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 *****************************************************************************/
36 #define snprintf _snprintf
37 #define X264_VERSION "" // no configure script for msvc
43 #define pthread_t HANDLE
44 #define pthread_create(t,u,f,d) *(t)=CreateThread(NULL,0,f,d,0,NULL)
45 #define pthread_join(t,s) { WaitForSingleObject(t,INFINITE); \
47 #define HAVE_PTHREAD 1
49 #elif defined(SYS_BEOS)
50 #include <kernel/OS.h>
51 #define pthread_t thread_id
52 #define pthread_create(t,u,f,d) { *(t)=spawn_thread(f,"",10,d); \
53 resume_thread(*(t)); }
54 #define pthread_join(t,s) { long tmp; \
55 wait_for_thread(t,(s)?(long*)(s):&tmp); }
56 #define HAVE_PTHREAD 1
58 #elif defined(HAVE_PTHREAD)
62 /****************************************************************************
64 ****************************************************************************/
65 #define X264_MIN(a,b) ( (a)<(b) ? (a) : (b) )
66 #define X264_MAX(a,b) ( (a)>(b) ? (a) : (b) )
67 #define X264_MIN3(a,b,c) X264_MIN((a),X264_MIN((b),(c)))
68 #define X264_MAX3(a,b,c) X264_MAX((a),X264_MAX((b),(c)))
69 #define X264_MIN4(a,b,c,d) X264_MIN((a),X264_MIN3((b),(c),(d)))
70 #define X264_MAX4(a,b,c,d) X264_MAX((a),X264_MAX3((b),(c),(d)))
71 #define XCHG(type,a,b) { type t = a; a = b; b = t; }
72 #define FIX8(f) ((int)(f*(1<<8)+.5))
74 #if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 0)
75 #define UNUSED __attribute__((unused))
80 #define CHECKED_MALLOC( var, size )\
82 var = x264_malloc( size );\
85 x264_log( h, X264_LOG_ERROR, "malloc failed\n" );\
90 #define X264_BFRAME_MAX 16
91 #define X264_SLICE_MAX 4
92 #define X264_NAL_MAX (4 + X264_SLICE_MAX)
94 /****************************************************************************
96 ****************************************************************************/
109 /****************************************************************************
111 ****************************************************************************/
112 /* x264_malloc : will do or emulate a memalign
113 * XXX you HAVE TO use x264_free for buffer allocated
116 void *x264_malloc( int );
117 void *x264_realloc( void *p, int i_size );
118 void x264_free( void * );
120 /* x264_slurp_file: malloc space for the whole file and read it */
121 char *x264_slurp_file( const char *filename );
123 /* mdate: return the current date in microsecond */
124 int64_t x264_mdate( void );
126 /* x264_param2string: return a (malloced) string containing most of
127 * the encoding options */
128 char *x264_param2string( x264_param_t *p, int b_res );
131 void x264_log( x264_t *h, int i_level, const char *psz_fmt, ... );
133 void x264_reduce_fraction( int *n, int *d );
135 static inline int x264_clip3( int v, int i_min, int i_max )
137 return ( (v < i_min) ? i_min : (v > i_max) ? i_max : v );
140 static inline float x264_clip3f( float v, float f_min, float f_max )
142 return ( (v < f_min) ? f_min : (v > f_max) ? f_max : v );
145 static inline int x264_median( int a, int b, int c )
151 max = b; /* no need to do 'b > max' (more consuming than always doing affectation) */
158 return a + b + c - min - max;
162 /****************************************************************************
164 ****************************************************************************/
174 static const char slice_type_to_char[] = { 'P', 'B', 'I', 'S', 'S' };
192 int i_idr_pic_id; /* -1 if nal_type != 5 */
195 int i_delta_poc_bottom;
198 int i_redundant_pic_cnt;
200 int b_direct_spatial_mv_pred;
202 int b_num_ref_idx_override;
203 int i_num_ref_idx_l0_active;
204 int i_num_ref_idx_l1_active;
206 int b_ref_pic_list_reordering_l0;
207 int b_ref_pic_list_reordering_l1;
211 } ref_pic_list_order[2][16];
213 int i_cabac_init_idc;
220 /* deblocking filter */
221 int i_disable_deblocking_filter_idc;
222 int i_alpha_c0_offset;
225 } x264_slice_header_t;
229 #define X264_SCAN8_SIZE (6*8)
230 #define X264_SCAN8_0 (4+1*8)
232 static const int x264_scan8[16+2*4] =
235 4+1*8, 5+1*8, 4+2*8, 5+2*8,
236 6+1*8, 7+1*8, 6+2*8, 7+2*8,
237 4+3*8, 5+3*8, 4+4*8, 5+4*8,
238 6+3*8, 7+3*8, 6+4*8, 7+4*8,
258 typedef struct x264_ratecontrol_t x264_ratecontrol_t;
259 typedef struct x264_vlc_table_t x264_vlc_table_t;
263 /* encoder parameters */
266 x264_t *thread[X264_SLICE_MAX];
268 /* bitstream output */
272 x264_nal_t nal[X264_NAL_MAX];
273 int i_bitstream; /* size of p_bitstream */
274 uint8_t *p_bitstream; /* will hold data for all nal */
278 /* frame number/poc */
281 int i_frame_offset; /* decoding only */
282 int i_frame_num; /* decoding only */
283 int i_poc_msb; /* decoding only */
284 int i_poc_lsb; /* decoding only */
285 int i_poc; /* decoding only */
287 int i_thread_num; /* threads only */
288 int i_nal_type; /* threads only */
289 int i_nal_ref_idc; /* threads only */
291 /* We use only one SPS and one PPS */
292 x264_sps_t sps_array[1];
294 x264_pps_t pps_array[1];
298 int (*dequant4_mf[4])[4][4]; /* [4][6][4][4] */
299 int (*dequant8_mf[2])[8][8]; /* [2][6][8][8] */
300 int (*quant4_mf[4])[4][4]; /* [4][6][4][4] */
301 int (*quant8_mf[2])[8][8]; /* [2][6][8][8] */
302 int (*unquant4_mf[4])[16]; /* [4][52][16] */
303 int (*unquant8_mf[2])[64]; /* [2][52][64] */
305 uint32_t nr_residual_sum[2][64];
306 uint32_t nr_offset[2][64];
307 uint32_t nr_count[2];
310 x264_slice_header_t sh;
317 /* Frames to be encoded (whose types have been decided) */
318 x264_frame_t *current[X264_BFRAME_MAX+3];
319 /* Temporary buffer (frames types not yet decided) */
320 x264_frame_t *next[X264_BFRAME_MAX+3];
322 x264_frame_t *unused[X264_BFRAME_MAX+3];
323 /* For adaptive B decision */
324 x264_frame_t *last_nonb;
326 /* frames used for reference +1 for decoding + sentinels */
327 x264_frame_t *reference[16+2+1+2];
329 int i_last_idr; /* Frame number of the last IDR */
331 int i_input; /* Number of input frames already accepted */
333 int i_max_dpb; /* Number of frames allocated in the decoded picture buffer */
336 int i_delay; /* Number of frames buffered for B reordering */
337 int b_have_lowres; /* Whether 1/2 resolution luma planes are being used */
340 /* current frame being encoded */
343 /* frame being reconstructed */
346 /* references lists */
348 x264_frame_t *fref0[16+3]; /* ref list 0 */
350 x264_frame_t *fref1[16+3]; /* ref list 1 */
351 int b_ref_reorder[2];
355 /* Current MB DCT coeffs */
358 DECLARE_ALIGNED( int, luma16x16_dc[16], 16 );
359 DECLARE_ALIGNED( int, chroma_dc[2][4], 16 );
360 // FIXME merge with union
361 DECLARE_ALIGNED( int, luma8x8[4][64], 16 );
364 DECLARE_ALIGNED( int, residual_ac[15], 16 );
365 DECLARE_ALIGNED( int, luma4x4[16], 16 );
369 /* MB table and cache for current frame/mb */
372 int i_mb_count; /* number of mbs in a frame */
386 /* Search parameters */
391 int b_noise_reduction;
393 /* Allowed qpel MV range to stay within the picture + emulated edge pixels */
396 /* Subpel MV range for motion search.
397 * same mv_min/max but includes levels' i_mv_range. */
400 /* Fullpel MV range for motion search */
404 /* neighboring MBs */
405 unsigned int i_neighbour;
406 unsigned int i_neighbour8[4]; /* neighbours of each 8x8 or 4x4 block that are available */
407 unsigned int i_neighbour4[16]; /* at the time the block is coded */
410 int i_mb_type_topleft;
411 int i_mb_type_topright;
414 int8_t *type; /* mb type */
415 int8_t *qp; /* mb qp */
416 int16_t *cbp; /* mb cbp: 0x0?: luma, 0x?0: chroma, 0x100: luma dc, 0x0200 and 0x0400: chroma dc (all set for PCM)*/
417 int8_t (*intra4x4_pred_mode)[7]; /* intra4x4 pred mode. for non I4x4 set to I_PRED_4x4_DC(2) */
418 uint8_t (*non_zero_count)[16+4+4]; /* nzc. for I_PCM set to 16 */
419 int8_t *chroma_pred_mode; /* chroma_pred_mode. cabac only. for non intra I_PRED_CHROMA_DC(0) */
420 int16_t (*mv[2])[2]; /* mb mv. set to 0 for intra mb */
421 int16_t (*mvd[2])[2]; /* mb mv difference with predict. set to 0 if intra. cabac only */
422 int8_t *ref[2]; /* mb ref. set to -1 if non used (intra or Lx only) */
423 int16_t (*mvr[2][16])[2]; /* 16x16 mv for each possible ref */
424 int8_t *skipbp; /* block pattern for SKIP or DIRECT (sub)mbs. B-frames + cabac only */
425 int8_t *mb_transform_size; /* transform_size_8x8_flag of each mb */
430 int i_sub_partition[4];
436 int i_intra16x16_pred_mode;
437 int i_chroma_pred_mode;
441 /* space for p_fenc and p_fdec */
442 #define FENC_STRIDE 16
443 #define FDEC_STRIDE 32
444 DECLARE_ALIGNED( uint8_t, fenc_buf[24*FENC_STRIDE], 16 );
445 DECLARE_ALIGNED( uint8_t, fdec_buf[27*FDEC_STRIDE], 16 );
447 /* pointer over mb of the frame to be compressed */
450 /* pointer over mb of the frame to be reconstrucated */
453 /* pointer over mb of the references */
454 uint8_t *p_fref[2][16][4+2]; /* last: lN, lH, lV, lHV, cU, cV */
455 uint16_t *p_integral[2][16];
464 /* real intra4x4_pred_mode if I_4X4 or I_8X8, I_PRED_4x4_DC if mb available, -1 if not */
465 int intra4x4_pred_mode[X264_SCAN8_SIZE];
467 /* i_non_zero_count if availble else 0x80 */
468 int non_zero_count[X264_SCAN8_SIZE];
470 /* -1 if unused, -2 if unavaible */
471 int8_t ref[2][X264_SCAN8_SIZE];
473 /* 0 if non avaible */
474 int16_t mv[2][X264_SCAN8_SIZE][2];
475 int16_t mvd[2][X264_SCAN8_SIZE][2];
477 /* 1 if SKIP or DIRECT. set only for B-frames + CABAC */
478 int8_t skip[X264_SCAN8_SIZE];
480 int16_t direct_mv[2][X264_SCAN8_SIZE][2];
481 int8_t direct_ref[2][X264_SCAN8_SIZE];
483 /* number of neighbors (top and left) that used 8x8 dct */
484 int i_neighbour_transform_size;
485 int b_transform_8x8_allowed;
489 int i_qp; /* current qp */
490 int i_last_qp; /* last qp */
491 int i_last_dqp; /* last delta qp */
492 int b_variable_qp; /* whether qp is allowed to vary per macroblock */
494 int b_direct_auto_read; /* take stats for --direct auto from the 2pass log */
495 int b_direct_auto_write; /* analyse direct modes, to use and/or save */
497 /* B_direct and weighted prediction */
498 int dist_scale_factor[16][16];
499 int bipred_weight[16][16];
500 /* maps fref1[0]'s ref indices into the current list0 */
501 int map_col_to_list0_buf[2]; // for negative indices
502 int map_col_to_list0[16];
505 /* rate control encoding only */
506 x264_ratecontrol_t *rc;
511 /* Current frame stats */
514 /* Headers bits (MV+Ref+MB Block Type */
516 /* Texture bits (Intra/Predicted) */
526 int i_mb_count_8x8dct[2];
527 int i_mb_count_size[7];
528 int i_mb_count_ref[16];
529 /* Estimated (SATD) cost as Intra/Predicted frame */
530 /* XXX: both omit the cost of MBs coded as P_SKIP */
533 /* Adaptive direct mv pred */
534 int i_direct_score[2];
537 /* Cummulated stats */
540 int i_slice_count[5];
541 int64_t i_slice_size[5];
544 int64_t i_sqe_global[5];
545 float f_psnr_average[5];
546 float f_psnr_mean_y[5];
547 float f_psnr_mean_u[5];
548 float f_psnr_mean_v[5];
550 int64_t i_mb_count[5][19];
551 int64_t i_mb_count_8x8dct[2];
552 int64_t i_mb_count_size[2][7];
553 int64_t i_mb_count_ref[2][16];
555 int i_direct_score[2];
556 int i_direct_frames[2];
560 /* CPU functions dependants */
561 x264_predict_t predict_16x16[4+3];
562 x264_predict_t predict_8x8c[4+3];
563 x264_predict8x8_t predict_8x8[9+3];
564 x264_predict_t predict_4x4[9+3];
566 x264_pixel_function_t pixf;
567 x264_mc_functions_t mc;
568 x264_dct_function_t dctf;
569 x264_csp_function_t csp;
570 x264_quant_function_t quantf;
571 x264_deblock_function_t loopf;
573 /* vlc table for decoding purpose only */
574 x264_vlc_table_t *x264_coeff_token_lookup[5];
575 x264_vlc_table_t *x264_level_prefix_lookup;
576 x264_vlc_table_t *x264_total_zeros_lookup[15];
577 x264_vlc_table_t *x264_total_zeros_dc_lookup[3];
578 x264_vlc_table_t *x264_run_before_lookup[7];
581 struct visualize_t *visualize;
585 // included at the end because it needs x264_t
586 #include "macroblock.h"