */
357 x264_frame_t *fref0[16+3]; /* ref list 0 */
359 x264_frame_t *fref1[16+3]; /* ref list 1 */
360 int b_ref_reorder[2];
364 /* Current MB DCT coeffs */
367 DECLARE_ALIGNED( int, luma16x16_dc[16], 16 );
368 DECLARE_ALIGNED( int, chroma_dc[2][4], 16 );
369 // FIXME merge with union
370 DECLARE_ALIGNED( int, luma8x8[4][64], 16 );
373 DECLARE_ALIGNED( int, residual_ac[15], 16 );
374 DECLARE_ALIGNED( int, luma4x4[16], 16 );
378 /* MB table and cache for current frame/mb */
381 int i_mb_count; /* number of mbs in a frame */
395 /* Search parameters */
400 int b_noise_reduction;
404 /* Allowed qpel MV range to stay within the picture + emulated edge pixels */
407 /* Subpel MV range for motion search.
408 * same mv_min/max but includes levels' i_mv_range. */
411 /* Fullpel MV range for motion search */
415 /* neighboring MBs */
416 unsigned int i_neighbour;
417 unsigned int i_neighbour8[4]; /* neighbours of each 8x8 or 4x4 block that are available */
418 unsigned int i_neighbour4[16]; /* at the time the block is coded */
421 int i_mb_type_topleft;
422 int i_mb_type_topright;
427 int8_t *type; /* mb type */
428 int8_t *qp; /* mb qp */
429 int16_t *cbp; /* mb cbp: 0x0?: luma, 0x?0: chroma, 0x100: luma dc, 0x0200 and 0x0400: chroma dc (all set for PCM)*/
430 int8_t (*intra4x4_pred_mode)[7]; /* intra4x4 pred mode. for non I4x4 set to I_PRED_4x4_DC(2) */
431 uint8_t (*non_zero_count)[16+4+4]; /* nzc. for I_PCM set to 16 */
432 int8_t *chroma_pred_mode; /* chroma_pred_mode. cabac only. for non intra I_PRED_CHROMA_DC(0) */
433 int16_t (*mv[2])[2]; /* mb mv. set to 0 for intra mb */
434 int16_t (*mvd[2])[2]; /* mb mv difference with predict. set to 0 if intra. cabac only */
435 int8_t *ref[2]; /* mb ref. set to -1 if non used (intra or Lx only) */
436 int16_t (*mvr[2][32])[2]; /* 16x16 mv for each possible ref */
437 int8_t *skipbp; /* block pattern for SKIP or DIRECT (sub)mbs. B-frames + cabac only */
438 int8_t *mb_transform_size; /* transform_size_8x8_flag of each mb */
443 int i_sub_partition[4];
449 int i_intra16x16_pred_mode;
450 int i_chroma_pred_mode;
454 /* space for p_fenc and p_fdec */
455 #define FENC_STRIDE 16
456 #define FDEC_STRIDE 32
457 DECLARE_ALIGNED( uint8_t, fenc_buf[24*FENC_STRIDE], 16 );
458 DECLARE_ALIGNED( uint8_t, fdec_buf[27*FDEC_STRIDE], 16 );
460 /* pointer over mb of the frame to be compressed */
463 /* pointer over mb of the frame to be reconstructed */
466 /* pointer over mb of the references */
468 uint8_t *p_fref[2][32][4+2]; /* last: lN, lH, lV, lHV, cU, cV */
469 uint16_t *p_integral[2][16];
478 /* real intra4x4_pred_mode if I_4X4 or I_8X8, I_PRED_4x4_DC if mb available, -1 if not */
479 int intra4x4_pred_mode[X264_SCAN8_SIZE];
481 /* i_non_zero_count if available else 0x80 */
482 int non_zero_count[X264_SCAN8_SIZE];
484 /* -1 if unused, -2 if unavailable */
485 int8_t ref[2][X264_SCAN8_SIZE];
487 /* 0 if not available */
488 int16_t mv[2][X264_SCAN8_SIZE][2];
489 int16_t mvd[2][X264_SCAN8_SIZE][2];
491 /* 1 if SKIP or DIRECT. set only for B-frames + CABAC */
492 int8_t skip[X264_SCAN8_SIZE];
494 int16_t direct_mv[2][X264_SCAN8_SIZE][2];
495 int8_t direct_ref[2][X264_SCAN8_SIZE];
497 /* number of neighbors (top and left) that used 8x8 dct */
498 int i_neighbour_transform_size;
499 int b_transform_8x8_allowed;
500 int i_neighbour_interlaced;
504 int i_qp; /* current qp */
505 int i_last_qp; /* last qp */
506 int i_last_dqp; /* last delta qp */
507 int b_variable_qp; /* whether qp is allowed to vary per macroblock */
509 int b_direct_auto_read; /* take stats for --direct auto from the 2pass log */
510 int b_direct_auto_write; /* analyse direct modes, to use and/or save */
512 /* B_direct and weighted prediction */
513 int dist_scale_factor[16][2];
514 int bipred_weight[32][4];
515 /* maps fref1[0]'s ref indices into the current list0 */
516 int map_col_to_list0_buf[2]; // for negative indices
517 int map_col_to_list0[16];
520 /* rate control encoding only */
521 x264_ratecontrol_t *rc;
526 /* Current frame stats */
529 /* Headers bits (MV+Ref+MB Block Type */
531 /* Texture bits (Intra/Predicted) */
541 int i_mb_count_8x8dct[2];
542 int i_mb_count_size[7];
543 int i_mb_count_ref[32];
544 /* Estimated (SATD) cost as Intra/Predicted frame */
545 /* XXX: both omit the cost of MBs coded as P_SKIP */
548 /* Adaptive direct mv pred */
549 int i_direct_score[2];
552 /* Cumulated stats */
555 int i_slice_count[5];
556 int64_t i_slice_size[5];
559 int64_t i_sqe_global[5];
560 float f_psnr_average[5];
561 float f_psnr_mean_y[5];
562 float f_psnr_mean_u[5];
563 float f_psnr_mean_v[5];
564 float f_ssim_mean_y[5];
566 int64_t i_mb_count[5][19];
567 int64_t i_mb_count_8x8dct[2];
568 int64_t i_mb_count_size[2][7];
569 int64_t i_mb_count_ref[2][32];
571 int i_direct_score[2];
572 int i_direct_frames[2];
576 /* CPU functions dependents */
577 x264_predict_t predict_16x16[4+3];
578 x264_predict_t predict_8x8c[4+3];
579 x264_predict8x8_t predict_8x8[9+3];
580 x264_predict_t predict_4x4[9+3];
582 x264_pixel_function_t pixf;
583 x264_mc_functions_t mc;
584 x264_dct_function_t dctf;
585 x264_zigzag_function_t zigzagf;
586 x264_csp_function_t csp;
587 x264_quant_function_t quantf;
588 x264_deblock_function_t loopf;
590 /* vlc table for decoding purpose only */
591 x264_vlc_table_t *x264_coeff_token_lookup[5];
592 x264_vlc_table_t *x264_level_prefix_lookup;
593 x264_vlc_table_t *x264_total_zeros_lookup[15];
594 x264_vlc_table_t *x264_total_zeros_dc_lookup[3];
595 x264_vlc_table_t *x264_run_before_lookup[7];
598 struct visualize_t *visualize;
602 // included at the end because it needs x264_t
603 #include "macroblock.h"