summaryrefslogtreecommitdiff
path: root/libavcodec/libxavs.c (plain)
blob: f257e5584d91ff229b30510aa239cc5e592428ca
1/*
2 * AVS encoding using the xavs library
3 * Copyright (C) 2010 Amanda, Y.N. Wu <amanda11192003@gmail.com>
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22#include <stdio.h>
23#include <stdlib.h>
24#include <string.h>
25#include <math.h>
26#include <stdint.h>
27#include <float.h>
28#include <xavs.h>
29#include "avcodec.h"
30#include "internal.h"
31#include "libavutil/internal.h"
32#include "libavutil/mem.h"
33#include "libavutil/opt.h"
34
35#define END_OF_STREAM 0x001
36
37#define XAVS_PART_I8X8 0x002 /* Analyze i8x8 (requires 8x8 transform) */
38#define XAVS_PART_P8X8 0x010 /* Analyze p16x8, p8x16 and p8x8 */
39#define XAVS_PART_B8X8 0x100 /* Analyze b16x8, b*/
40
41typedef struct XavsContext {
42 AVClass *class;
43 xavs_param_t params;
44 xavs_t *enc;
45 xavs_picture_t pic;
46 uint8_t *sei;
47 int sei_size;
48 int end_of_stream;
49 float crf;
50 int cqp;
51 int b_bias;
52 float cplxblur;
53 int direct_pred;
54 int aud;
55 int fast_pskip;
56 int motion_est;
57 int mbtree;
58 int mixed_refs;
59 int b_frame_strategy;
60 int chroma_offset;
61 int scenechange_threshold;
62 int noise_reduction;
63
64 int64_t *pts_buffer;
65 int out_frame_count;
66} XavsContext;
67
68static void XAVS_log(void *p, int level, const char *fmt, va_list args)
69{
70 static const int level_map[] = {
71 [XAVS_LOG_ERROR] = AV_LOG_ERROR,
72 [XAVS_LOG_WARNING] = AV_LOG_WARNING,
73 [XAVS_LOG_INFO] = AV_LOG_INFO,
74 [XAVS_LOG_DEBUG] = AV_LOG_DEBUG
75 };
76
77 if (level < 0 || level > XAVS_LOG_DEBUG)
78 return;
79
80 av_vlog(p, level_map[level], fmt, args);
81}
82
83static int encode_nals(AVCodecContext *ctx, AVPacket *pkt,
84 xavs_nal_t *nals, int nnal)
85{
86 XavsContext *x4 = ctx->priv_data;
87 uint8_t *p;
88 int i, s, ret, size = x4->sei_size + AV_INPUT_BUFFER_MIN_SIZE;
89
90 if (!nnal)
91 return 0;
92
93 for (i = 0; i < nnal; i++)
94 size += nals[i].i_payload;
95
96 if ((ret = ff_alloc_packet2(ctx, pkt, size, 0)) < 0)
97 return ret;
98 p = pkt->data;
99
100 /* Write the SEI as part of the first frame. */
101 if (x4->sei_size > 0 && nnal > 0) {
102 memcpy(p, x4->sei, x4->sei_size);
103 p += x4->sei_size;
104 x4->sei_size = 0;
105 }
106
107 for (i = 0; i < nnal; i++) {
108 s = xavs_nal_encode(p, &size, 1, nals + i);
109 if (s < 0)
110 return -1;
111 p += s;
112 }
113 pkt->size = p - pkt->data;
114
115 return 1;
116}
117
118static int XAVS_frame(AVCodecContext *avctx, AVPacket *pkt,
119 const AVFrame *frame, int *got_packet)
120{
121 XavsContext *x4 = avctx->priv_data;
122 xavs_nal_t *nal;
123 int nnal, i, ret;
124 xavs_picture_t pic_out;
125 int pict_type;
126
127 x4->pic.img.i_csp = XAVS_CSP_I420;
128 x4->pic.img.i_plane = 3;
129
130 if (frame) {
131 for (i = 0; i < 3; i++) {
132 x4->pic.img.plane[i] = frame->data[i];
133 x4->pic.img.i_stride[i] = frame->linesize[i];
134 }
135
136 x4->pic.i_pts = frame->pts;
137 x4->pic.i_type = XAVS_TYPE_AUTO;
138 x4->pts_buffer[avctx->frame_number % (avctx->max_b_frames+1)] = frame->pts;
139 }
140
141 if (xavs_encoder_encode(x4->enc, &nal, &nnal,
142 frame? &x4->pic: NULL, &pic_out) < 0)
143 return -1;
144
145 ret = encode_nals(avctx, pkt, nal, nnal);
146
147 if (ret < 0)
148 return -1;
149
150 if (!ret) {
151 if (!frame && !(x4->end_of_stream)) {
152 if ((ret = ff_alloc_packet2(avctx, pkt, 4, 0)) < 0)
153 return ret;
154
155 pkt->data[0] = 0x0;
156 pkt->data[1] = 0x0;
157 pkt->data[2] = 0x01;
158 pkt->data[3] = 0xb1;
159 pkt->dts = 2*x4->pts_buffer[(x4->out_frame_count-1)%(avctx->max_b_frames+1)] -
160 x4->pts_buffer[(x4->out_frame_count-2)%(avctx->max_b_frames+1)];
161 x4->end_of_stream = END_OF_STREAM;
162 *got_packet = 1;
163 }
164 return 0;
165 }
166
167#if FF_API_CODED_FRAME
168FF_DISABLE_DEPRECATION_WARNINGS
169 avctx->coded_frame->pts = pic_out.i_pts;
170FF_ENABLE_DEPRECATION_WARNINGS
171#endif
172 pkt->pts = pic_out.i_pts;
173 if (avctx->has_b_frames) {
174 if (!x4->out_frame_count)
175 pkt->dts = pkt->pts - (x4->pts_buffer[1] - x4->pts_buffer[0]);
176 else
177 pkt->dts = x4->pts_buffer[(x4->out_frame_count-1)%(avctx->max_b_frames+1)];
178 } else
179 pkt->dts = pkt->pts;
180
181 switch (pic_out.i_type) {
182 case XAVS_TYPE_IDR:
183 case XAVS_TYPE_I:
184 pict_type = AV_PICTURE_TYPE_I;
185 break;
186 case XAVS_TYPE_P:
187 pict_type = AV_PICTURE_TYPE_P;
188 break;
189 case XAVS_TYPE_B:
190 case XAVS_TYPE_BREF:
191 pict_type = AV_PICTURE_TYPE_B;
192 break;
193 default:
194 pict_type = AV_PICTURE_TYPE_NONE;
195 }
196#if FF_API_CODED_FRAME
197FF_DISABLE_DEPRECATION_WARNINGS
198 avctx->coded_frame->pict_type = pict_type;
199FF_ENABLE_DEPRECATION_WARNINGS
200#endif
201
202 /* There is no IDR frame in AVS JiZhun */
203 /* Sequence header is used as a flag */
204 if (pic_out.i_type == XAVS_TYPE_I) {
205#if FF_API_CODED_FRAME
206FF_DISABLE_DEPRECATION_WARNINGS
207 avctx->coded_frame->key_frame = 1;
208FF_ENABLE_DEPRECATION_WARNINGS
209#endif
210 pkt->flags |= AV_PKT_FLAG_KEY;
211 }
212
213#if FF_API_CODED_FRAME
214FF_DISABLE_DEPRECATION_WARNINGS
215 avctx->coded_frame->quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
216FF_ENABLE_DEPRECATION_WARNINGS
217#endif
218
219 ff_side_data_set_encoder_stats(pkt, (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA, NULL, 0, pict_type);
220
221 x4->out_frame_count++;
222 *got_packet = ret;
223 return 0;
224}
225
226static av_cold int XAVS_close(AVCodecContext *avctx)
227{
228 XavsContext *x4 = avctx->priv_data;
229
230 av_freep(&avctx->extradata);
231 av_freep(&x4->sei);
232 av_freep(&x4->pts_buffer);
233
234 if (x4->enc)
235 xavs_encoder_close(x4->enc);
236
237 return 0;
238}
239
240static av_cold int XAVS_init(AVCodecContext *avctx)
241{
242 XavsContext *x4 = avctx->priv_data;
243
244 x4->sei_size = 0;
245 xavs_param_default(&x4->params);
246
247 x4->params.pf_log = XAVS_log;
248 x4->params.p_log_private = avctx;
249 x4->params.i_keyint_max = avctx->gop_size;
250 if (avctx->bit_rate) {
251 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
252 x4->params.rc.i_rc_method = XAVS_RC_ABR;
253 }
254 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
255 x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
256 x4->params.rc.b_stat_write = avctx->flags & AV_CODEC_FLAG_PASS1;
257 if (avctx->flags & AV_CODEC_FLAG_PASS2) {
258 x4->params.rc.b_stat_read = 1;
259 } else {
260 if (x4->crf >= 0) {
261 x4->params.rc.i_rc_method = XAVS_RC_CRF;
262 x4->params.rc.f_rf_constant = x4->crf;
263 } else if (x4->cqp >= 0) {
264 x4->params.rc.i_rc_method = XAVS_RC_CQP;
265 x4->params.rc.i_qp_constant = x4->cqp;
266 }
267 }
268
269 if (x4->aud >= 0)
270 x4->params.b_aud = x4->aud;
271 if (x4->mbtree >= 0)
272 x4->params.rc.b_mb_tree = x4->mbtree;
273 if (x4->direct_pred >= 0)
274 x4->params.analyse.i_direct_mv_pred = x4->direct_pred;
275 if (x4->fast_pskip >= 0)
276 x4->params.analyse.b_fast_pskip = x4->fast_pskip;
277 if (x4->motion_est >= 0)
278 x4->params.analyse.i_me_method = x4->motion_est;
279 if (x4->mixed_refs >= 0)
280 x4->params.analyse.b_mixed_references = x4->mixed_refs;
281 if (x4->b_bias != INT_MIN)
282 x4->params.i_bframe_bias = x4->b_bias;
283 if (x4->cplxblur >= 0)
284 x4->params.rc.f_complexity_blur = x4->cplxblur;
285
286#if FF_API_MOTION_EST
287FF_DISABLE_DEPRECATION_WARNINGS
288 if (x4->motion_est < 0) {
289 switch (avctx->me_method) {
290 case ME_EPZS:
291 x4->params.analyse.i_me_method = XAVS_ME_DIA;
292 break;
293 case ME_HEX:
294 x4->params.analyse.i_me_method = XAVS_ME_HEX;
295 break;
296 case ME_UMH:
297 x4->params.analyse.i_me_method = XAVS_ME_UMH;
298 break;
299 case ME_FULL:
300 x4->params.analyse.i_me_method = XAVS_ME_ESA;
301 break;
302 case ME_TESA:
303 x4->params.analyse.i_me_method = XAVS_ME_TESA;
304 break;
305 default:
306 x4->params.analyse.i_me_method = XAVS_ME_HEX;
307 }
308 }
309FF_ENABLE_DEPRECATION_WARNINGS
310#endif
311
312 x4->params.i_bframe = avctx->max_b_frames;
313 /* cabac is not included in AVS JiZhun Profile */
314 x4->params.b_cabac = 0;
315
316#if FF_API_PRIVATE_OPT
317FF_DISABLE_DEPRECATION_WARNINGS
318 if (avctx->b_frame_strategy)
319 x4->b_frame_strategy = avctx->b_frame_strategy;
320FF_ENABLE_DEPRECATION_WARNINGS
321#endif
322
323 x4->params.i_bframe_adaptive = x4->b_frame_strategy;
324
325 avctx->has_b_frames = !!avctx->max_b_frames;
326
327 /* AVS doesn't allow B picture as reference */
328 /* The max allowed reference frame number of B is 2 */
329 x4->params.i_keyint_min = avctx->keyint_min;
330 if (x4->params.i_keyint_min > x4->params.i_keyint_max)
331 x4->params.i_keyint_min = x4->params.i_keyint_max;
332
333#if FF_API_PRIVATE_OPT
334FF_DISABLE_DEPRECATION_WARNINGS
335 if (avctx->scenechange_threshold)
336 x4->scenechange_threshold = avctx->scenechange_threshold;
337FF_ENABLE_DEPRECATION_WARNINGS
338#endif
339
340 x4->params.i_scenecut_threshold = x4->scenechange_threshold;
341
342 // x4->params.b_deblocking_filter = avctx->flags & AV_CODEC_FLAG_LOOP_FILTER;
343
344 x4->params.rc.i_qp_min = avctx->qmin;
345 x4->params.rc.i_qp_max = avctx->qmax;
346 x4->params.rc.i_qp_step = avctx->max_qdiff;
347
348 x4->params.rc.f_qcompress = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
349 x4->params.rc.f_qblur = avctx->qblur; /* temporally blur quants */
350
351 x4->params.i_frame_reference = avctx->refs;
352
353 x4->params.i_width = avctx->width;
354 x4->params.i_height = avctx->height;
355 x4->params.vui.i_sar_width = avctx->sample_aspect_ratio.num;
356 x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
357 /* This is only used for counting the fps */
358 x4->params.i_fps_num = avctx->time_base.den;
359 x4->params.i_fps_den = avctx->time_base.num;
360 x4->params.analyse.inter = XAVS_ANALYSE_I8x8 |XAVS_ANALYSE_PSUB16x16| XAVS_ANALYSE_BSUB16x16;
361
362 x4->params.analyse.i_me_range = avctx->me_range;
363 x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
364
365 x4->params.analyse.b_chroma_me = avctx->me_cmp & FF_CMP_CHROMA;
366 /* AVS P2 only enables 8x8 transform */
367 x4->params.analyse.b_transform_8x8 = 1; //avctx->flags2 & AV_CODEC_FLAG2_8X8DCT;
368
369 x4->params.analyse.i_trellis = avctx->trellis;
370
371#if FF_API_PRIVATE_OPT
372 FF_DISABLE_DEPRECATION_WARNINGS
373 if (avctx->noise_reduction >= 0)
374 x4->noise_reduction = avctx->noise_reduction;
375 FF_ENABLE_DEPRECATION_WARNINGS
376#endif
377
378 x4->params.analyse.i_noise_reduction = x4->noise_reduction;
379
380 if (avctx->level > 0)
381 x4->params.i_level_idc = avctx->level;
382
383 if (avctx->bit_rate > 0)
384 x4->params.rc.f_rate_tolerance =
385 (float)avctx->bit_rate_tolerance / avctx->bit_rate;
386
387 if ((avctx->rc_buffer_size) &&
388 (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
389 x4->params.rc.f_vbv_buffer_init =
390 (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
391 } else
392 x4->params.rc.f_vbv_buffer_init = 0.9;
393
394 /* TAG:do we have MB tree RC method */
395 /* what is the RC method we are now using? Default NO */
396 x4->params.rc.f_ip_factor = 1 / fabs(avctx->i_quant_factor);
397 x4->params.rc.f_pb_factor = avctx->b_quant_factor;
398
399#if FF_API_PRIVATE_OPT
400FF_DISABLE_DEPRECATION_WARNINGS
401 if (avctx->chromaoffset)
402 x4->chroma_offset = avctx->chromaoffset;
403FF_ENABLE_DEPRECATION_WARNINGS
404#endif
405
406 x4->params.analyse.i_chroma_qp_offset = x4->chroma_offset;
407
408 x4->params.analyse.b_psnr = avctx->flags & AV_CODEC_FLAG_PSNR;
409 x4->params.i_log_level = XAVS_LOG_DEBUG;
410 x4->params.i_threads = avctx->thread_count;
411 x4->params.b_interlaced = avctx->flags & AV_CODEC_FLAG_INTERLACED_DCT;
412
413 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
414 x4->params.b_repeat_headers = 0;
415
416 x4->enc = xavs_encoder_open(&x4->params);
417 if (!x4->enc)
418 return -1;
419
420 if (!(x4->pts_buffer = av_mallocz_array((avctx->max_b_frames+1), sizeof(*x4->pts_buffer))))
421 return AVERROR(ENOMEM);
422
423 /* TAG: Do we have GLOBAL HEADER in AVS */
424 /* We Have PPS and SPS in AVS */
425 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER && 0) {
426 xavs_nal_t *nal;
427 int nnal, s, i, size;
428 uint8_t *p;
429
430 s = xavs_encoder_headers(x4->enc, &nal, &nnal);
431
432 avctx->extradata = p = av_malloc(s);
433 for (i = 0; i < nnal; i++) {
434 /* Don't put the SEI in extradata. */
435 if (nal[i].i_type == NAL_SEI) {
436 x4->sei = av_malloc( 5 + nal[i].i_payload * 4 / 3 );
437 if (xavs_nal_encode(x4->sei, &x4->sei_size, 1, nal + i) < 0)
438 return -1;
439
440 continue;
441 }
442 size = xavs_nal_encode(p, &s, 1, nal + i);
443 if (size < 0)
444 return -1;
445 p += size;
446 }
447 avctx->extradata_size = p - avctx->extradata;
448 }
449 return 0;
450}
451
452#define OFFSET(x) offsetof(XavsContext, x)
453#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
454static const AVOption options[] = {
455 { "crf", "Select the quality for constant quality mode", OFFSET(crf), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE },
456 { "qp", "Constant quantization parameter rate control method",OFFSET(cqp), AV_OPT_TYPE_INT, {.i64 = -1 }, -1, INT_MAX, VE },
457 { "b-bias", "Influences how often B-frames are used", OFFSET(b_bias), AV_OPT_TYPE_INT, {.i64 = INT_MIN}, INT_MIN, INT_MAX, VE },
458 { "cplxblur", "Reduce fluctuations in QP (before curve compression)", OFFSET(cplxblur), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE},
459 { "direct-pred", "Direct MV prediction mode", OFFSET(direct_pred), AV_OPT_TYPE_INT, {.i64 = -1 }, -1, INT_MAX, VE, "direct-pred" },
460 { "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_DIRECT_PRED_NONE }, 0, 0, VE, "direct-pred" },
461 { "spatial", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_DIRECT_PRED_SPATIAL }, 0, 0, VE, "direct-pred" },
462 { "temporal", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_DIRECT_PRED_TEMPORAL }, 0, 0, VE, "direct-pred" },
463 { "auto", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_DIRECT_PRED_AUTO }, 0, 0, VE, "direct-pred" },
464 { "aud", "Use access unit delimiters.", OFFSET(aud), AV_OPT_TYPE_BOOL, {.i64 = -1 }, -1, 1, VE},
465 { "mbtree", "Use macroblock tree ratecontrol.", OFFSET(mbtree), AV_OPT_TYPE_BOOL, {.i64 = -1 }, -1, 1, VE},
466 { "mixed-refs", "One reference per partition, as opposed to one reference per macroblock", OFFSET(mixed_refs), AV_OPT_TYPE_BOOL, {.i64 = -1}, -1, 1, VE },
467 { "fast-pskip", NULL, OFFSET(fast_pskip), AV_OPT_TYPE_BOOL, {.i64 = -1 }, -1, 1, VE},
468 { "motion-est", "Set motion estimation method", OFFSET(motion_est), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, XAVS_ME_TESA, VE, "motion-est"},
469 { "dia", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_ME_DIA }, INT_MIN, INT_MAX, VE, "motion-est" },
470 { "hex", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_ME_HEX }, INT_MIN, INT_MAX, VE, "motion-est" },
471 { "umh", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_ME_UMH }, INT_MIN, INT_MAX, VE, "motion-est" },
472 { "esa", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_ME_ESA }, INT_MIN, INT_MAX, VE, "motion-est" },
473 { "tesa", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = XAVS_ME_TESA }, INT_MIN, INT_MAX, VE, "motion-est" },
474 { "b_strategy", "Strategy to choose between I/P/B-frames", OFFSET(b_frame_strategy), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 2, VE},
475 { "chromaoffset", "QP difference between chroma and luma", OFFSET(chroma_offset), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, VE},
476 { "sc_threshold", "Scene change threshold", OFFSET(scenechange_threshold), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, INT_MAX, VE},
477 { "noise_reduction", "Noise reduction", OFFSET(noise_reduction), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, INT_MAX, VE},
478
479 { NULL },
480};
481
482static const AVClass xavs_class = {
483 .class_name = "libxavs",
484 .item_name = av_default_item_name,
485 .option = options,
486 .version = LIBAVUTIL_VERSION_INT,
487};
488
489static const AVCodecDefault xavs_defaults[] = {
490 { "b", "0" },
491 { NULL },
492};
493
494AVCodec ff_libxavs_encoder = {
495 .name = "libxavs",
496 .long_name = NULL_IF_CONFIG_SMALL("libxavs Chinese AVS (Audio Video Standard)"),
497 .type = AVMEDIA_TYPE_VIDEO,
498 .id = AV_CODEC_ID_CAVS,
499 .priv_data_size = sizeof(XavsContext),
500 .init = XAVS_init,
501 .encode2 = XAVS_frame,
502 .close = XAVS_close,
503 .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_AUTO_THREADS,
504 .pix_fmts = (const enum AVPixelFormat[]) { AV_PIX_FMT_YUV420P, AV_PIX_FMT_NONE },
505 .priv_class = &xavs_class,
506 .defaults = xavs_defaults,
507};
508
generated by cgit at 2024-05-17 02:05:48 (GMT)