AOMedia AV1 Codec
aomenc
1/*
2 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
3 *
4 * This source code is subject to the terms of the BSD 2 Clause License and
5 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
6 * was not distributed with this source code in the LICENSE file, you can
7 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
8 * Media Patent License 1.0 was not distributed with this source code in the
9 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
10 */
11
12#include "apps/aomenc.h"
13
14#include "config/aom_config.h"
15
16#include <assert.h>
17#include <limits.h>
18#include <math.h>
19#include <stdarg.h>
20#include <stdio.h>
21#include <stdlib.h>
22#include <string.h>
23
24#if CONFIG_AV1_DECODER
25#include "aom/aom_decoder.h"
26#include "aom/aomdx.h"
27#endif
28
29#include "aom/aom_encoder.h"
30#include "aom/aom_integer.h"
31#include "aom/aomcx.h"
32#include "aom_dsp/aom_dsp_common.h"
33#include "aom_ports/aom_timer.h"
34#include "aom_ports/mem_ops.h"
35#include "common/args.h"
36#include "common/ivfenc.h"
37#include "common/tools_common.h"
38#include "common/warnings.h"
39
40#if CONFIG_WEBM_IO
41#include "common/webmenc.h"
42#endif
43
44#include "common/y4minput.h"
45#include "examples/encoder_util.h"
46#include "stats/aomstats.h"
47#include "stats/rate_hist.h"
48
49#if CONFIG_LIBYUV
50#include <libyuv/scale.h>
51#endif
52
53/* Swallow warnings about unused results of fread/fwrite */
54static size_t wrap_fread(void *ptr, size_t size, size_t nmemb, FILE *stream) {
55 return fread(ptr, size, nmemb, stream);
56}
57#define fread wrap_fread
58
59static size_t wrap_fwrite(const void *ptr, size_t size, size_t nmemb,
60 FILE *stream) {
61 return fwrite(ptr, size, nmemb, stream);
62}
63#define fwrite wrap_fwrite
64
65static const char *exec_name;
66
67static AOM_TOOLS_FORMAT_PRINTF(3, 0) void warn_or_exit_on_errorv(
68 aom_codec_ctx_t *ctx, int fatal, const char *s, va_list ap) {
69 if (ctx->err) {
70 const char *detail = aom_codec_error_detail(ctx);
71
72 vfprintf(stderr, s, ap);
73 fprintf(stderr, ": %s\n", aom_codec_error(ctx));
74
75 if (detail) fprintf(stderr, " %s\n", detail);
76
77 if (fatal) {
79 exit(EXIT_FAILURE);
80 }
81 }
82}
83
84static AOM_TOOLS_FORMAT_PRINTF(2,
85 3) void ctx_exit_on_error(aom_codec_ctx_t *ctx,
86 const char *s, ...) {
87 va_list ap;
88
89 va_start(ap, s);
90 warn_or_exit_on_errorv(ctx, 1, s, ap);
91 va_end(ap);
92}
93
94static AOM_TOOLS_FORMAT_PRINTF(3, 4) void warn_or_exit_on_error(
95 aom_codec_ctx_t *ctx, int fatal, const char *s, ...) {
96 va_list ap;
97
98 va_start(ap, s);
99 warn_or_exit_on_errorv(ctx, fatal, s, ap);
100 va_end(ap);
101}
102
103static int read_frame(struct AvxInputContext *input_ctx, aom_image_t *img) {
104 FILE *f = input_ctx->file;
105 y4m_input *y4m = &input_ctx->y4m;
106 int shortread = 0;
107
108 if (input_ctx->file_type == FILE_TYPE_Y4M) {
109 if (y4m_input_fetch_frame(y4m, f, img) < 1) return 0;
110 } else {
111 shortread = read_yuv_frame(input_ctx, img);
112 }
113
114 return !shortread;
115}
116
117static int file_is_y4m(const char detect[4]) {
118 if (memcmp(detect, "YUV4", 4) == 0) {
119 return 1;
120 }
121 return 0;
122}
123
124static int fourcc_is_ivf(const char detect[4]) {
125 if (memcmp(detect, "DKIF", 4) == 0) {
126 return 1;
127 }
128 return 0;
129}
130
131static const int av1_arg_ctrl_map[] = { AOME_SET_CPUUSED,
220#if CONFIG_DENOISE
223 AV1E_SET_ENABLE_DNL_DENOISING,
224#endif // CONFIG_DENOISE
234#if CONFIG_TUNE_VMAF
236#endif
245 0 };
246
247const arg_def_t *main_args[] = { &g_av1_codec_arg_defs.help,
248 &g_av1_codec_arg_defs.use_cfg,
249 &g_av1_codec_arg_defs.debugmode,
250 &g_av1_codec_arg_defs.outputfile,
251 &g_av1_codec_arg_defs.codecarg,
252 &g_av1_codec_arg_defs.passes,
253 &g_av1_codec_arg_defs.pass_arg,
254 &g_av1_codec_arg_defs.fpf_name,
255 &g_av1_codec_arg_defs.limit,
256 &g_av1_codec_arg_defs.skip,
257 &g_av1_codec_arg_defs.good_dl,
258 &g_av1_codec_arg_defs.rt_dl,
259 &g_av1_codec_arg_defs.ai_dl,
260 &g_av1_codec_arg_defs.quietarg,
261 &g_av1_codec_arg_defs.verbosearg,
262 &g_av1_codec_arg_defs.psnrarg,
263 &g_av1_codec_arg_defs.use_webm,
264 &g_av1_codec_arg_defs.use_ivf,
265 &g_av1_codec_arg_defs.use_obu,
266 &g_av1_codec_arg_defs.q_hist_n,
267 &g_av1_codec_arg_defs.rate_hist_n,
268 &g_av1_codec_arg_defs.disable_warnings,
269 &g_av1_codec_arg_defs.disable_warning_prompt,
270 &g_av1_codec_arg_defs.recontest,
271 NULL };
272
273const arg_def_t *global_args[] = {
274 &g_av1_codec_arg_defs.use_nv12,
275 &g_av1_codec_arg_defs.use_yv12,
276 &g_av1_codec_arg_defs.use_i420,
277 &g_av1_codec_arg_defs.use_i422,
278 &g_av1_codec_arg_defs.use_i444,
279 &g_av1_codec_arg_defs.usage,
280 &g_av1_codec_arg_defs.threads,
281 &g_av1_codec_arg_defs.profile,
282 &g_av1_codec_arg_defs.width,
283 &g_av1_codec_arg_defs.height,
284 &g_av1_codec_arg_defs.forced_max_frame_width,
285 &g_av1_codec_arg_defs.forced_max_frame_height,
286#if CONFIG_WEBM_IO
287 &g_av1_codec_arg_defs.stereo_mode,
288#endif
289 &g_av1_codec_arg_defs.timebase,
290 &g_av1_codec_arg_defs.framerate,
291 &g_av1_codec_arg_defs.global_error_resilient,
292 &g_av1_codec_arg_defs.bitdeptharg,
293 &g_av1_codec_arg_defs.inbitdeptharg,
294 &g_av1_codec_arg_defs.lag_in_frames,
295 &g_av1_codec_arg_defs.large_scale_tile,
296 &g_av1_codec_arg_defs.monochrome,
297 &g_av1_codec_arg_defs.full_still_picture_hdr,
298 &g_av1_codec_arg_defs.use_16bit_internal,
299 &g_av1_codec_arg_defs.save_as_annexb,
300 NULL
301};
302
303const arg_def_t *rc_args[] = { &g_av1_codec_arg_defs.dropframe_thresh,
304 &g_av1_codec_arg_defs.resize_mode,
305 &g_av1_codec_arg_defs.resize_denominator,
306 &g_av1_codec_arg_defs.resize_kf_denominator,
307 &g_av1_codec_arg_defs.superres_mode,
308 &g_av1_codec_arg_defs.superres_denominator,
309 &g_av1_codec_arg_defs.superres_kf_denominator,
310 &g_av1_codec_arg_defs.superres_qthresh,
311 &g_av1_codec_arg_defs.superres_kf_qthresh,
312 &g_av1_codec_arg_defs.end_usage,
313 &g_av1_codec_arg_defs.target_bitrate,
314 &g_av1_codec_arg_defs.min_quantizer,
315 &g_av1_codec_arg_defs.max_quantizer,
316 &g_av1_codec_arg_defs.undershoot_pct,
317 &g_av1_codec_arg_defs.overshoot_pct,
318 &g_av1_codec_arg_defs.buf_sz,
319 &g_av1_codec_arg_defs.buf_initial_sz,
320 &g_av1_codec_arg_defs.buf_optimal_sz,
321 &g_av1_codec_arg_defs.bias_pct,
322 &g_av1_codec_arg_defs.minsection_pct,
323 &g_av1_codec_arg_defs.maxsection_pct,
324 NULL };
325
326const arg_def_t *kf_args[] = { &g_av1_codec_arg_defs.fwd_kf_enabled,
327 &g_av1_codec_arg_defs.kf_min_dist,
328 &g_av1_codec_arg_defs.kf_max_dist,
329 &g_av1_codec_arg_defs.kf_disabled,
330 &g_av1_codec_arg_defs.sframe_dist,
331 &g_av1_codec_arg_defs.sframe_mode,
332 NULL };
333
334// TODO(bohanli): Currently all options are supported by the key & value API.
335// Consider removing the control ID usages?
336const arg_def_t *av1_ctrl_args[] = {
337 &g_av1_codec_arg_defs.cpu_used_av1,
338 &g_av1_codec_arg_defs.auto_altref,
339 &g_av1_codec_arg_defs.sharpness,
340 &g_av1_codec_arg_defs.static_thresh,
341 &g_av1_codec_arg_defs.rowmtarg,
342 &g_av1_codec_arg_defs.fpmtarg,
343 &g_av1_codec_arg_defs.tile_cols,
344 &g_av1_codec_arg_defs.tile_rows,
345 &g_av1_codec_arg_defs.enable_tpl_model,
346 &g_av1_codec_arg_defs.enable_keyframe_filtering,
347 &g_av1_codec_arg_defs.arnr_maxframes,
348 &g_av1_codec_arg_defs.arnr_strength,
349 &g_av1_codec_arg_defs.tune_metric,
350 &g_av1_codec_arg_defs.cq_level,
351 &g_av1_codec_arg_defs.max_intra_rate_pct,
352 &g_av1_codec_arg_defs.max_inter_rate_pct,
353 &g_av1_codec_arg_defs.gf_cbr_boost_pct,
354 &g_av1_codec_arg_defs.lossless,
355 &g_av1_codec_arg_defs.enable_cdef,
356 &g_av1_codec_arg_defs.enable_restoration,
357 &g_av1_codec_arg_defs.enable_rect_partitions,
358 &g_av1_codec_arg_defs.enable_ab_partitions,
359 &g_av1_codec_arg_defs.enable_1to4_partitions,
360 &g_av1_codec_arg_defs.min_partition_size,
361 &g_av1_codec_arg_defs.max_partition_size,
362 &g_av1_codec_arg_defs.enable_dual_filter,
363 &g_av1_codec_arg_defs.enable_chroma_deltaq,
364 &g_av1_codec_arg_defs.enable_intra_edge_filter,
365 &g_av1_codec_arg_defs.enable_order_hint,
366 &g_av1_codec_arg_defs.enable_tx64,
367 &g_av1_codec_arg_defs.enable_flip_idtx,
368 &g_av1_codec_arg_defs.enable_rect_tx,
369 &g_av1_codec_arg_defs.enable_dist_wtd_comp,
370 &g_av1_codec_arg_defs.enable_masked_comp,
371 &g_av1_codec_arg_defs.enable_onesided_comp,
372 &g_av1_codec_arg_defs.enable_interintra_comp,
373 &g_av1_codec_arg_defs.enable_smooth_interintra,
374 &g_av1_codec_arg_defs.enable_diff_wtd_comp,
375 &g_av1_codec_arg_defs.enable_interinter_wedge,
376 &g_av1_codec_arg_defs.enable_interintra_wedge,
377 &g_av1_codec_arg_defs.enable_global_motion,
378 &g_av1_codec_arg_defs.enable_warped_motion,
379 &g_av1_codec_arg_defs.enable_filter_intra,
380 &g_av1_codec_arg_defs.enable_smooth_intra,
381 &g_av1_codec_arg_defs.enable_paeth_intra,
382 &g_av1_codec_arg_defs.enable_cfl_intra,
383 &g_av1_codec_arg_defs.enable_diagonal_intra,
384 &g_av1_codec_arg_defs.force_video_mode,
385 &g_av1_codec_arg_defs.enable_obmc,
386 &g_av1_codec_arg_defs.enable_overlay,
387 &g_av1_codec_arg_defs.enable_palette,
388 &g_av1_codec_arg_defs.enable_intrabc,
389 &g_av1_codec_arg_defs.enable_angle_delta,
390 &g_av1_codec_arg_defs.disable_trellis_quant,
391 &g_av1_codec_arg_defs.enable_qm,
392 &g_av1_codec_arg_defs.qm_min,
393 &g_av1_codec_arg_defs.qm_max,
394 &g_av1_codec_arg_defs.reduced_tx_type_set,
395 &g_av1_codec_arg_defs.use_intra_dct_only,
396 &g_av1_codec_arg_defs.use_inter_dct_only,
397 &g_av1_codec_arg_defs.use_intra_default_tx_only,
398 &g_av1_codec_arg_defs.quant_b_adapt,
399 &g_av1_codec_arg_defs.coeff_cost_upd_freq,
400 &g_av1_codec_arg_defs.mode_cost_upd_freq,
401 &g_av1_codec_arg_defs.mv_cost_upd_freq,
402 &g_av1_codec_arg_defs.frame_parallel_decoding,
403 &g_av1_codec_arg_defs.error_resilient_mode,
404 &g_av1_codec_arg_defs.aq_mode,
405 &g_av1_codec_arg_defs.deltaq_mode,
406 &g_av1_codec_arg_defs.deltaq_strength,
407 &g_av1_codec_arg_defs.deltalf_mode,
408 &g_av1_codec_arg_defs.frame_periodic_boost,
409 &g_av1_codec_arg_defs.noise_sens,
410 &g_av1_codec_arg_defs.tune_content,
411 &g_av1_codec_arg_defs.cdf_update_mode,
412 &g_av1_codec_arg_defs.input_color_primaries,
413 &g_av1_codec_arg_defs.input_transfer_characteristics,
414 &g_av1_codec_arg_defs.input_matrix_coefficients,
415 &g_av1_codec_arg_defs.input_chroma_sample_position,
416 &g_av1_codec_arg_defs.min_gf_interval,
417 &g_av1_codec_arg_defs.max_gf_interval,
418 &g_av1_codec_arg_defs.gf_min_pyr_height,
419 &g_av1_codec_arg_defs.gf_max_pyr_height,
420 &g_av1_codec_arg_defs.superblock_size,
421 &g_av1_codec_arg_defs.num_tg,
422 &g_av1_codec_arg_defs.mtu_size,
423 &g_av1_codec_arg_defs.timing_info,
424 &g_av1_codec_arg_defs.film_grain_test,
425 &g_av1_codec_arg_defs.film_grain_table,
426#if CONFIG_DENOISE
427 &g_av1_codec_arg_defs.denoise_noise_level,
428 &g_av1_codec_arg_defs.denoise_block_size,
429 &g_av1_codec_arg_defs.enable_dnl_denoising,
430#endif // CONFIG_DENOISE
431 &g_av1_codec_arg_defs.max_reference_frames,
432 &g_av1_codec_arg_defs.reduced_reference_set,
433 &g_av1_codec_arg_defs.enable_ref_frame_mvs,
434 &g_av1_codec_arg_defs.target_seq_level_idx,
435 &g_av1_codec_arg_defs.set_tier_mask,
436 &g_av1_codec_arg_defs.set_min_cr,
437 &g_av1_codec_arg_defs.vbr_corpus_complexity_lap,
438 &g_av1_codec_arg_defs.input_chroma_subsampling_x,
439 &g_av1_codec_arg_defs.input_chroma_subsampling_y,
440#if CONFIG_TUNE_VMAF
441 &g_av1_codec_arg_defs.vmaf_model_path,
442#endif
443 &g_av1_codec_arg_defs.dv_cost_upd_freq,
444 &g_av1_codec_arg_defs.partition_info_path,
445 &g_av1_codec_arg_defs.enable_rate_guide_deltaq,
446 &g_av1_codec_arg_defs.rate_distribution_info,
447 &g_av1_codec_arg_defs.enable_directional_intra,
448 &g_av1_codec_arg_defs.enable_tx_size_search,
449 &g_av1_codec_arg_defs.loopfilter_control,
450 &g_av1_codec_arg_defs.auto_intra_tools_off,
451 NULL,
452};
453
454const arg_def_t *av1_key_val_args[] = {
455 &g_av1_codec_arg_defs.passes,
456 &g_av1_codec_arg_defs.two_pass_output,
457 &g_av1_codec_arg_defs.second_pass_log,
458 &g_av1_codec_arg_defs.fwd_kf_dist,
459 &g_av1_codec_arg_defs.strict_level_conformance,
460 &g_av1_codec_arg_defs.sb_qp_sweep,
461 &g_av1_codec_arg_defs.dist_metric,
462 &g_av1_codec_arg_defs.kf_max_pyr_height,
463 NULL,
464};
465
466static const arg_def_t *no_args[] = { NULL };
467
468static void show_help(FILE *fout, int shorthelp) {
469 fprintf(fout, "Usage: %s <options> -o dst_filename src_filename\n",
470 exec_name);
471
472 if (shorthelp) {
473 fprintf(fout, "Use --help to see the full list of options.\n");
474 return;
475 }
476
477 fprintf(fout, "\nOptions:\n");
478 arg_show_usage(fout, main_args);
479 fprintf(fout, "\nEncoder Global Options:\n");
480 arg_show_usage(fout, global_args);
481 fprintf(fout, "\nRate Control Options:\n");
482 arg_show_usage(fout, rc_args);
483 fprintf(fout, "\nKeyframe Placement Options:\n");
484 arg_show_usage(fout, kf_args);
485#if CONFIG_AV1_ENCODER
486 fprintf(fout, "\nAV1 Specific Options:\n");
487 arg_show_usage(fout, av1_ctrl_args);
488 arg_show_usage(fout, av1_key_val_args);
489#endif
490 fprintf(fout,
491 "\nStream timebase (--timebase):\n"
492 " The desired precision of timestamps in the output, expressed\n"
493 " in fractional seconds. Default is 1/1000.\n");
494 fprintf(fout, "\nIncluded encoders:\n\n");
495
496 const int num_encoder = get_aom_encoder_count();
497 for (int i = 0; i < num_encoder; ++i) {
498 aom_codec_iface_t *encoder = get_aom_encoder_by_index(i);
499 const char *defstr = (i == (num_encoder - 1)) ? "(default)" : "";
500 fprintf(fout, " %-6s - %s %s\n", get_short_name_by_aom_encoder(encoder),
501 aom_codec_iface_name(encoder), defstr);
502 }
503 fprintf(fout, "\n ");
504 fprintf(fout, "Use --codec to switch to a non-default encoder.\n\n");
505}
506
507void usage_exit(void) {
508 show_help(stderr, 1);
509 exit(EXIT_FAILURE);
510}
511
512#if CONFIG_AV1_ENCODER
513#define ARG_CTRL_CNT_MAX NELEMENTS(av1_arg_ctrl_map)
514#define ARG_KEY_VAL_CNT_MAX NELEMENTS(av1_key_val_args)
515#endif
516
517#if !CONFIG_WEBM_IO
518typedef int stereo_format_t;
519struct WebmOutputContext {
520 int debug;
521};
522#endif
523
524/* Per-stream configuration */
525struct stream_config {
526 struct aom_codec_enc_cfg cfg;
527 const char *out_fn;
528 const char *stats_fn;
529 stereo_format_t stereo_fmt;
530 int arg_ctrls[ARG_CTRL_CNT_MAX][2];
531 int arg_ctrl_cnt;
532 const char *arg_key_vals[ARG_KEY_VAL_CNT_MAX][2];
533 int arg_key_val_cnt;
534 int write_webm;
535 const char *film_grain_filename;
536 int write_ivf;
537 // whether to use 16bit internal buffers
538 int use_16bit_internal;
539#if CONFIG_TUNE_VMAF
540 const char *vmaf_model_path;
541#endif
542 const char *partition_info_path;
543 unsigned int enable_rate_guide_deltaq;
544 const char *rate_distribution_info;
545 aom_color_range_t color_range;
546 const char *two_pass_input;
547 const char *two_pass_output;
548 int two_pass_width;
549 int two_pass_height;
550};
551
552struct stream_state {
553 int index;
554 struct stream_state *next;
555 struct stream_config config;
556 FILE *file;
557 struct rate_hist *rate_hist;
558 struct WebmOutputContext webm_ctx;
559 uint64_t psnr_sse_total[2];
560 uint64_t psnr_samples_total[2];
561 double psnr_totals[2][4];
562 int psnr_count[2];
563 int counts[64];
564 aom_codec_ctx_t encoder;
565 unsigned int frames_out;
566 uint64_t cx_time;
567 size_t nbytes;
568 stats_io_t stats;
569 struct aom_image *img;
570 aom_codec_ctx_t decoder;
571 int mismatch_seen;
572 unsigned int chroma_subsampling_x;
573 unsigned int chroma_subsampling_y;
574 const char *orig_out_fn;
575 unsigned int orig_width;
576 unsigned int orig_height;
577 int orig_write_webm;
578 int orig_write_ivf;
579 char tmp_out_fn[1000];
580};
581
582static void validate_positive_rational(const char *msg,
583 struct aom_rational *rat) {
584 if (rat->den < 0) {
585 rat->num *= -1;
586 rat->den *= -1;
587 }
588
589 if (rat->num < 0) die("Error: %s must be positive\n", msg);
590
591 if (!rat->den) die("Error: %s has zero denominator\n", msg);
592}
593
594static void init_config(cfg_options_t *config) {
595 memset(config, 0, sizeof(cfg_options_t));
596 config->super_block_size = 0; // Dynamic
597 config->max_partition_size = 128;
598 config->min_partition_size = 4;
599 config->disable_trellis_quant = 3;
600}
601
602/* Parses global config arguments into the AvxEncoderConfig. Note that
603 * argv is modified and overwrites all parsed arguments.
604 */
605static void parse_global_config(struct AvxEncoderConfig *global, char ***argv) {
606 char **argi, **argj;
607 struct arg arg;
608 const int num_encoder = get_aom_encoder_count();
609 char **argv_local = (char **)*argv;
610 if (num_encoder < 1) die("Error: no valid encoder available\n");
611
612 /* Initialize default parameters */
613 memset(global, 0, sizeof(*global));
614 global->codec = get_aom_encoder_by_index(num_encoder - 1);
615 global->passes = 0;
616 global->color_type = I420;
617 global->csp = AOM_CSP_UNKNOWN;
618 global->show_psnr = 0;
619
620 int cfg_included = 0;
621 init_config(&global->encoder_config);
622
623 for (argi = argj = argv_local; (*argj = *argi); argi += arg.argv_step) {
624 arg.argv_step = 1;
625
626 if (arg_match(&arg, &g_av1_codec_arg_defs.use_cfg, argi)) {
627 if (!cfg_included) {
628 parse_cfg(arg.val, &global->encoder_config);
629 cfg_included = 1;
630 }
631 } else if (arg_match(&arg, &g_av1_codec_arg_defs.help, argi)) {
632 show_help(stdout, 0);
633 exit(EXIT_SUCCESS);
634 } else if (arg_match(&arg, &g_av1_codec_arg_defs.codecarg, argi)) {
635 global->codec = get_aom_encoder_by_short_name(arg.val);
636 if (!global->codec)
637 die("Error: Unrecognized argument (%s) to --codec\n", arg.val);
638 } else if (arg_match(&arg, &g_av1_codec_arg_defs.passes, argi)) {
639 global->passes = arg_parse_uint(&arg);
640
641 if (global->passes < 1 || global->passes > 3)
642 die("Error: Invalid number of passes (%d)\n", global->passes);
643 } else if (arg_match(&arg, &g_av1_codec_arg_defs.pass_arg, argi)) {
644 global->pass = arg_parse_uint(&arg);
645
646 if (global->pass < 1 || global->pass > 3)
647 die("Error: Invalid pass selected (%d)\n", global->pass);
648 } else if (arg_match(&arg,
649 &g_av1_codec_arg_defs.input_chroma_sample_position,
650 argi)) {
651 global->csp = arg_parse_enum(&arg);
652 /* Flag is used by later code as well, preserve it. */
653 argj++;
654 } else if (arg_match(&arg, &g_av1_codec_arg_defs.usage, argi)) {
655 global->usage = arg_parse_uint(&arg);
656 } else if (arg_match(&arg, &g_av1_codec_arg_defs.good_dl, argi)) {
657 global->usage = AOM_USAGE_GOOD_QUALITY; // Good quality usage
658 } else if (arg_match(&arg, &g_av1_codec_arg_defs.rt_dl, argi)) {
659 global->usage = AOM_USAGE_REALTIME; // Real-time usage
660 } else if (arg_match(&arg, &g_av1_codec_arg_defs.ai_dl, argi)) {
661 global->usage = AOM_USAGE_ALL_INTRA; // All intra usage
662 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_nv12, argi)) {
663 global->color_type = NV12;
664 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_yv12, argi)) {
665 global->color_type = YV12;
666 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_i420, argi)) {
667 global->color_type = I420;
668 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_i422, argi)) {
669 global->color_type = I422;
670 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_i444, argi)) {
671 global->color_type = I444;
672 } else if (arg_match(&arg, &g_av1_codec_arg_defs.quietarg, argi)) {
673 global->quiet = 1;
674 } else if (arg_match(&arg, &g_av1_codec_arg_defs.verbosearg, argi)) {
675 global->verbose = 1;
676 } else if (arg_match(&arg, &g_av1_codec_arg_defs.limit, argi)) {
677 global->limit = arg_parse_uint(&arg);
678 } else if (arg_match(&arg, &g_av1_codec_arg_defs.skip, argi)) {
679 global->skip_frames = arg_parse_uint(&arg);
680 } else if (arg_match(&arg, &g_av1_codec_arg_defs.psnrarg, argi)) {
681 if (arg.val)
682 global->show_psnr = arg_parse_int(&arg);
683 else
684 global->show_psnr = 1;
685 } else if (arg_match(&arg, &g_av1_codec_arg_defs.recontest, argi)) {
686 global->test_decode = arg_parse_enum_or_int(&arg);
687 } else if (arg_match(&arg, &g_av1_codec_arg_defs.framerate, argi)) {
688 global->framerate = arg_parse_rational(&arg);
689 validate_positive_rational(arg.name, &global->framerate);
690 global->have_framerate = 1;
691 } else if (arg_match(&arg, &g_av1_codec_arg_defs.debugmode, argi)) {
692 global->debug = 1;
693 } else if (arg_match(&arg, &g_av1_codec_arg_defs.q_hist_n, argi)) {
694 global->show_q_hist_buckets = arg_parse_uint(&arg);
695 } else if (arg_match(&arg, &g_av1_codec_arg_defs.rate_hist_n, argi)) {
696 global->show_rate_hist_buckets = arg_parse_uint(&arg);
697 } else if (arg_match(&arg, &g_av1_codec_arg_defs.disable_warnings, argi)) {
698 global->disable_warnings = 1;
699 } else if (arg_match(&arg, &g_av1_codec_arg_defs.disable_warning_prompt,
700 argi)) {
701 global->disable_warning_prompt = 1;
702 } else {
703 argj++;
704 }
705 }
706
707 if (global->pass) {
708 /* DWIM: Assume the user meant passes=2 if pass=2 is specified */
709 if (global->pass > global->passes) {
710 aom_tools_warn("Assuming --pass=%d implies --passes=%d\n", global->pass,
711 global->pass);
712 global->passes = global->pass;
713 }
714 }
715 /* Validate global config */
716 if (global->passes == 0) {
717#if CONFIG_AV1_ENCODER
718 // Make default AV1 passes = 2 until there is a better quality 1-pass
719 // encoder
720 if (global->codec != NULL)
721 global->passes =
722 (strcmp(get_short_name_by_aom_encoder(global->codec), "av1") == 0 &&
723 global->usage != AOM_USAGE_REALTIME)
724 ? 2
725 : 1;
726#else
727 global->passes = 1;
728#endif
729 }
730
731 if (global->usage == AOM_USAGE_REALTIME && global->passes > 1) {
732 aom_tools_warn("Enforcing one-pass encoding in realtime mode\n");
733 if (global->pass > 1)
734 die("Error: Invalid --pass=%d for one-pass encoding\n", global->pass);
735 global->passes = 1;
736 }
737
738 if (global->usage == AOM_USAGE_ALL_INTRA && global->passes > 1) {
739 aom_tools_warn("Enforcing one-pass encoding in all intra mode\n");
740 global->passes = 1;
741 }
742}
743
744static void open_input_file(struct AvxInputContext *input,
746 /* Parse certain options from the input file, if possible */
747 input->file = strcmp(input->filename, "-") ? fopen(input->filename, "rb")
748 : set_binary_mode(stdin);
749
750 if (!input->file) fatal("Failed to open input file");
751
752 if (!fseeko(input->file, 0, SEEK_END)) {
753 /* Input file is seekable. Figure out how long it is, so we can get
754 * progress info.
755 */
756 input->length = ftello(input->file);
757 rewind(input->file);
758 }
759
760 /* Default to 1:1 pixel aspect ratio. */
761 input->pixel_aspect_ratio.numerator = 1;
762 input->pixel_aspect_ratio.denominator = 1;
763
764 /* For RAW input sources, these bytes will applied on the first frame
765 * in read_frame().
766 */
767 input->detect.buf_read = fread(input->detect.buf, 1, 4, input->file);
768 input->detect.position = 0;
769
770 if (input->detect.buf_read == 4 && file_is_y4m(input->detect.buf)) {
771 if (y4m_input_open(&input->y4m, input->file, input->detect.buf, 4, csp,
772 input->only_i420) >= 0) {
773 input->file_type = FILE_TYPE_Y4M;
774 input->width = input->y4m.pic_w;
775 input->height = input->y4m.pic_h;
776 input->pixel_aspect_ratio.numerator = input->y4m.par_n;
777 input->pixel_aspect_ratio.denominator = input->y4m.par_d;
778 input->framerate.numerator = input->y4m.fps_n;
779 input->framerate.denominator = input->y4m.fps_d;
780 input->fmt = input->y4m.aom_fmt;
781 input->bit_depth = input->y4m.bit_depth;
782 input->color_range = input->y4m.color_range;
783 } else
784 fatal("Unsupported Y4M stream.");
785 } else if (input->detect.buf_read == 4 && fourcc_is_ivf(input->detect.buf)) {
786 fatal("IVF is not supported as input.");
787 } else {
788 input->file_type = FILE_TYPE_RAW;
789 }
790}
791
792static void close_input_file(struct AvxInputContext *input) {
793 fclose(input->file);
794 if (input->file_type == FILE_TYPE_Y4M) y4m_input_close(&input->y4m);
795}
796
797static struct stream_state *new_stream(struct AvxEncoderConfig *global,
798 struct stream_state *prev) {
799 struct stream_state *stream;
800
801 stream = calloc(1, sizeof(*stream));
802 if (stream == NULL) {
803 fatal("Failed to allocate new stream.");
804 }
805
806 if (prev) {
807 memcpy(stream, prev, sizeof(*stream));
808 stream->index++;
809 prev->next = stream;
810 } else {
811 aom_codec_err_t res;
812
813 /* Populate encoder configuration */
814 res = aom_codec_enc_config_default(global->codec, &stream->config.cfg,
815 global->usage);
816 if (res) fatal("Failed to get config: %s\n", aom_codec_err_to_string(res));
817
818 /* Change the default timebase to a high enough value so that the
819 * encoder will always create strictly increasing timestamps.
820 */
821 stream->config.cfg.g_timebase.den = 1000;
822
823 /* Never use the library's default resolution, require it be parsed
824 * from the file or set on the command line.
825 */
826 stream->config.cfg.g_w = 0;
827 stream->config.cfg.g_h = 0;
828
829 /* Initialize remaining stream parameters */
830 stream->config.write_webm = 1;
831 stream->config.write_ivf = 0;
832
833#if CONFIG_WEBM_IO
834 stream->config.stereo_fmt = STEREO_FORMAT_MONO;
835 stream->webm_ctx.last_pts_ns = -1;
836 stream->webm_ctx.writer = NULL;
837 stream->webm_ctx.segment = NULL;
838#endif
839
840 /* Allows removal of the application version from the EBML tags */
841 stream->webm_ctx.debug = global->debug;
842 memcpy(&stream->config.cfg.encoder_cfg, &global->encoder_config,
843 sizeof(stream->config.cfg.encoder_cfg));
844 }
845
846 /* Output files must be specified for each stream */
847 stream->config.out_fn = NULL;
848 stream->config.two_pass_input = NULL;
849 stream->config.two_pass_output = NULL;
850 stream->config.two_pass_width = 0;
851 stream->config.two_pass_height = 0;
852
853 stream->next = NULL;
854 return stream;
855}
856
857static void set_config_arg_ctrls(struct stream_config *config, int key,
858 const struct arg *arg) {
859 int j;
860 if (key == AV1E_SET_FILM_GRAIN_TABLE) {
861 config->film_grain_filename = arg->val;
862 return;
863 }
864
865 // For target level, the settings should accumulate rather than overwrite,
866 // so we simply append it.
868 j = config->arg_ctrl_cnt;
869 assert(j < ARG_CTRL_CNT_MAX);
870 config->arg_ctrls[j][0] = key;
871 config->arg_ctrls[j][1] = arg_parse_enum_or_int(arg);
872 ++config->arg_ctrl_cnt;
873 return;
874 }
875
876 /* Point either to the next free element or the first instance of this
877 * control.
878 */
879 for (j = 0; j < config->arg_ctrl_cnt; j++)
880 if (config->arg_ctrls[j][0] == key) break;
881
882 /* Update/insert */
883 assert(j < ARG_CTRL_CNT_MAX);
884 config->arg_ctrls[j][0] = key;
885 config->arg_ctrls[j][1] = arg_parse_enum_or_int(arg);
886
887 if (key == AOME_SET_ENABLEAUTOALTREF && config->arg_ctrls[j][1] > 1) {
888 aom_tools_warn(
889 "auto-alt-ref > 1 is deprecated... setting auto-alt-ref=1\n");
890 config->arg_ctrls[j][1] = 1;
891 }
892
893 if (j == config->arg_ctrl_cnt) config->arg_ctrl_cnt++;
894}
895
896static void set_config_arg_key_vals(struct stream_config *config,
897 const char *name, const struct arg *arg) {
898 int j;
899 const char *val = arg->val;
900 // For target level, the settings should accumulate rather than overwrite,
901 // so we simply append it.
902 if (strcmp(name, "target-seq-level-idx") == 0) {
903 j = config->arg_key_val_cnt;
904 assert(j < ARG_KEY_VAL_CNT_MAX);
905 config->arg_key_vals[j][0] = name;
906 config->arg_key_vals[j][1] = val;
907 ++config->arg_key_val_cnt;
908 return;
909 }
910
911 /* Point either to the next free element or the first instance of this
912 * option.
913 */
914 for (j = 0; j < config->arg_key_val_cnt; j++)
915 if (strcmp(name, config->arg_key_vals[j][0]) == 0) break;
916
917 /* Update/insert */
918 assert(j < ARG_KEY_VAL_CNT_MAX);
919 config->arg_key_vals[j][0] = name;
920 config->arg_key_vals[j][1] = val;
921
922 if (strcmp(name, g_av1_codec_arg_defs.auto_altref.long_name) == 0) {
923 int auto_altref = arg_parse_int(arg);
924 if (auto_altref > 1) {
925 aom_tools_warn(
926 "auto-alt-ref > 1 is deprecated... setting auto-alt-ref=1\n");
927 config->arg_key_vals[j][1] = "1";
928 }
929 }
930
931 if (j == config->arg_key_val_cnt) config->arg_key_val_cnt++;
932}
933
934static int parse_stream_params(struct AvxEncoderConfig *global,
935 struct stream_state *stream, char **argv) {
936 char **argi, **argj;
937 struct arg arg;
938 static const arg_def_t **ctrl_args = no_args;
939 static const arg_def_t **key_val_args = no_args;
940 static const int *ctrl_args_map = NULL;
941 struct stream_config *config = &stream->config;
942 int eos_mark_found = 0;
943 int webm_forced = 0;
944
945 // Handle codec specific options
946 if (0) {
947#if CONFIG_AV1_ENCODER
948 } else if (strcmp(get_short_name_by_aom_encoder(global->codec), "av1") == 0) {
949 // TODO(jingning): Reuse AV1 specific encoder configuration parameters.
950 // Consider to expand this set for AV1 encoder control.
951#if __STDC_VERSION__ >= 201112L
952 _Static_assert(NELEMENTS(av1_ctrl_args) == NELEMENTS(av1_arg_ctrl_map),
953 "The av1_ctrl_args and av1_arg_ctrl_map arrays must be of "
954 "the same size.");
955#else
956 assert(NELEMENTS(av1_ctrl_args) == NELEMENTS(av1_arg_ctrl_map));
957#endif
958 ctrl_args = av1_ctrl_args;
959 ctrl_args_map = av1_arg_ctrl_map;
960 key_val_args = av1_key_val_args;
961#endif
962 }
963
964 for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step) {
965 arg.argv_step = 1;
966
967 /* Once we've found an end-of-stream marker (--) we want to continue
968 * shifting arguments but not consuming them.
969 */
970 if (eos_mark_found) {
971 argj++;
972 continue;
973 } else if (!strcmp(*argj, "--")) {
974 eos_mark_found = 1;
975 continue;
976 }
977
978 if (arg_match(&arg, &g_av1_codec_arg_defs.outputfile, argi)) {
979 config->out_fn = arg.val;
980 if (!webm_forced) {
981 const size_t out_fn_len = strlen(config->out_fn);
982 if (out_fn_len >= 4 &&
983 !strcmp(config->out_fn + out_fn_len - 4, ".ivf")) {
984 config->write_webm = 0;
985 config->write_ivf = 1;
986 } else if (out_fn_len >= 4 &&
987 !strcmp(config->out_fn + out_fn_len - 4, ".obu")) {
988 config->write_webm = 0;
989 config->write_ivf = 0;
990 }
991 }
992 } else if (arg_match(&arg, &g_av1_codec_arg_defs.fpf_name, argi)) {
993 config->stats_fn = arg.val;
994 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_webm, argi)) {
995#if CONFIG_WEBM_IO
996 config->write_webm = 1;
997 webm_forced = 1;
998#else
999 die("Error: --webm specified but webm is disabled.");
1000#endif
1001 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_ivf, argi)) {
1002 config->write_webm = 0;
1003 config->write_ivf = 1;
1004 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_obu, argi)) {
1005 config->write_webm = 0;
1006 config->write_ivf = 0;
1007 } else if (arg_match(&arg, &g_av1_codec_arg_defs.threads, argi)) {
1008 config->cfg.g_threads = arg_parse_uint(&arg);
1009 } else if (arg_match(&arg, &g_av1_codec_arg_defs.profile, argi)) {
1010 config->cfg.g_profile = arg_parse_uint(&arg);
1011 } else if (arg_match(&arg, &g_av1_codec_arg_defs.width, argi)) {
1012 config->cfg.g_w = arg_parse_uint(&arg);
1013 } else if (arg_match(&arg, &g_av1_codec_arg_defs.height, argi)) {
1014 config->cfg.g_h = arg_parse_uint(&arg);
1015 } else if (arg_match(&arg, &g_av1_codec_arg_defs.forced_max_frame_width,
1016 argi)) {
1017 config->cfg.g_forced_max_frame_width = arg_parse_uint(&arg);
1018 } else if (arg_match(&arg, &g_av1_codec_arg_defs.forced_max_frame_height,
1019 argi)) {
1020 config->cfg.g_forced_max_frame_height = arg_parse_uint(&arg);
1021 } else if (arg_match(&arg, &g_av1_codec_arg_defs.bitdeptharg, argi)) {
1022 config->cfg.g_bit_depth = arg_parse_enum_or_int(&arg);
1023 } else if (arg_match(&arg, &g_av1_codec_arg_defs.inbitdeptharg, argi)) {
1024 config->cfg.g_input_bit_depth = arg_parse_uint(&arg);
1025 } else if (arg_match(&arg, &g_av1_codec_arg_defs.input_chroma_subsampling_x,
1026 argi)) {
1027 stream->chroma_subsampling_x = arg_parse_uint(&arg);
1028 } else if (arg_match(&arg, &g_av1_codec_arg_defs.input_chroma_subsampling_y,
1029 argi)) {
1030 stream->chroma_subsampling_y = arg_parse_uint(&arg);
1031#if CONFIG_WEBM_IO
1032 } else if (arg_match(&arg, &g_av1_codec_arg_defs.stereo_mode, argi)) {
1033 config->stereo_fmt = arg_parse_enum_or_int(&arg);
1034#endif
1035 } else if (arg_match(&arg, &g_av1_codec_arg_defs.timebase, argi)) {
1036 config->cfg.g_timebase = arg_parse_rational(&arg);
1037 validate_positive_rational(arg.name, &config->cfg.g_timebase);
1038 } else if (arg_match(&arg, &g_av1_codec_arg_defs.global_error_resilient,
1039 argi)) {
1040 config->cfg.g_error_resilient = arg_parse_uint(&arg);
1041 } else if (arg_match(&arg, &g_av1_codec_arg_defs.lag_in_frames, argi)) {
1042 config->cfg.g_lag_in_frames = arg_parse_uint(&arg);
1043 } else if (arg_match(&arg, &g_av1_codec_arg_defs.large_scale_tile, argi)) {
1044 config->cfg.large_scale_tile = arg_parse_uint(&arg);
1045 if (config->cfg.large_scale_tile) {
1046 global->codec = get_aom_encoder_by_short_name("av1");
1047 }
1048 } else if (arg_match(&arg, &g_av1_codec_arg_defs.monochrome, argi)) {
1049 config->cfg.monochrome = 1;
1050 } else if (arg_match(&arg, &g_av1_codec_arg_defs.full_still_picture_hdr,
1051 argi)) {
1052 config->cfg.full_still_picture_hdr = 1;
1053 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_16bit_internal,
1054 argi)) {
1055 config->use_16bit_internal = CONFIG_AV1_HIGHBITDEPTH;
1056 if (!config->use_16bit_internal) {
1057 aom_tools_warn("%s option ignored with CONFIG_AV1_HIGHBITDEPTH=0.\n",
1058 arg.name);
1059 }
1060 } else if (arg_match(&arg, &g_av1_codec_arg_defs.dropframe_thresh, argi)) {
1061 config->cfg.rc_dropframe_thresh = arg_parse_uint(&arg);
1062 } else if (arg_match(&arg, &g_av1_codec_arg_defs.resize_mode, argi)) {
1063 config->cfg.rc_resize_mode = arg_parse_uint(&arg);
1064 } else if (arg_match(&arg, &g_av1_codec_arg_defs.resize_denominator,
1065 argi)) {
1066 config->cfg.rc_resize_denominator = arg_parse_uint(&arg);
1067 } else if (arg_match(&arg, &g_av1_codec_arg_defs.resize_kf_denominator,
1068 argi)) {
1069 config->cfg.rc_resize_kf_denominator = arg_parse_uint(&arg);
1070 } else if (arg_match(&arg, &g_av1_codec_arg_defs.superres_mode, argi)) {
1071 config->cfg.rc_superres_mode = arg_parse_uint(&arg);
1072 } else if (arg_match(&arg, &g_av1_codec_arg_defs.superres_denominator,
1073 argi)) {
1074 config->cfg.rc_superres_denominator = arg_parse_uint(&arg);
1075 } else if (arg_match(&arg, &g_av1_codec_arg_defs.superres_kf_denominator,
1076 argi)) {
1077 config->cfg.rc_superres_kf_denominator = arg_parse_uint(&arg);
1078 } else if (arg_match(&arg, &g_av1_codec_arg_defs.superres_qthresh, argi)) {
1079 config->cfg.rc_superres_qthresh = arg_parse_uint(&arg);
1080 } else if (arg_match(&arg, &g_av1_codec_arg_defs.superres_kf_qthresh,
1081 argi)) {
1082 config->cfg.rc_superres_kf_qthresh = arg_parse_uint(&arg);
1083 } else if (arg_match(&arg, &g_av1_codec_arg_defs.end_usage, argi)) {
1084 config->cfg.rc_end_usage = arg_parse_enum_or_int(&arg);
1085 } else if (arg_match(&arg, &g_av1_codec_arg_defs.target_bitrate, argi)) {
1086 config->cfg.rc_target_bitrate = arg_parse_uint(&arg);
1087 } else if (arg_match(&arg, &g_av1_codec_arg_defs.min_quantizer, argi)) {
1088 config->cfg.rc_min_quantizer = arg_parse_uint(&arg);
1089 } else if (arg_match(&arg, &g_av1_codec_arg_defs.max_quantizer, argi)) {
1090 config->cfg.rc_max_quantizer = arg_parse_uint(&arg);
1091 } else if (arg_match(&arg, &g_av1_codec_arg_defs.undershoot_pct, argi)) {
1092 config->cfg.rc_undershoot_pct = arg_parse_uint(&arg);
1093 } else if (arg_match(&arg, &g_av1_codec_arg_defs.overshoot_pct, argi)) {
1094 config->cfg.rc_overshoot_pct = arg_parse_uint(&arg);
1095 } else if (arg_match(&arg, &g_av1_codec_arg_defs.buf_sz, argi)) {
1096 config->cfg.rc_buf_sz = arg_parse_uint(&arg);
1097 } else if (arg_match(&arg, &g_av1_codec_arg_defs.buf_initial_sz, argi)) {
1098 config->cfg.rc_buf_initial_sz = arg_parse_uint(&arg);
1099 } else if (arg_match(&arg, &g_av1_codec_arg_defs.buf_optimal_sz, argi)) {
1100 config->cfg.rc_buf_optimal_sz = arg_parse_uint(&arg);
1101 } else if (arg_match(&arg, &g_av1_codec_arg_defs.bias_pct, argi)) {
1102 config->cfg.rc_2pass_vbr_bias_pct = arg_parse_uint(&arg);
1103 if (global->passes < 2)
1104 aom_tools_warn("option %s ignored in one-pass mode.\n", arg.name);
1105 } else if (arg_match(&arg, &g_av1_codec_arg_defs.minsection_pct, argi)) {
1106 config->cfg.rc_2pass_vbr_minsection_pct = arg_parse_uint(&arg);
1107
1108 if (global->passes < 2)
1109 aom_tools_warn("option %s ignored in one-pass mode.\n", arg.name);
1110 } else if (arg_match(&arg, &g_av1_codec_arg_defs.maxsection_pct, argi)) {
1111 config->cfg.rc_2pass_vbr_maxsection_pct = arg_parse_uint(&arg);
1112
1113 if (global->passes < 2)
1114 aom_tools_warn("option %s ignored in one-pass mode.\n", arg.name);
1115 } else if (arg_match(&arg, &g_av1_codec_arg_defs.fwd_kf_enabled, argi)) {
1116 config->cfg.fwd_kf_enabled = arg_parse_uint(&arg);
1117 } else if (arg_match(&arg, &g_av1_codec_arg_defs.kf_min_dist, argi)) {
1118 config->cfg.kf_min_dist = arg_parse_uint(&arg);
1119 } else if (arg_match(&arg, &g_av1_codec_arg_defs.kf_max_dist, argi)) {
1120 config->cfg.kf_max_dist = arg_parse_uint(&arg);
1121 } else if (arg_match(&arg, &g_av1_codec_arg_defs.kf_disabled, argi)) {
1122 config->cfg.kf_mode = AOM_KF_DISABLED;
1123 } else if (arg_match(&arg, &g_av1_codec_arg_defs.sframe_dist, argi)) {
1124 config->cfg.sframe_dist = arg_parse_uint(&arg);
1125 } else if (arg_match(&arg, &g_av1_codec_arg_defs.sframe_mode, argi)) {
1126 config->cfg.sframe_mode = arg_parse_uint(&arg);
1127 } else if (arg_match(&arg, &g_av1_codec_arg_defs.save_as_annexb, argi)) {
1128 config->cfg.save_as_annexb = arg_parse_uint(&arg);
1129 } else if (arg_match(&arg, &g_av1_codec_arg_defs.tile_width, argi)) {
1130 config->cfg.tile_width_count =
1131 arg_parse_list(&arg, config->cfg.tile_widths, MAX_TILE_WIDTHS);
1132 } else if (arg_match(&arg, &g_av1_codec_arg_defs.tile_height, argi)) {
1133 config->cfg.tile_height_count =
1134 arg_parse_list(&arg, config->cfg.tile_heights, MAX_TILE_HEIGHTS);
1135#if CONFIG_TUNE_VMAF
1136 } else if (arg_match(&arg, &g_av1_codec_arg_defs.vmaf_model_path, argi)) {
1137 config->vmaf_model_path = arg.val;
1138#endif
1139 } else if (arg_match(&arg, &g_av1_codec_arg_defs.partition_info_path,
1140 argi)) {
1141 config->partition_info_path = arg.val;
1142 } else if (arg_match(&arg, &g_av1_codec_arg_defs.enable_rate_guide_deltaq,
1143 argi)) {
1144 config->enable_rate_guide_deltaq = arg_parse_uint(&arg);
1145 } else if (arg_match(&arg, &g_av1_codec_arg_defs.rate_distribution_info,
1146 argi)) {
1147 config->rate_distribution_info = arg.val;
1148 } else if (arg_match(&arg, &g_av1_codec_arg_defs.use_fixed_qp_offsets,
1149 argi)) {
1150 config->cfg.use_fixed_qp_offsets = arg_parse_uint(&arg);
1151 } else if (arg_match(&arg, &g_av1_codec_arg_defs.fixed_qp_offsets, argi)) {
1152 config->cfg.use_fixed_qp_offsets = 1;
1153 } else if (global->usage == AOM_USAGE_REALTIME &&
1154 arg_match(&arg, &g_av1_codec_arg_defs.enable_restoration,
1155 argi)) {
1156 if (arg_parse_uint(&arg) == 1) {
1157 aom_tools_warn("non-zero %s option ignored in realtime mode.\n",
1158 arg.name);
1159 }
1160 } else if (arg_match(&arg, &g_av1_codec_arg_defs.two_pass_input, argi)) {
1161 config->two_pass_input = arg.val;
1162 } else if (arg_match(&arg, &g_av1_codec_arg_defs.two_pass_output, argi)) {
1163 config->two_pass_output = arg.val;
1164 } else if (arg_match(&arg, &g_av1_codec_arg_defs.two_pass_width, argi)) {
1165 config->two_pass_width = arg_parse_int(&arg);
1166 } else if (arg_match(&arg, &g_av1_codec_arg_defs.two_pass_height, argi)) {
1167 config->two_pass_height = arg_parse_int(&arg);
1168 } else {
1169 int i, match = 0;
1170 // check if the control ID API supports this arg
1171 if (ctrl_args_map) {
1172 for (i = 0; ctrl_args[i]; i++) {
1173 if (arg_match(&arg, ctrl_args[i], argi)) {
1174 match = 1;
1175 set_config_arg_ctrls(config, ctrl_args_map[i], &arg);
1176 break;
1177 }
1178 }
1179 }
1180 if (!match) {
1181 // check if the key & value API supports this arg
1182 for (i = 0; key_val_args[i]; i++) {
1183 if (arg_match(&arg, key_val_args[i], argi)) {
1184 match = 1;
1185 set_config_arg_key_vals(config, key_val_args[i]->long_name, &arg);
1186 break;
1187 }
1188 }
1189 }
1190 if (!match) argj++;
1191 }
1192 }
1193 config->use_16bit_internal |= config->cfg.g_bit_depth > AOM_BITS_8;
1194
1195 if (global->usage == AOM_USAGE_REALTIME && config->cfg.g_lag_in_frames != 0) {
1196 aom_tools_warn("non-zero lag-in-frames option ignored in realtime mode.\n");
1197 config->cfg.g_lag_in_frames = 0;
1198 }
1199
1200 if (global->usage == AOM_USAGE_ALL_INTRA) {
1201 if (config->cfg.g_lag_in_frames != 0) {
1202 aom_tools_warn(
1203 "non-zero lag-in-frames option ignored in all intra mode.\n");
1204 config->cfg.g_lag_in_frames = 0;
1205 }
1206 if (config->cfg.kf_max_dist != 0) {
1207 aom_tools_warn(
1208 "non-zero max key frame distance option ignored in all intra "
1209 "mode.\n");
1210 config->cfg.kf_max_dist = 0;
1211 }
1212 }
1213
1214 // set the passes field using key & val API
1215 if (config->arg_key_val_cnt >= ARG_KEY_VAL_CNT_MAX) {
1216 die("Not enough buffer for the key & value API.");
1217 }
1218 config->arg_key_vals[config->arg_key_val_cnt][0] = "passes";
1219 switch (global->passes) {
1220 case 0: config->arg_key_vals[config->arg_key_val_cnt][1] = "0"; break;
1221 case 1: config->arg_key_vals[config->arg_key_val_cnt][1] = "1"; break;
1222 case 2: config->arg_key_vals[config->arg_key_val_cnt][1] = "2"; break;
1223 case 3: config->arg_key_vals[config->arg_key_val_cnt][1] = "3"; break;
1224 default: die("Invalid value of --passes.");
1225 }
1226 config->arg_key_val_cnt++;
1227
1228 // set the two_pass_output field
1229 if (!config->two_pass_output && global->passes == 3) {
1230 // If not specified, set the name of two_pass_output file here.
1231 snprintf(stream->tmp_out_fn, sizeof(stream->tmp_out_fn),
1232 "%.980s_pass2_%d.ivf", stream->config.out_fn, stream->index);
1233 stream->config.two_pass_output = stream->tmp_out_fn;
1234 }
1235 if (config->two_pass_output) {
1236 config->arg_key_vals[config->arg_key_val_cnt][0] = "two-pass-output";
1237 config->arg_key_vals[config->arg_key_val_cnt][1] = config->two_pass_output;
1238 config->arg_key_val_cnt++;
1239 }
1240
1241 return eos_mark_found;
1242}
1243
1244#define FOREACH_STREAM(iterator, list) \
1245 for (struct stream_state *iterator = list; iterator; \
1246 iterator = iterator->next)
1247
1248static void validate_stream_config(const struct stream_state *stream,
1249 const struct AvxEncoderConfig *global) {
1250 const struct stream_state *streami;
1251 (void)global;
1252
1253 if (!stream->config.cfg.g_w || !stream->config.cfg.g_h)
1254 fatal(
1255 "Stream %d: Specify stream dimensions with --width (-w) "
1256 " and --height (-h)",
1257 stream->index);
1258
1259 /* Even if bit depth is set on the command line flag to be lower,
1260 * it is upgraded to at least match the input bit depth.
1261 */
1262 assert(stream->config.cfg.g_input_bit_depth <=
1263 (unsigned int)stream->config.cfg.g_bit_depth);
1264
1265 for (streami = stream; streami; streami = streami->next) {
1266 /* All streams require output files */
1267 if (!streami->config.out_fn)
1268 fatal("Stream %d: Output file is required (specify with -o)",
1269 streami->index);
1270
1271 /* Check for two streams outputting to the same file */
1272 if (streami != stream) {
1273 const char *a = stream->config.out_fn;
1274 const char *b = streami->config.out_fn;
1275 if (!strcmp(a, b) && strcmp(a, "/dev/null") && strcmp(a, ":nul"))
1276 fatal("Stream %d: duplicate output file (from stream %d)",
1277 streami->index, stream->index);
1278 }
1279
1280 /* Check for two streams sharing a stats file. */
1281 if (streami != stream) {
1282 const char *a = stream->config.stats_fn;
1283 const char *b = streami->config.stats_fn;
1284 if (a && b && !strcmp(a, b))
1285 fatal("Stream %d: duplicate stats file (from stream %d)",
1286 streami->index, stream->index);
1287 }
1288 }
1289}
1290
1291static void set_stream_dimensions(struct stream_state *stream, unsigned int w,
1292 unsigned int h) {
1293 if (!stream->config.cfg.g_w) {
1294 if (!stream->config.cfg.g_h)
1295 stream->config.cfg.g_w = w;
1296 else
1297 stream->config.cfg.g_w = w * stream->config.cfg.g_h / h;
1298 }
1299 if (!stream->config.cfg.g_h) {
1300 stream->config.cfg.g_h = h * stream->config.cfg.g_w / w;
1301 }
1302}
1303
1304static const char *file_type_to_string(enum VideoFileType t) {
1305 switch (t) {
1306 case FILE_TYPE_RAW: return "RAW";
1307 case FILE_TYPE_Y4M: return "Y4M";
1308 default: return "Other";
1309 }
1310}
1311
1312static void show_stream_config(struct stream_state *stream,
1313 struct AvxEncoderConfig *global,
1314 struct AvxInputContext *input) {
1315#define SHOW(field) \
1316 fprintf(stderr, " %-28s = %d\n", #field, stream->config.cfg.field)
1317
1318 if (stream->index == 0) {
1319 fprintf(stderr, "Codec: %s\n", aom_codec_iface_name(global->codec));
1320 fprintf(stderr, "Source file: %s File Type: %s Format: %s\n",
1321 input->filename, file_type_to_string(input->file_type),
1322 image_format_to_string(input->fmt));
1323 }
1324 if (stream->next || stream->index)
1325 fprintf(stderr, "\nStream Index: %d\n", stream->index);
1326 fprintf(stderr, "Destination file: %s\n", stream->config.out_fn);
1327 fprintf(stderr, "Coding path: %s\n",
1328 stream->config.use_16bit_internal ? "HBD" : "LBD");
1329 fprintf(stderr, "Encoder parameters:\n");
1330
1331 SHOW(g_usage);
1332 SHOW(g_threads);
1333 SHOW(g_profile);
1334 SHOW(g_w);
1335 SHOW(g_h);
1336 SHOW(g_bit_depth);
1337 SHOW(g_input_bit_depth);
1338 SHOW(g_timebase.num);
1339 SHOW(g_timebase.den);
1340 SHOW(g_error_resilient);
1341 SHOW(g_pass);
1342 SHOW(g_lag_in_frames);
1343 SHOW(large_scale_tile);
1344 SHOW(rc_dropframe_thresh);
1345 SHOW(rc_resize_mode);
1346 SHOW(rc_resize_denominator);
1347 SHOW(rc_resize_kf_denominator);
1348 SHOW(rc_superres_mode);
1349 SHOW(rc_superres_denominator);
1350 SHOW(rc_superres_kf_denominator);
1351 SHOW(rc_superres_qthresh);
1352 SHOW(rc_superres_kf_qthresh);
1353 SHOW(rc_end_usage);
1354 SHOW(rc_target_bitrate);
1355 SHOW(rc_min_quantizer);
1356 SHOW(rc_max_quantizer);
1357 SHOW(rc_undershoot_pct);
1358 SHOW(rc_overshoot_pct);
1359 SHOW(rc_buf_sz);
1360 SHOW(rc_buf_initial_sz);
1361 SHOW(rc_buf_optimal_sz);
1362 SHOW(rc_2pass_vbr_bias_pct);
1363 SHOW(rc_2pass_vbr_minsection_pct);
1364 SHOW(rc_2pass_vbr_maxsection_pct);
1365 SHOW(fwd_kf_enabled);
1366 SHOW(kf_mode);
1367 SHOW(kf_min_dist);
1368 SHOW(kf_max_dist);
1369
1370#define SHOW_PARAMS(field) \
1371 fprintf(stderr, " %-28s = %d\n", #field, \
1372 stream->config.cfg.encoder_cfg.field)
1373 if (global->encoder_config.init_by_cfg_file) {
1374 SHOW_PARAMS(super_block_size);
1375 SHOW_PARAMS(max_partition_size);
1376 SHOW_PARAMS(min_partition_size);
1377 SHOW_PARAMS(disable_ab_partition_type);
1378 SHOW_PARAMS(disable_rect_partition_type);
1379 SHOW_PARAMS(disable_1to4_partition_type);
1380 SHOW_PARAMS(disable_flip_idtx);
1381 SHOW_PARAMS(disable_cdef);
1382 SHOW_PARAMS(disable_lr);
1383 SHOW_PARAMS(disable_obmc);
1384 SHOW_PARAMS(disable_warp_motion);
1385 SHOW_PARAMS(disable_global_motion);
1386 SHOW_PARAMS(disable_dist_wtd_comp);
1387 SHOW_PARAMS(disable_diff_wtd_comp);
1388 SHOW_PARAMS(disable_inter_intra_comp);
1389 SHOW_PARAMS(disable_masked_comp);
1390 SHOW_PARAMS(disable_one_sided_comp);
1391 SHOW_PARAMS(disable_palette);
1392 SHOW_PARAMS(disable_intrabc);
1393 SHOW_PARAMS(disable_cfl);
1394 SHOW_PARAMS(disable_smooth_intra);
1395 SHOW_PARAMS(disable_filter_intra);
1396 SHOW_PARAMS(disable_dual_filter);
1397 SHOW_PARAMS(disable_intra_angle_delta);
1398 SHOW_PARAMS(disable_intra_edge_filter);
1399 SHOW_PARAMS(disable_tx_64x64);
1400 SHOW_PARAMS(disable_smooth_inter_intra);
1401 SHOW_PARAMS(disable_inter_inter_wedge);
1402 SHOW_PARAMS(disable_inter_intra_wedge);
1403 SHOW_PARAMS(disable_paeth_intra);
1404 SHOW_PARAMS(disable_trellis_quant);
1405 SHOW_PARAMS(disable_ref_frame_mv);
1406 SHOW_PARAMS(reduced_reference_set);
1407 SHOW_PARAMS(reduced_tx_type_set);
1408 }
1409}
1410
1411static void open_output_file(struct stream_state *stream,
1412 struct AvxEncoderConfig *global,
1413 const struct AvxRational *pixel_aspect_ratio,
1414 const char *encoder_settings) {
1415 const char *fn = stream->config.out_fn;
1416 const struct aom_codec_enc_cfg *const cfg = &stream->config.cfg;
1417
1418 if (cfg->g_pass == AOM_RC_FIRST_PASS) return;
1419
1420 stream->file = strcmp(fn, "-") ? fopen(fn, "wb") : set_binary_mode(stdout);
1421
1422 if (!stream->file) fatal("Failed to open output file");
1423
1424 if (stream->config.write_webm && fseek(stream->file, 0, SEEK_CUR))
1425 fatal("WebM output to pipes not supported.");
1426
1427#if CONFIG_WEBM_IO
1428 if (stream->config.write_webm) {
1429 stream->webm_ctx.stream = stream->file;
1430 if (write_webm_file_header(&stream->webm_ctx, &stream->encoder, cfg,
1431 stream->config.stereo_fmt,
1432 get_fourcc_by_aom_encoder(global->codec),
1433 pixel_aspect_ratio, encoder_settings) != 0) {
1434 fatal("WebM writer initialization failed.");
1435 }
1436 }
1437#else
1438 (void)pixel_aspect_ratio;
1439 (void)encoder_settings;
1440#endif
1441
1442 if (!stream->config.write_webm && stream->config.write_ivf) {
1443 ivf_write_file_header(stream->file, cfg,
1444 get_fourcc_by_aom_encoder(global->codec), 0);
1445 }
1446}
1447
1448static void close_output_file(struct stream_state *stream,
1449 unsigned int fourcc) {
1450 const struct aom_codec_enc_cfg *const cfg = &stream->config.cfg;
1451
1452 if (cfg->g_pass == AOM_RC_FIRST_PASS) return;
1453
1454#if CONFIG_WEBM_IO
1455 if (stream->config.write_webm) {
1456 if (write_webm_file_footer(&stream->webm_ctx) != 0) {
1457 fatal("WebM writer finalization failed.");
1458 }
1459 }
1460#endif
1461
1462 if (!stream->config.write_webm && stream->config.write_ivf) {
1463 if (!fseek(stream->file, 0, SEEK_SET))
1464 ivf_write_file_header(stream->file, &stream->config.cfg, fourcc,
1465 stream->frames_out);
1466 }
1467
1468 fclose(stream->file);
1469}
1470
1471static void setup_pass(struct stream_state *stream,
1472 struct AvxEncoderConfig *global, int pass) {
1473 if (stream->config.stats_fn) {
1474 if (!stats_open_file(&stream->stats, stream->config.stats_fn, pass))
1475 fatal("Failed to open statistics store");
1476 } else {
1477 if (!stats_open_mem(&stream->stats, pass))
1478 fatal("Failed to open statistics store");
1479 }
1480
1481 if (global->passes == 1) {
1482 stream->config.cfg.g_pass = AOM_RC_ONE_PASS;
1483 } else {
1484 switch (pass) {
1485 case 0: stream->config.cfg.g_pass = AOM_RC_FIRST_PASS; break;
1486 case 1: stream->config.cfg.g_pass = AOM_RC_SECOND_PASS; break;
1487 case 2: stream->config.cfg.g_pass = AOM_RC_THIRD_PASS; break;
1488 default: fatal("Failed to set pass");
1489 }
1490 }
1491
1492 if (pass) {
1493 stream->config.cfg.rc_twopass_stats_in = stats_get(&stream->stats);
1494 }
1495
1496 stream->cx_time = 0;
1497 stream->nbytes = 0;
1498 stream->frames_out = 0;
1499}
1500
1501static void initialize_encoder(struct stream_state *stream,
1502 struct AvxEncoderConfig *global) {
1503 int i;
1504 int flags = 0;
1505
1506 flags |= (global->show_psnr >= 1) ? AOM_CODEC_USE_PSNR : 0;
1507 flags |= stream->config.use_16bit_internal ? AOM_CODEC_USE_HIGHBITDEPTH : 0;
1508
1509 /* Construct Encoder Context */
1510 aom_codec_enc_init(&stream->encoder, global->codec, &stream->config.cfg,
1511 flags);
1512 ctx_exit_on_error(&stream->encoder, "Failed to initialize encoder");
1513
1514 for (i = 0; i < stream->config.arg_ctrl_cnt; i++) {
1515 int ctrl = stream->config.arg_ctrls[i][0];
1516 int value = stream->config.arg_ctrls[i][1];
1517 if (aom_codec_control(&stream->encoder, ctrl, value))
1518 fprintf(stderr, "Error: Tried to set control %d = %d\n", ctrl, value);
1519
1520 ctx_exit_on_error(&stream->encoder, "Failed to control codec");
1521 }
1522
1523 for (i = 0; i < stream->config.arg_key_val_cnt; i++) {
1524 const char *name = stream->config.arg_key_vals[i][0];
1525 const char *val = stream->config.arg_key_vals[i][1];
1526 if (aom_codec_set_option(&stream->encoder, name, val))
1527 fprintf(stderr, "Error: Tried to set option %s = %s\n", name, val);
1528
1529 ctx_exit_on_error(&stream->encoder, "Failed to set codec option");
1530 }
1531
1532#if CONFIG_TUNE_VMAF
1533 if (stream->config.vmaf_model_path) {
1535 stream->config.vmaf_model_path);
1536 }
1537#endif
1538 if (stream->config.partition_info_path) {
1539 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
1541 stream->config.partition_info_path);
1542 }
1543 if (stream->config.enable_rate_guide_deltaq) {
1544 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
1546 stream->config.enable_rate_guide_deltaq);
1547 }
1548 if (stream->config.rate_distribution_info) {
1549 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
1551 stream->config.rate_distribution_info);
1552 }
1553
1554 if (stream->config.film_grain_filename) {
1556 stream->config.film_grain_filename);
1557 }
1559 stream->config.color_range);
1560
1561#if CONFIG_AV1_DECODER
1562 if (global->test_decode != TEST_DECODE_OFF) {
1563 aom_codec_iface_t *decoder = get_aom_decoder_by_short_name(
1564 get_short_name_by_aom_encoder(global->codec));
1565 aom_codec_dec_cfg_t cfg = { 0, 0, 0, !stream->config.use_16bit_internal };
1566 aom_codec_dec_init(&stream->decoder, decoder, &cfg, 0);
1567
1568 if (strcmp(get_short_name_by_aom_encoder(global->codec), "av1") == 0) {
1570 stream->config.cfg.large_scale_tile);
1571 ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_mode");
1572
1574 stream->config.cfg.save_as_annexb);
1575 ctx_exit_on_error(&stream->decoder, "Failed to set is_annexb");
1576
1578 -1);
1579 ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_row");
1580
1581 AOM_CODEC_CONTROL_TYPECHECKED(&stream->decoder, AV1_SET_DECODE_TILE_COL,
1582 -1);
1583 ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_col");
1584 }
1585 }
1586#endif
1587}
1588
1589// Convert the input image 'img' to a monochrome image. The Y plane of the
1590// output image is a shallow copy of the Y plane of the input image, therefore
1591// the input image must remain valid for the lifetime of the output image. The U
1592// and V planes of the output image are set to null pointers. The output image
1593// format is AOM_IMG_FMT_I420 because libaom does not have AOM_IMG_FMT_I400.
1594static void convert_image_to_monochrome(const struct aom_image *img,
1595 struct aom_image *monochrome_img) {
1596 *monochrome_img = *img;
1597 monochrome_img->fmt = AOM_IMG_FMT_I420;
1598 if (img->fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
1599 monochrome_img->fmt |= AOM_IMG_FMT_HIGHBITDEPTH;
1600 }
1601 monochrome_img->monochrome = 1;
1602 monochrome_img->csp = AOM_CSP_UNKNOWN;
1603 monochrome_img->x_chroma_shift = 1;
1604 monochrome_img->y_chroma_shift = 1;
1605 monochrome_img->planes[AOM_PLANE_U] = NULL;
1606 monochrome_img->planes[AOM_PLANE_V] = NULL;
1607 monochrome_img->stride[AOM_PLANE_U] = 0;
1608 monochrome_img->stride[AOM_PLANE_V] = 0;
1609 monochrome_img->sz = 0;
1610 monochrome_img->bps = (img->fmt & AOM_IMG_FMT_HIGHBITDEPTH) ? 16 : 8;
1611 monochrome_img->img_data = NULL;
1612 monochrome_img->img_data_owner = 0;
1613 monochrome_img->self_allocd = 0;
1614}
1615
1616static void encode_frame(struct stream_state *stream,
1617 struct AvxEncoderConfig *global, struct aom_image *img,
1618 unsigned int frames_in) {
1619 aom_codec_pts_t frame_start, next_frame_start;
1620 struct aom_codec_enc_cfg *cfg = &stream->config.cfg;
1621 struct aom_usec_timer timer;
1622
1623 frame_start =
1624 (cfg->g_timebase.den * (int64_t)(frames_in - 1) * global->framerate.den) /
1625 cfg->g_timebase.num / global->framerate.num;
1626 next_frame_start =
1627 (cfg->g_timebase.den * (int64_t)(frames_in)*global->framerate.den) /
1628 cfg->g_timebase.num / global->framerate.num;
1629
1630 /* Scale if necessary */
1631 if (img) {
1632 if ((img->fmt & AOM_IMG_FMT_HIGHBITDEPTH) &&
1633 (img->d_w != cfg->g_w || img->d_h != cfg->g_h)) {
1634 if (img->fmt != AOM_IMG_FMT_I42016) {
1635 fprintf(stderr, "%s can only scale 4:2:0 inputs\n", exec_name);
1636 exit(EXIT_FAILURE);
1637 }
1638#if CONFIG_LIBYUV
1639 if (!stream->img) {
1640 stream->img =
1641 aom_img_alloc(NULL, AOM_IMG_FMT_I42016, cfg->g_w, cfg->g_h, 16);
1642 }
1643 I420Scale_16(
1644 (uint16_t *)img->planes[AOM_PLANE_Y], img->stride[AOM_PLANE_Y] / 2,
1645 (uint16_t *)img->planes[AOM_PLANE_U], img->stride[AOM_PLANE_U] / 2,
1646 (uint16_t *)img->planes[AOM_PLANE_V], img->stride[AOM_PLANE_V] / 2,
1647 img->d_w, img->d_h, (uint16_t *)stream->img->planes[AOM_PLANE_Y],
1648 stream->img->stride[AOM_PLANE_Y] / 2,
1649 (uint16_t *)stream->img->planes[AOM_PLANE_U],
1650 stream->img->stride[AOM_PLANE_U] / 2,
1651 (uint16_t *)stream->img->planes[AOM_PLANE_V],
1652 stream->img->stride[AOM_PLANE_V] / 2, stream->img->d_w,
1653 stream->img->d_h, kFilterBox);
1654 img = stream->img;
1655#else
1656 stream->encoder.err = 1;
1657 ctx_exit_on_error(&stream->encoder,
1658 "Stream %d: Failed to encode frame.\n"
1659 "libyuv is required for scaling but is currently "
1660 "disabled.\n"
1661 "Be sure to specify -DCONFIG_LIBYUV=1 when running "
1662 "cmake.\n",
1663 stream->index);
1664#endif
1665 }
1666 }
1667 if (img && (img->d_w != cfg->g_w || img->d_h != cfg->g_h)) {
1668 if (img->fmt != AOM_IMG_FMT_I420 && img->fmt != AOM_IMG_FMT_YV12) {
1669 fprintf(stderr, "%s can only scale 4:2:0 8bpp inputs\n", exec_name);
1670 exit(EXIT_FAILURE);
1671 }
1672#if CONFIG_LIBYUV
1673 if (!stream->img)
1674 stream->img =
1675 aom_img_alloc(NULL, AOM_IMG_FMT_I420, cfg->g_w, cfg->g_h, 16);
1676 I420Scale(
1677 img->planes[AOM_PLANE_Y], img->stride[AOM_PLANE_Y],
1678 img->planes[AOM_PLANE_U], img->stride[AOM_PLANE_U],
1679 img->planes[AOM_PLANE_V], img->stride[AOM_PLANE_V], img->d_w, img->d_h,
1680 stream->img->planes[AOM_PLANE_Y], stream->img->stride[AOM_PLANE_Y],
1681 stream->img->planes[AOM_PLANE_U], stream->img->stride[AOM_PLANE_U],
1682 stream->img->planes[AOM_PLANE_V], stream->img->stride[AOM_PLANE_V],
1683 stream->img->d_w, stream->img->d_h, kFilterBox);
1684 img = stream->img;
1685#else
1686 stream->encoder.err = 1;
1687 ctx_exit_on_error(&stream->encoder,
1688 "Stream %d: Failed to encode frame.\n"
1689 "Scaling disabled in this configuration. \n"
1690 "To enable, configure with --enable-libyuv\n",
1691 stream->index);
1692#endif
1693 }
1694
1695 struct aom_image monochrome_img;
1696 if (img && cfg->monochrome) {
1697 convert_image_to_monochrome(img, &monochrome_img);
1698 img = &monochrome_img;
1699 }
1700
1701 aom_usec_timer_start(&timer);
1702 aom_codec_encode(&stream->encoder, img, frame_start,
1703 (uint32_t)(next_frame_start - frame_start), 0);
1704 aom_usec_timer_mark(&timer);
1705 stream->cx_time += aom_usec_timer_elapsed(&timer);
1706 ctx_exit_on_error(&stream->encoder, "Stream %d: Failed to encode frame",
1707 stream->index);
1708}
1709
1710static void update_quantizer_histogram(struct stream_state *stream) {
1711 if (stream->config.cfg.g_pass != AOM_RC_FIRST_PASS) {
1712 int q;
1713
1715 &q);
1716 ctx_exit_on_error(&stream->encoder, "Failed to read quantizer");
1717 stream->counts[q]++;
1718 }
1719}
1720
1721static void get_cx_data(struct stream_state *stream,
1722 struct AvxEncoderConfig *global, int *got_data) {
1723 const aom_codec_cx_pkt_t *pkt;
1724 const struct aom_codec_enc_cfg *cfg = &stream->config.cfg;
1725 aom_codec_iter_t iter = NULL;
1726
1727 *got_data = 0;
1728 while ((pkt = aom_codec_get_cx_data(&stream->encoder, &iter))) {
1729 static size_t fsize = 0;
1730 static FileOffset ivf_header_pos = 0;
1731
1732 switch (pkt->kind) {
1734 ++stream->frames_out;
1735 if (!global->quiet)
1736 fprintf(stderr, " %6luF", (unsigned long)pkt->data.frame.sz);
1737
1738 update_rate_histogram(stream->rate_hist, cfg, pkt);
1739#if CONFIG_WEBM_IO
1740 if (stream->config.write_webm) {
1741 if (write_webm_block(&stream->webm_ctx, cfg, pkt) != 0) {
1742 fatal("WebM writer failed.");
1743 }
1744 }
1745#endif
1746 if (!stream->config.write_webm) {
1747 if (stream->config.write_ivf) {
1748 if (pkt->data.frame.partition_id <= 0) {
1749 ivf_header_pos = ftello(stream->file);
1750 fsize = pkt->data.frame.sz;
1751
1752 ivf_write_frame_header(stream->file, pkt->data.frame.pts, fsize);
1753 } else {
1754 fsize += pkt->data.frame.sz;
1755
1756 const FileOffset currpos = ftello(stream->file);
1757 fseeko(stream->file, ivf_header_pos, SEEK_SET);
1758 ivf_write_frame_size(stream->file, fsize);
1759 fseeko(stream->file, currpos, SEEK_SET);
1760 }
1761 }
1762
1763 (void)fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
1764 stream->file);
1765 }
1766 stream->nbytes += pkt->data.raw.sz;
1767
1768 *got_data = 1;
1769#if CONFIG_AV1_DECODER
1770 if (global->test_decode != TEST_DECODE_OFF && !stream->mismatch_seen) {
1771 aom_codec_decode(&stream->decoder, pkt->data.frame.buf,
1772 pkt->data.frame.sz, NULL);
1773 if (stream->decoder.err) {
1774 warn_or_exit_on_error(&stream->decoder,
1775 global->test_decode == TEST_DECODE_FATAL,
1776 "Failed to decode frame %d in stream %d",
1777 stream->frames_out + 1, stream->index);
1778 stream->mismatch_seen = stream->frames_out + 1;
1779 }
1780 }
1781#endif
1782 break;
1784 stream->frames_out++;
1785 stats_write(&stream->stats, pkt->data.twopass_stats.buf,
1786 pkt->data.twopass_stats.sz);
1787 stream->nbytes += pkt->data.raw.sz;
1788 break;
1789 case AOM_CODEC_PSNR_PKT:
1790
1791 if (global->show_psnr >= 1) {
1792 int i;
1793
1794 stream->psnr_sse_total[0] += pkt->data.psnr.sse[0];
1795 stream->psnr_samples_total[0] += pkt->data.psnr.samples[0];
1796 for (i = 0; i < 4; i++) {
1797 if (!global->quiet)
1798 fprintf(stderr, "%.3f ", pkt->data.psnr.psnr[i]);
1799 stream->psnr_totals[0][i] += pkt->data.psnr.psnr[i];
1800 }
1801 stream->psnr_count[0]++;
1802
1803#if CONFIG_AV1_HIGHBITDEPTH
1804 if (stream->config.cfg.g_input_bit_depth <
1805 (unsigned int)stream->config.cfg.g_bit_depth) {
1806 stream->psnr_sse_total[1] += pkt->data.psnr.sse_hbd[0];
1807 stream->psnr_samples_total[1] += pkt->data.psnr.samples_hbd[0];
1808 for (i = 0; i < 4; i++) {
1809 if (!global->quiet)
1810 fprintf(stderr, "%.3f ", pkt->data.psnr.psnr_hbd[i]);
1811 stream->psnr_totals[1][i] += pkt->data.psnr.psnr_hbd[i];
1812 }
1813 stream->psnr_count[1]++;
1814 }
1815#endif
1816 }
1817
1818 break;
1819 default: break;
1820 }
1821 }
1822}
1823
1824static void show_psnr(struct stream_state *stream, double peak, int64_t bps) {
1825 int i;
1826 double ovpsnr;
1827
1828 if (!stream->psnr_count[0]) return;
1829
1830 fprintf(stderr, "Stream %d PSNR (Overall/Avg/Y/U/V)", stream->index);
1831 ovpsnr = sse_to_psnr((double)stream->psnr_samples_total[0], peak,
1832 (double)stream->psnr_sse_total[0]);
1833 fprintf(stderr, " %.3f", ovpsnr);
1834
1835 for (i = 0; i < 4; i++) {
1836 fprintf(stderr, " %.3f", stream->psnr_totals[0][i] / stream->psnr_count[0]);
1837 }
1838 if (bps > 0) {
1839 fprintf(stderr, " %7" PRId64 " bps", bps);
1840 }
1841 fprintf(stderr, " %7" PRId64 " ms", stream->cx_time / 1000);
1842 fprintf(stderr, "\n");
1843}
1844
1845#if CONFIG_AV1_HIGHBITDEPTH
1846static void show_psnr_hbd(struct stream_state *stream, double peak,
1847 int64_t bps) {
1848 int i;
1849 double ovpsnr;
1850 // Compute PSNR based on stream bit depth
1851 if (!stream->psnr_count[1]) return;
1852
1853 fprintf(stderr, "Stream %d PSNR (Overall/Avg/Y/U/V)", stream->index);
1854 ovpsnr = sse_to_psnr((double)stream->psnr_samples_total[1], peak,
1855 (double)stream->psnr_sse_total[1]);
1856 fprintf(stderr, " %.3f", ovpsnr);
1857
1858 for (i = 0; i < 4; i++) {
1859 fprintf(stderr, " %.3f", stream->psnr_totals[1][i] / stream->psnr_count[1]);
1860 }
1861 if (bps > 0) {
1862 fprintf(stderr, " %7" PRId64 " bps", bps);
1863 }
1864 fprintf(stderr, " %7" PRId64 " ms", stream->cx_time / 1000);
1865 fprintf(stderr, "\n");
1866}
1867#endif
1868
1869static float usec_to_fps(uint64_t usec, unsigned int frames) {
1870 return (float)(usec > 0 ? frames * 1000000.0 / (float)usec : 0);
1871}
1872
1873static void test_decode(struct stream_state *stream,
1874 enum TestDecodeFatality fatal) {
1875 aom_image_t enc_img, dec_img;
1876
1877 if (stream->mismatch_seen) return;
1878
1879 /* Get the internal reference frame */
1881 &enc_img);
1883 &dec_img);
1884
1885 if ((enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) !=
1886 (dec_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH)) {
1887 if (enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
1888 aom_image_t enc_hbd_img;
1889 aom_img_alloc(&enc_hbd_img, enc_img.fmt - AOM_IMG_FMT_HIGHBITDEPTH,
1890 enc_img.d_w, enc_img.d_h, 16);
1891 aom_img_truncate_16_to_8(&enc_hbd_img, &enc_img);
1892 enc_img = enc_hbd_img;
1893 }
1894 if (dec_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
1895 aom_image_t dec_hbd_img;
1896 aom_img_alloc(&dec_hbd_img, dec_img.fmt - AOM_IMG_FMT_HIGHBITDEPTH,
1897 dec_img.d_w, dec_img.d_h, 16);
1898 aom_img_truncate_16_to_8(&dec_hbd_img, &dec_img);
1899 dec_img = dec_hbd_img;
1900 }
1901 }
1902
1903 ctx_exit_on_error(&stream->encoder, "Failed to get encoder reference frame");
1904 ctx_exit_on_error(&stream->decoder, "Failed to get decoder reference frame");
1905
1906 if (!aom_compare_img(&enc_img, &dec_img)) {
1907 int y[4], u[4], v[4];
1908 if (enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
1909 aom_find_mismatch_high(&enc_img, &dec_img, y, u, v);
1910 } else {
1911 aom_find_mismatch(&enc_img, &dec_img, y, u, v);
1912 }
1913 stream->decoder.err = 1;
1914 warn_or_exit_on_error(&stream->decoder, fatal == TEST_DECODE_FATAL,
1915 "Stream %d: Encode/decode mismatch on frame %d at"
1916 " Y[%d, %d] {%d/%d},"
1917 " U[%d, %d] {%d/%d},"
1918 " V[%d, %d] {%d/%d}",
1919 stream->index, stream->frames_out, y[0], y[1], y[2],
1920 y[3], u[0], u[1], u[2], u[3], v[0], v[1], v[2], v[3]);
1921 stream->mismatch_seen = stream->frames_out;
1922 }
1923
1924 aom_img_free(&enc_img);
1925 aom_img_free(&dec_img);
1926}
1927
1928static void print_time(const char *label, int64_t etl) {
1929 int64_t hours;
1930 int64_t mins;
1931 int64_t secs;
1932
1933 if (etl >= 0) {
1934 hours = etl / 3600;
1935 etl -= hours * 3600;
1936 mins = etl / 60;
1937 etl -= mins * 60;
1938 secs = etl;
1939
1940 fprintf(stderr, "[%3s %2" PRId64 ":%02" PRId64 ":%02" PRId64 "] ", label,
1941 hours, mins, secs);
1942 } else {
1943 fprintf(stderr, "[%3s unknown] ", label);
1944 }
1945}
1946
1947static void clear_stream_count_state(struct stream_state *stream) {
1948 // PSNR counters
1949 for (int k = 0; k < 2; k++) {
1950 stream->psnr_sse_total[k] = 0;
1951 stream->psnr_samples_total[k] = 0;
1952 for (int i = 0; i < 4; i++) {
1953 stream->psnr_totals[k][i] = 0;
1954 }
1955 stream->psnr_count[k] = 0;
1956 }
1957 // q hist
1958 memset(stream->counts, 0, sizeof(stream->counts));
1959}
1960
1961// aomenc will downscale the second pass if:
1962// 1. the specific pass is not given by commandline (aomenc will perform all
1963// passes)
1964// 2. there are more than 2 passes in total
1965// 3. current pass is the second pass (the parameter pass starts with 0 so
1966// pass == 1)
1967static int pass_need_downscale(int global_pass, int global_passes, int pass) {
1968 return !global_pass && global_passes > 2 && pass == 1;
1969}
1970
1971int main(int argc, const char **argv_) {
1972 int pass;
1973 aom_image_t raw;
1974 aom_image_t raw_shift;
1975 int allocated_raw_shift = 0;
1976 int do_16bit_internal = 0;
1977 int input_shift = 0;
1978 int frame_avail, got_data;
1979
1980 struct AvxInputContext input;
1981 struct AvxEncoderConfig global;
1982 struct stream_state *streams = NULL;
1983 char **argv, **argi;
1984 uint64_t cx_time = 0;
1985 int stream_cnt = 0;
1986 int res = 0;
1987 int profile_updated = 0;
1988
1989 memset(&input, 0, sizeof(input));
1990 memset(&raw, 0, sizeof(raw));
1991 exec_name = argv_[0];
1992
1993 /* Setup default input stream settings */
1994 input.framerate.numerator = 30;
1995 input.framerate.denominator = 1;
1996 input.only_i420 = 1;
1997 input.bit_depth = 0;
1998
1999 /* First parse the global configuration values, because we want to apply
2000 * other parameters on top of the default configuration provided by the
2001 * codec.
2002 */
2003 argv = argv_dup(argc - 1, argv_ + 1);
2004 if (!argv) {
2005 fprintf(stderr, "Error allocating argument list\n");
2006 return EXIT_FAILURE;
2007 }
2008 parse_global_config(&global, &argv);
2009
2010 if (argc < 2) usage_exit();
2011
2012 switch (global.color_type) {
2013 case I420: input.fmt = AOM_IMG_FMT_I420; break;
2014 case I422: input.fmt = AOM_IMG_FMT_I422; break;
2015 case I444: input.fmt = AOM_IMG_FMT_I444; break;
2016 case YV12: input.fmt = AOM_IMG_FMT_YV12; break;
2017 case NV12: input.fmt = AOM_IMG_FMT_NV12; break;
2018 }
2019
2020 {
2021 /* Now parse each stream's parameters. Using a local scope here
2022 * due to the use of 'stream' as loop variable in FOREACH_STREAM
2023 * loops
2024 */
2025 struct stream_state *stream = NULL;
2026
2027 do {
2028 stream = new_stream(&global, stream);
2029 stream_cnt++;
2030 if (!streams) streams = stream;
2031 } while (parse_stream_params(&global, stream, argv));
2032 }
2033
2034 /* Check for unrecognized options */
2035 for (argi = argv; *argi; argi++)
2036 if (argi[0][0] == '-' && argi[0][1])
2037 die("Error: Unrecognized option %s\n", *argi);
2038
2039 FOREACH_STREAM(stream, streams) {
2040 check_encoder_config(global.disable_warning_prompt, &global,
2041 &stream->config.cfg);
2042
2043 // If large_scale_tile = 1, only support to output to ivf format.
2044 if (stream->config.cfg.large_scale_tile && !stream->config.write_ivf)
2045 die("only support ivf output format while large-scale-tile=1\n");
2046 }
2047
2048 /* Handle non-option arguments */
2049 input.filename = argv[0];
2050 const char *orig_input_filename = input.filename;
2051 FOREACH_STREAM(stream, streams) {
2052 stream->orig_out_fn = stream->config.out_fn;
2053 stream->orig_width = stream->config.cfg.g_w;
2054 stream->orig_height = stream->config.cfg.g_h;
2055 stream->orig_write_ivf = stream->config.write_ivf;
2056 stream->orig_write_webm = stream->config.write_webm;
2057 }
2058
2059 if (!input.filename) {
2060 fprintf(stderr, "No input file specified!\n");
2061 usage_exit();
2062 }
2063
2064 /* Decide if other chroma subsamplings than 4:2:0 are supported */
2065 if (get_fourcc_by_aom_encoder(global.codec) == AV1_FOURCC)
2066 input.only_i420 = 0;
2067
2068 for (pass = global.pass ? global.pass - 1 : 0; pass < global.passes; pass++) {
2069 if (pass > 1) {
2070 FOREACH_STREAM(stream, streams) { clear_stream_count_state(stream); }
2071 }
2072
2073 int frames_in = 0, seen_frames = 0;
2074 int64_t estimated_time_left = -1;
2075 int64_t average_rate = -1;
2076 int64_t lagged_count = 0;
2077 const int need_downscale =
2078 pass_need_downscale(global.pass, global.passes, pass);
2079
2080 // Set the output to the specified two-pass output file, and
2081 // restore the width and height to the original values.
2082 FOREACH_STREAM(stream, streams) {
2083 if (need_downscale) {
2084 stream->config.out_fn = stream->config.two_pass_output;
2085 // Libaom currently only supports the ivf format for the third pass.
2086 stream->config.write_ivf = 1;
2087 stream->config.write_webm = 0;
2088 } else {
2089 stream->config.out_fn = stream->orig_out_fn;
2090 stream->config.write_ivf = stream->orig_write_ivf;
2091 stream->config.write_webm = stream->orig_write_webm;
2092 }
2093 stream->config.cfg.g_w = stream->orig_width;
2094 stream->config.cfg.g_h = stream->orig_height;
2095 }
2096
2097 // For second pass in three-pass encoding, set the input to
2098 // the given two-pass-input file if available. If the scaled input is not
2099 // given, we will attempt to re-scale the original input.
2100 input.filename = orig_input_filename;
2101 const char *two_pass_input = NULL;
2102 if (need_downscale) {
2103 FOREACH_STREAM(stream, streams) {
2104 if (stream->config.two_pass_input) {
2105 two_pass_input = stream->config.two_pass_input;
2106 input.filename = two_pass_input;
2107 break;
2108 }
2109 }
2110 }
2111
2112 open_input_file(&input, global.csp);
2113
2114 /* If the input file doesn't specify its w/h (raw files), try to get
2115 * the data from the first stream's configuration.
2116 */
2117 if (!input.width || !input.height) {
2118 if (two_pass_input) {
2119 FOREACH_STREAM(stream, streams) {
2120 if (stream->config.two_pass_width && stream->config.two_pass_height) {
2121 input.width = stream->config.two_pass_width;
2122 input.height = stream->config.two_pass_height;
2123 break;
2124 }
2125 }
2126 } else {
2127 FOREACH_STREAM(stream, streams) {
2128 if (stream->config.cfg.g_w && stream->config.cfg.g_h) {
2129 input.width = stream->config.cfg.g_w;
2130 input.height = stream->config.cfg.g_h;
2131 break;
2132 }
2133 }
2134 }
2135 }
2136
2137 /* Update stream configurations from the input file's parameters */
2138 if (!input.width || !input.height) {
2139 if (two_pass_input) {
2140 fatal(
2141 "Specify downscaled stream dimensions with --two-pass-width "
2142 " and --two-pass-height");
2143 } else {
2144 fatal(
2145 "Specify stream dimensions with --width (-w) "
2146 " and --height (-h)");
2147 }
2148 }
2149
2150 if (need_downscale) {
2151 FOREACH_STREAM(stream, streams) {
2152 if (stream->config.two_pass_width && stream->config.two_pass_height) {
2153 stream->config.cfg.g_w = stream->config.two_pass_width;
2154 stream->config.cfg.g_h = stream->config.two_pass_height;
2155 } else if (two_pass_input) {
2156 stream->config.cfg.g_w = input.width;
2157 stream->config.cfg.g_h = input.height;
2158 } else if (stream->orig_width && stream->orig_height) {
2159#if CONFIG_BITRATE_ACCURACY || CONFIG_BITRATE_ACCURACY_BL
2160 stream->config.cfg.g_w = stream->orig_width;
2161 stream->config.cfg.g_h = stream->orig_height;
2162#else // CONFIG_BITRATE_ACCURACY || CONFIG_BITRATE_ACCURACY_BL
2163 stream->config.cfg.g_w = (stream->orig_width + 1) / 2;
2164 stream->config.cfg.g_h = (stream->orig_height + 1) / 2;
2165#endif // CONFIG_BITRATE_ACCURACY || CONFIG_BITRATE_ACCURACY_BL
2166 } else {
2167#if CONFIG_BITRATE_ACCURACY || CONFIG_BITRATE_ACCURACY_BL
2168 stream->config.cfg.g_w = input.width;
2169 stream->config.cfg.g_h = input.height;
2170#else // CONFIG_BITRATE_ACCURACY || CONFIG_BITRATE_ACCURACY_BL
2171 stream->config.cfg.g_w = (input.width + 1) / 2;
2172 stream->config.cfg.g_h = (input.height + 1) / 2;
2173#endif // CONFIG_BITRATE_ACCURACY || CONFIG_BITRATE_ACCURACY_BL
2174 }
2175 }
2176 }
2177
2178 /* If input file does not specify bit-depth but input-bit-depth parameter
2179 * exists, assume that to be the input bit-depth. However, if the
2180 * input-bit-depth paramter does not exist, assume the input bit-depth
2181 * to be the same as the codec bit-depth.
2182 */
2183 if (!input.bit_depth) {
2184 FOREACH_STREAM(stream, streams) {
2185 if (stream->config.cfg.g_input_bit_depth)
2186 input.bit_depth = stream->config.cfg.g_input_bit_depth;
2187 else
2188 input.bit_depth = stream->config.cfg.g_input_bit_depth =
2189 (int)stream->config.cfg.g_bit_depth;
2190 }
2191 if (input.bit_depth > 8) input.fmt |= AOM_IMG_FMT_HIGHBITDEPTH;
2192 } else {
2193 FOREACH_STREAM(stream, streams) {
2194 stream->config.cfg.g_input_bit_depth = input.bit_depth;
2195 }
2196 }
2197
2198 FOREACH_STREAM(stream, streams) {
2199 if (input.fmt != AOM_IMG_FMT_I420 && input.fmt != AOM_IMG_FMT_I42016 &&
2200 input.fmt != AOM_IMG_FMT_NV12) {
2201 /* Automatically upgrade if input is non-4:2:0 but a 4:2:0 profile
2202 was selected. */
2203 switch (stream->config.cfg.g_profile) {
2204 case 0:
2205 if (input.bit_depth < 12 && (input.fmt == AOM_IMG_FMT_I444 ||
2206 input.fmt == AOM_IMG_FMT_I44416)) {
2207 if (!stream->config.cfg.monochrome) {
2208 stream->config.cfg.g_profile = 1;
2209 profile_updated = 1;
2210 }
2211 } else if (input.bit_depth == 12 ||
2212 ((input.fmt == AOM_IMG_FMT_I422 ||
2213 input.fmt == AOM_IMG_FMT_I42216) &&
2214 !stream->config.cfg.monochrome)) {
2215 stream->config.cfg.g_profile = 2;
2216 profile_updated = 1;
2217 }
2218 break;
2219 case 1:
2220 if (input.bit_depth == 12 || input.fmt == AOM_IMG_FMT_I422 ||
2221 input.fmt == AOM_IMG_FMT_I42216) {
2222 stream->config.cfg.g_profile = 2;
2223 profile_updated = 1;
2224 } else if (input.bit_depth < 12 &&
2225 (input.fmt == AOM_IMG_FMT_I420 ||
2226 input.fmt == AOM_IMG_FMT_I42016)) {
2227 stream->config.cfg.g_profile = 0;
2228 profile_updated = 1;
2229 }
2230 break;
2231 case 2:
2232 if (input.bit_depth < 12 && (input.fmt == AOM_IMG_FMT_I444 ||
2233 input.fmt == AOM_IMG_FMT_I44416)) {
2234 stream->config.cfg.g_profile = 1;
2235 profile_updated = 1;
2236 } else if (input.bit_depth < 12 &&
2237 (input.fmt == AOM_IMG_FMT_I420 ||
2238 input.fmt == AOM_IMG_FMT_I42016)) {
2239 stream->config.cfg.g_profile = 0;
2240 profile_updated = 1;
2241 } else if (input.bit_depth == 12 &&
2242 input.file_type == FILE_TYPE_Y4M) {
2243 // Note that here the input file values for chroma subsampling
2244 // are used instead of those from the command line.
2245 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2247 input.y4m.dst_c_dec_h >> 1);
2248 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2250 input.y4m.dst_c_dec_v >> 1);
2251 } else if (input.bit_depth == 12 &&
2252 input.file_type == FILE_TYPE_RAW) {
2253 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2255 stream->chroma_subsampling_x);
2256 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2258 stream->chroma_subsampling_y);
2259 }
2260 break;
2261 default: break;
2262 }
2263 }
2264 /* Automatically set the codec bit depth to match the input bit depth.
2265 * Upgrade the profile if required. */
2266 if (stream->config.cfg.g_input_bit_depth >
2267 (unsigned int)stream->config.cfg.g_bit_depth) {
2268 stream->config.cfg.g_bit_depth = stream->config.cfg.g_input_bit_depth;
2269 if (!global.quiet) {
2270 fprintf(stderr,
2271 "Warning: automatically updating bit depth to %d to "
2272 "match input format.\n",
2273 stream->config.cfg.g_input_bit_depth);
2274 }
2275 }
2276#if !CONFIG_AV1_HIGHBITDEPTH
2277 if (stream->config.cfg.g_bit_depth > 8) {
2278 fatal("Unsupported bit-depth with CONFIG_AV1_HIGHBITDEPTH=0\n");
2279 }
2280#endif // CONFIG_AV1_HIGHBITDEPTH
2281 if (stream->config.cfg.g_bit_depth > 10) {
2282 switch (stream->config.cfg.g_profile) {
2283 case 0:
2284 case 1:
2285 stream->config.cfg.g_profile = 2;
2286 profile_updated = 1;
2287 break;
2288 default: break;
2289 }
2290 }
2291 if (stream->config.cfg.g_bit_depth > 8) {
2292 stream->config.use_16bit_internal = 1;
2293 }
2294 if (profile_updated && !global.quiet) {
2295 fprintf(stderr,
2296 "Warning: automatically updating to profile %d to "
2297 "match input format.\n",
2298 stream->config.cfg.g_profile);
2299 }
2300 if ((global.show_psnr == 2) && (stream->config.cfg.g_input_bit_depth ==
2301 stream->config.cfg.g_bit_depth)) {
2302 fprintf(stderr,
2303 "Warning: --psnr==2 and --psnr==1 will provide same "
2304 "results when input bit-depth == stream bit-depth, "
2305 "falling back to default psnr value\n");
2306 global.show_psnr = 1;
2307 }
2308 if (global.show_psnr < 0 || global.show_psnr > 2) {
2309 fprintf(stderr,
2310 "Warning: --psnr can take only 0,1,2 as values,"
2311 "falling back to default psnr value\n");
2312 global.show_psnr = 1;
2313 }
2314 /* Set limit */
2315 stream->config.cfg.g_limit = global.limit;
2316 }
2317
2318 FOREACH_STREAM(stream, streams) {
2319 set_stream_dimensions(stream, input.width, input.height);
2320 stream->config.color_range = input.color_range;
2321 }
2322 FOREACH_STREAM(stream, streams) { validate_stream_config(stream, &global); }
2323
2324 /* Ensure that --passes and --pass are consistent. If --pass is set and
2325 * --passes >= 2, ensure --fpf was set.
2326 */
2327 if (global.pass > 0 && global.pass <= 3 && global.passes >= 2) {
2328 FOREACH_STREAM(stream, streams) {
2329 if (!stream->config.stats_fn)
2330 die("Stream %d: Must specify --fpf when --pass=%d"
2331 " and --passes=%d\n",
2332 stream->index, global.pass, global.passes);
2333 }
2334 }
2335
2336#if !CONFIG_WEBM_IO
2337 FOREACH_STREAM(stream, streams) {
2338 if (stream->config.write_webm) {
2339 stream->config.write_webm = 0;
2340 stream->config.write_ivf = 0;
2341 aom_tools_warn("aomenc compiled w/o WebM support. Writing OBU stream.");
2342 }
2343 }
2344#endif
2345
2346 /* Use the frame rate from the file only if none was specified
2347 * on the command-line.
2348 */
2349 if (!global.have_framerate) {
2350 global.framerate.num = input.framerate.numerator;
2351 global.framerate.den = input.framerate.denominator;
2352 }
2353 FOREACH_STREAM(stream, streams) {
2354 stream->config.cfg.g_timebase.den = global.framerate.num;
2355 stream->config.cfg.g_timebase.num = global.framerate.den;
2356 }
2357 /* Show configuration */
2358 if (global.verbose && pass == 0) {
2359 FOREACH_STREAM(stream, streams) {
2360 show_stream_config(stream, &global, &input);
2361 }
2362 }
2363
2364 if (pass == (global.pass ? global.pass - 1 : 0)) {
2365 // The Y4M reader does its own allocation.
2366 if (input.file_type != FILE_TYPE_Y4M) {
2367 aom_img_alloc(&raw, input.fmt, input.width, input.height, 32);
2368 }
2369 FOREACH_STREAM(stream, streams) {
2370 stream->rate_hist =
2371 init_rate_histogram(&stream->config.cfg, &global.framerate);
2372 }
2373 }
2374
2375 FOREACH_STREAM(stream, streams) { setup_pass(stream, &global, pass); }
2376 FOREACH_STREAM(stream, streams) { initialize_encoder(stream, &global); }
2377 FOREACH_STREAM(stream, streams) {
2378 char *encoder_settings = NULL;
2379#if CONFIG_WEBM_IO
2380 // Test frameworks may compare outputs from different versions, but only
2381 // wish to check for bitstream changes. The encoder-settings tag, however,
2382 // can vary if the version is updated, even if no encoder algorithm
2383 // changes were made. To work around this issue, do not output
2384 // the encoder-settings tag when --debug is enabled (which is the flag
2385 // that test frameworks should use, when they want deterministic output
2386 // from the container format).
2387 if (stream->config.write_webm && !stream->webm_ctx.debug) {
2388 encoder_settings = extract_encoder_settings(
2389 aom_codec_version_str(), argv_, argc, input.filename);
2390 if (encoder_settings == NULL) {
2391 fprintf(
2392 stderr,
2393 "Warning: unable to extract encoder settings. Continuing...\n");
2394 }
2395 }
2396#endif
2397 open_output_file(stream, &global, &input.pixel_aspect_ratio,
2398 encoder_settings);
2399 free(encoder_settings);
2400 }
2401
2402 if (strcmp(get_short_name_by_aom_encoder(global.codec), "av1") == 0) {
2403 // Check to see if at least one stream uses 16 bit internal.
2404 // Currently assume that the bit_depths for all streams using
2405 // highbitdepth are the same.
2406 FOREACH_STREAM(stream, streams) {
2407 if (stream->config.use_16bit_internal) {
2408 do_16bit_internal = 1;
2409 }
2410 input_shift = (int)stream->config.cfg.g_bit_depth -
2411 stream->config.cfg.g_input_bit_depth;
2412 }
2413 }
2414
2415 frame_avail = 1;
2416 got_data = 0;
2417
2418 while (frame_avail || got_data) {
2419 struct aom_usec_timer timer;
2420
2421 if (!global.limit || frames_in < global.limit) {
2422 frame_avail = read_frame(&input, &raw);
2423
2424 if (frame_avail) frames_in++;
2425 seen_frames =
2426 frames_in > global.skip_frames ? frames_in - global.skip_frames : 0;
2427
2428 if (!global.quiet) {
2429 float fps = usec_to_fps(cx_time, seen_frames);
2430 fprintf(stderr, "\rPass %d/%d ", pass + 1, global.passes);
2431
2432 if (stream_cnt == 1)
2433 fprintf(stderr, "frame %4d/%-4d %7" PRId64 "B ", frames_in,
2434 streams->frames_out, (int64_t)streams->nbytes);
2435 else
2436 fprintf(stderr, "frame %4d ", frames_in);
2437
2438 fprintf(stderr, "%7" PRId64 " %s %.2f %s ",
2439 cx_time > 9999999 ? cx_time / 1000 : cx_time,
2440 cx_time > 9999999 ? "ms" : "us", fps >= 1.0 ? fps : fps * 60,
2441 fps >= 1.0 ? "fps" : "fpm");
2442 print_time("ETA", estimated_time_left);
2443 // mingw-w64 gcc does not match msvc for stderr buffering behavior
2444 // and uses line buffering, thus the progress output is not
2445 // real-time. The fflush() is here to make sure the progress output
2446 // is sent out while the clip is being processed.
2447 fflush(stderr);
2448 }
2449
2450 } else {
2451 frame_avail = 0;
2452 }
2453
2454 if (frames_in > global.skip_frames) {
2455 aom_image_t *frame_to_encode;
2456 if (input_shift || (do_16bit_internal && input.bit_depth == 8)) {
2457 assert(do_16bit_internal);
2458 // Input bit depth and stream bit depth do not match, so up
2459 // shift frame to stream bit depth
2460 if (!allocated_raw_shift) {
2461 aom_img_alloc(&raw_shift, raw.fmt | AOM_IMG_FMT_HIGHBITDEPTH,
2462 input.width, input.height, 32);
2463 allocated_raw_shift = 1;
2464 }
2465 aom_img_upshift(&raw_shift, &raw, input_shift);
2466 frame_to_encode = &raw_shift;
2467 } else {
2468 frame_to_encode = &raw;
2469 }
2470 aom_usec_timer_start(&timer);
2471 if (do_16bit_internal) {
2472 assert(frame_to_encode->fmt & AOM_IMG_FMT_HIGHBITDEPTH);
2473 FOREACH_STREAM(stream, streams) {
2474 if (stream->config.use_16bit_internal)
2475 encode_frame(stream, &global,
2476 frame_avail ? frame_to_encode : NULL, frames_in);
2477 else
2478 assert(0);
2479 }
2480 } else {
2481 assert((frame_to_encode->fmt & AOM_IMG_FMT_HIGHBITDEPTH) == 0);
2482 FOREACH_STREAM(stream, streams) {
2483 encode_frame(stream, &global, frame_avail ? frame_to_encode : NULL,
2484 frames_in);
2485 }
2486 }
2487 aom_usec_timer_mark(&timer);
2488 cx_time += aom_usec_timer_elapsed(&timer);
2489
2490 FOREACH_STREAM(stream, streams) { update_quantizer_histogram(stream); }
2491
2492 got_data = 0;
2493 FOREACH_STREAM(stream, streams) {
2494 get_cx_data(stream, &global, &got_data);
2495 }
2496
2497 if (!got_data && input.length && streams != NULL &&
2498 !streams->frames_out) {
2499 lagged_count = global.limit ? seen_frames : ftello(input.file);
2500 } else if (input.length) {
2501 int64_t remaining;
2502 int64_t rate;
2503
2504 if (global.limit) {
2505 const int64_t frame_in_lagged = (seen_frames - lagged_count) * 1000;
2506
2507 rate = cx_time ? frame_in_lagged * (int64_t)1000000 / cx_time : 0;
2508 remaining = 1000 * (global.limit - global.skip_frames -
2509 seen_frames + lagged_count);
2510 } else {
2511 const int64_t input_pos = ftello(input.file);
2512 const int64_t input_pos_lagged = input_pos - lagged_count;
2513 const int64_t input_limit = input.length;
2514
2515 rate = cx_time ? input_pos_lagged * (int64_t)1000000 / cx_time : 0;
2516 remaining = input_limit - input_pos + lagged_count;
2517 }
2518
2519 average_rate =
2520 (average_rate <= 0) ? rate : (average_rate * 7 + rate) / 8;
2521 estimated_time_left = average_rate ? remaining / average_rate : -1;
2522 }
2523
2524 if (got_data && global.test_decode != TEST_DECODE_OFF) {
2525 FOREACH_STREAM(stream, streams) {
2526 test_decode(stream, global.test_decode);
2527 }
2528 }
2529 }
2530
2531 fflush(stdout);
2532 if (!global.quiet) fprintf(stderr, "\033[K");
2533 }
2534
2535 if (stream_cnt > 1) fprintf(stderr, "\n");
2536
2537 if (!global.quiet) {
2538 FOREACH_STREAM(stream, streams) {
2539 const int64_t bpf =
2540 seen_frames ? (int64_t)(stream->nbytes * 8 / seen_frames) : 0;
2541 const int64_t bps = bpf * global.framerate.num / global.framerate.den;
2542 fprintf(stderr,
2543 "\rPass %d/%d frame %4d/%-4d %7" PRId64 "B %7" PRId64
2544 "b/f %7" PRId64
2545 "b/s"
2546 " %7" PRId64 " %s (%.2f fps)\033[K\n",
2547 pass + 1, global.passes, frames_in, stream->frames_out,
2548 (int64_t)stream->nbytes, bpf, bps,
2549 stream->cx_time > 9999999 ? stream->cx_time / 1000
2550 : stream->cx_time,
2551 stream->cx_time > 9999999 ? "ms" : "us",
2552 usec_to_fps(stream->cx_time, seen_frames));
2553 // This instance of cr does not need fflush as it is followed by a
2554 // newline in the same string.
2555 }
2556 }
2557
2558 if (global.show_psnr >= 1) {
2559 if (get_fourcc_by_aom_encoder(global.codec) == AV1_FOURCC) {
2560 FOREACH_STREAM(stream, streams) {
2561 int64_t bps = 0;
2562 if (global.show_psnr == 1) {
2563 if (stream->psnr_count[0] && seen_frames && global.framerate.den) {
2564 bps = (int64_t)stream->nbytes * 8 *
2565 (int64_t)global.framerate.num / global.framerate.den /
2566 seen_frames;
2567 }
2568 show_psnr(stream, (1 << stream->config.cfg.g_input_bit_depth) - 1,
2569 bps);
2570 }
2571 if (global.show_psnr == 2) {
2572#if CONFIG_AV1_HIGHBITDEPTH
2573 if (stream->config.cfg.g_input_bit_depth <
2574 (unsigned int)stream->config.cfg.g_bit_depth)
2575 show_psnr_hbd(stream, (1 << stream->config.cfg.g_bit_depth) - 1,
2576 bps);
2577#endif
2578 }
2579 }
2580 } else {
2581 FOREACH_STREAM(stream, streams) { show_psnr(stream, 255.0, 0); }
2582 }
2583 }
2584
2585 if (pass == global.passes - 1) {
2586 FOREACH_STREAM(stream, streams) {
2587 int num_operating_points;
2588 int levels[32];
2589 int target_levels[32];
2591 &num_operating_points);
2592 aom_codec_control(&stream->encoder, AV1E_GET_SEQ_LEVEL_IDX, levels);
2594 target_levels);
2595
2596 for (int i = 0; i < num_operating_points; i++) {
2597 if (levels[i] > target_levels[i]) {
2598 if (levels[i] == 31) {
2599 aom_tools_warn(
2600 "Failed to encode to target level %d.%d for operating point "
2601 "%d. The output level is SEQ_LEVEL_MAX",
2602 2 + (target_levels[i] >> 2), target_levels[i] & 3, i);
2603 } else {
2604 aom_tools_warn(
2605 "Failed to encode to target level %d.%d for operating point "
2606 "%d. The output level is %d.%d",
2607 2 + (target_levels[i] >> 2), target_levels[i] & 3, i,
2608 2 + (levels[i] >> 2), levels[i] & 3);
2609 }
2610 }
2611 }
2612 }
2613 }
2614
2615 FOREACH_STREAM(stream, streams) { aom_codec_destroy(&stream->encoder); }
2616
2617 if (global.test_decode != TEST_DECODE_OFF) {
2618 FOREACH_STREAM(stream, streams) { aom_codec_destroy(&stream->decoder); }
2619 }
2620
2621 close_input_file(&input);
2622
2623 if (global.test_decode == TEST_DECODE_FATAL) {
2624 FOREACH_STREAM(stream, streams) { res |= stream->mismatch_seen; }
2625 }
2626 FOREACH_STREAM(stream, streams) {
2627 close_output_file(stream, get_fourcc_by_aom_encoder(global.codec));
2628 }
2629
2630 FOREACH_STREAM(stream, streams) {
2631 stats_close(&stream->stats, global.passes - 1);
2632 }
2633
2634 if (global.pass) break;
2635 }
2636
2637 if (global.show_q_hist_buckets) {
2638 FOREACH_STREAM(stream, streams) {
2639 show_q_histogram(stream->counts, global.show_q_hist_buckets);
2640 }
2641 }
2642
2643 if (global.show_rate_hist_buckets) {
2644 FOREACH_STREAM(stream, streams) {
2645 show_rate_histogram(stream->rate_hist, &stream->config.cfg,
2646 global.show_rate_hist_buckets);
2647 }
2648 }
2649 FOREACH_STREAM(stream, streams) { destroy_rate_histogram(stream->rate_hist); }
2650
2651#if CONFIG_INTERNAL_STATS
2652 /* TODO(jkoleszar): This doesn't belong in this executable. Do it for now,
2653 * to match some existing utilities.
2654 */
2655 if (!(global.pass == 1 && global.passes == 2)) {
2656 FOREACH_STREAM(stream, streams) {
2657 FILE *f = fopen("opsnr.stt", "a");
2658 if (stream->mismatch_seen) {
2659 fprintf(f, "First mismatch occurred in frame %d\n",
2660 stream->mismatch_seen);
2661 } else {
2662 fprintf(f, "No mismatch detected in recon buffers\n");
2663 }
2664 fclose(f);
2665 }
2666 }
2667#endif
2668
2669 if (allocated_raw_shift) aom_img_free(&raw_shift);
2670 aom_img_free(&raw);
2671 free(argv);
2672 free(streams);
2673 return res ? EXIT_FAILURE : EXIT_SUCCESS;
2674}
Describes the decoder algorithm interface to applications.
Describes the encoder algorithm interface to applications.
#define MAX_TILE_WIDTHS
Maximum number of tile widths in tile widths array.
Definition aom_encoder.h:861
#define MAX_TILE_HEIGHTS
Maximum number of tile heights in tile heights array.
Definition aom_encoder.h:874
#define AOM_PLANE_U
Definition aom_image.h:209
@ AOM_CSP_UNKNOWN
Definition aom_image.h:142
enum aom_chroma_sample_position aom_chroma_sample_position_t
List of chroma sample positions.
#define AOM_PLANE_Y
Definition aom_image.h:208
#define AOM_PLANE_V
Definition aom_image.h:210
enum aom_color_range aom_color_range_t
List of supported color range.
#define AOM_IMG_FMT_HIGHBITDEPTH
Definition aom_image.h:38
aom_image_t * aom_img_alloc(aom_image_t *img, aom_img_fmt_t fmt, unsigned int d_w, unsigned int d_h, unsigned int align)
Open a descriptor, allocating storage for the underlying image.
@ AOM_IMG_FMT_I42216
Definition aom_image.h:58
@ AOM_IMG_FMT_I42016
Definition aom_image.h:56
@ AOM_IMG_FMT_I444
Definition aom_image.h:50
@ AOM_IMG_FMT_I422
Definition aom_image.h:49
@ AOM_IMG_FMT_I44416
Definition aom_image.h:59
@ AOM_IMG_FMT_I420
Definition aom_image.h:45
@ AOM_IMG_FMT_NV12
Definition aom_image.h:54
@ AOM_IMG_FMT_YV12
Definition aom_image.h:43
void aom_img_free(aom_image_t *img)
Close an image descriptor.
Provides definitions for using AOM or AV1 encoder algorithm within the aom Codec Interface.
Provides definitions for using AOM or AV1 within the aom Decoder interface.
@ AV1_SET_TILE_MODE
Codec control function to set the tile coding mode, unsigned int parameter.
Definition aomdx.h:313
@ AV1D_SET_IS_ANNEXB
Codec control function to indicate whether bitstream is in Annex-B format, unsigned int parameter.
Definition aomdx.h:349
@ AV1_SET_DECODE_TILE_ROW
Codec control function to set the range of tile decoding, int parameter.
Definition aomdx.h:304
@ AV1E_SET_MATRIX_COEFFICIENTS
Codec control function to set transfer function info, int parameter.
Definition aomcx.h:573
@ AV1E_SET_ENABLE_INTERINTER_WEDGE
Codec control function to turn on / off interinter wedge compound, int parameter.
Definition aomcx.h:1010
@ AV1E_SET_ENABLE_DIAGONAL_INTRA
Codec control function to turn on / off D45 to D203 intra mode usage, int parameter.
Definition aomcx.h:1348
@ AV1E_SET_MAX_GF_INTERVAL
Codec control function to set minimum interval between GF/ARF frames, unsigned int parameter.
Definition aomcx.h:594
@ AV1E_SET_ROW_MT
Codec control function to enable the row based multi-threading of the encoder, unsigned int parameter...
Definition aomcx.h:361
@ AV1E_SET_ENABLE_SMOOTH_INTRA
Codec control function to turn on / off smooth intra modes usage, int parameter.
Definition aomcx.h:1070
@ AOME_SET_SHARPNESS
Codec control function to set the sharpness parameter, unsigned int parameter.
Definition aomcx.h:241
@ AV1E_GET_TARGET_SEQ_LEVEL_IDX
Codec control function to get the target sequence level index for each operating point....
Definition aomcx.h:1455
@ AV1E_SET_RATE_DISTRIBUTION_INFO
Codec control to set the input file for rate distribution used in all intra mode, const char * parame...
Definition aomcx.h:1514
@ AV1E_SET_ENABLE_TPL_MODEL
Codec control function to enable RDO modulated by frame temporal dependency, unsigned int parameter.
Definition aomcx.h:408
@ AOME_GET_LAST_QUANTIZER_64
Codec control function to get last quantizer chosen by the encoder, int* parameter.
Definition aomcx.h:263
@ AV1E_SET_AQ_MODE
Codec control function to set adaptive quantization mode, unsigned int parameter.
Definition aomcx.h:468
@ AV1E_SET_REDUCED_REFERENCE_SET
Control to use reduced set of single and compound references, int parameter.
Definition aomcx.h:1224
@ AV1E_GET_NUM_OPERATING_POINTS
Codec control function to get the number of operating points. int* parameter.
Definition aomcx.h:1460
@ AV1E_SET_GF_MIN_PYRAMID_HEIGHT
Control to select minimum height for the GF group pyramid structure, unsigned int parameter.
Definition aomcx.h:1320
@ AV1E_SET_ENABLE_PAETH_INTRA
Codec control function to turn on / off Paeth intra mode usage, int parameter.
Definition aomcx.h:1078
@ AV1E_SET_TUNE_CONTENT
Codec control function to set content type, aom_tune_content parameter.
Definition aomcx.h:497
@ AV1E_SET_CDF_UPDATE_MODE
Codec control function to set CDF update mode, unsigned int parameter.
Definition aomcx.h:506
@ AV1E_SET_CHROMA_SUBSAMPLING_X
Sets the chroma subsampling x value, unsigned int parameter.
Definition aomcx.h:1187
@ AV1E_SET_COLOR_RANGE
Codec control function to set color range bit, int parameter.
Definition aomcx.h:606
@ AV1E_SET_ENABLE_RESTORATION
Codec control function to encode with Loop Restoration Filter, unsigned int parameter.
Definition aomcx.h:680
@ AV1E_SET_ENABLE_ANGLE_DELTA
Codec control function to turn on/off intra angle delta, int parameter.
Definition aomcx.h:1117
@ AV1E_SET_MIN_GF_INTERVAL
Codec control function to set minimum interval between GF/ARF frames, unsigned int parameter.
Definition aomcx.h:587
@ AOME_SET_ARNR_MAXFRAMES
Codec control function to set the max no of frames to create arf, unsigned int parameter.
Definition aomcx.h:268
@ AV1E_SET_MV_COST_UPD_FREQ
Control to set frequency of the cost updates for motion vectors, unsigned int parameter.
Definition aomcx.h:1254
@ AV1E_SET_INTRA_DEFAULT_TX_ONLY
Control to use default tx type only for intra modes, int parameter.
Definition aomcx.h:1203
@ AV1E_SET_TRANSFER_CHARACTERISTICS
Codec control function to set transfer function info, int parameter.
Definition aomcx.h:552
@ AV1E_SET_MTU
Codec control function to set an MTU size for a tile group, unsigned int parameter.
Definition aomcx.h:800
@ AV1E_SET_DISABLE_TRELLIS_QUANT
Codec control function to encode without trellis quantization, unsigned int parameter.
Definition aomcx.h:707
@ AV1E_SET_ENABLE_INTRABC
Codec control function to turn on/off intra block copy mode, int parameter.
Definition aomcx.h:1113
@ AV1E_SET_ENABLE_AB_PARTITIONS
Codec control function to enable/disable AB partitions, int parameter.
Definition aomcx.h:818
@ AV1E_SET_ENABLE_INTERINTRA_COMP
Codec control function to turn on / off interintra compound for a sequence, int parameter.
Definition aomcx.h:986
@ AV1E_SET_FILM_GRAIN_TEST_VECTOR
Codec control function to add film grain parameters (one of several preset types) info in the bitstre...
Definition aomcx.h:1173
@ AV1E_SET_ENABLE_CHROMA_DELTAQ
Codec control function to turn on / off delta quantization in chroma planes for a sequence,...
Definition aomcx.h:962
@ AV1E_SET_ENABLE_DUAL_FILTER
Codec control function to turn on / off dual interpolation filter for a sequence, int parameter.
Definition aomcx.h:954
@ AV1E_SET_FRAME_PARALLEL_DECODING
Codec control function to enable frame parallel decoding feature, unsigned int parameter.
Definition aomcx.h:431
@ AV1E_SET_MIN_PARTITION_SIZE
Codec control function to set min partition size, int parameter.
Definition aomcx.h:837
@ AV1E_SET_ENABLE_WARPED_MOTION
Codec control function to turn on / off warped motion usage at sequence level, int parameter.
Definition aomcx.h:1038
@ AV1E_SET_FORCE_VIDEO_MODE
Codec control function to force video mode, unsigned int parameter.
Definition aomcx.h:687
@ AV1E_SET_CHROMA_SUBSAMPLING_Y
Sets the chroma subsampling y value, unsigned int parameter.
Definition aomcx.h:1190
@ AV1E_SET_ENABLE_INTRA_EDGE_FILTER
Codec control function to turn on / off intra edge filter at sequence level, int parameter.
Definition aomcx.h:856
@ AV1E_SET_COEFF_COST_UPD_FREQ
Control to set frequency of the cost updates for coefficients, unsigned int parameter.
Definition aomcx.h:1234
@ AV1E_SET_ENABLE_DIRECTIONAL_INTRA
Codec control function to turn on / off directional intra mode usage, int parameter.
Definition aomcx.h:1377
@ AV1E_SET_MAX_INTER_BITRATE_PCT
Codec control function to set max data rate for inter frames, unsigned int parameter.
Definition aomcx.h:325
@ AV1E_SET_DENOISE_NOISE_LEVEL
Sets the noise level, int parameter.
Definition aomcx.h:1181
@ AV1E_SET_INTRA_DCT_ONLY
Control to use dct only for intra modes, int parameter.
Definition aomcx.h:1196
@ AV1E_SET_TILE_ROWS
Codec control function to set number of tile rows, unsigned int parameter.
Definition aomcx.h:398
@ AV1E_SET_ENABLE_REF_FRAME_MVS
Codec control function to turn on / off ref frame mvs (mfmv) usage at sequence level,...
Definition aomcx.h:935
@ AV1E_SET_FP_MT
Codec control function to enable frame parallel multi-threading of the encoder, unsigned int paramete...
Definition aomcx.h:1435
@ AV1E_SET_ENABLE_MASKED_COMP
Codec control function to turn on / off masked compound usage (wedge and diff-wtd compound modes) for...
Definition aomcx.h:970
@ AV1E_SET_VBR_CORPUS_COMPLEXITY_LAP
Control to set average complexity of the corpus in the case of single pass vbr based on LAP,...
Definition aomcx.h:1325
@ AV1E_SET_GF_MAX_PYRAMID_HEIGHT
Control to select maximum height for the GF group pyramid structure, unsigned int parameter.
Definition aomcx.h:1213
@ AV1E_SET_ENABLE_CDEF
Codec control function to encode with CDEF, unsigned int parameter.
Definition aomcx.h:670
@ AV1E_SET_ENABLE_FLIP_IDTX
Codec control function to turn on / off flip and identity transforms, int parameter.
Definition aomcx.h:900
@ AV1E_GET_SEQ_LEVEL_IDX
Codec control function to get sequence level index for each operating point. int* parameter....
Definition aomcx.h:643
@ AV1E_SET_FRAME_PERIODIC_BOOST
Codec control function to enable/disable periodic Q boost, unsigned int parameter.
Definition aomcx.h:480
@ AV1E_SET_DV_COST_UPD_FREQ
Control to set frequency of the cost updates for intrabc motion vectors, unsigned int parameter.
Definition aomcx.h:1358
@ AV1E_SET_AUTO_INTRA_TOOLS_OFF
Codec control to automatically turn off several intra coding tools, unsigned int parameter.
Definition aomcx.h:1419
@ AV1E_SET_ENABLE_RECT_TX
Codec control function to turn on / off rectangular transforms, int parameter.
Definition aomcx.h:912
@ AV1E_SET_ENABLE_DIST_WTD_COMP
Codec control function to turn on / off dist-wtd compound mode at sequence level, int parameter.
Definition aomcx.h:924
@ AV1E_SET_TIMING_INFO_TYPE
Codec control function to signal picture timing info in the bitstream, aom_timing_info_type_t paramet...
Definition aomcx.h:1166
@ AV1E_SET_SUPERBLOCK_SIZE
Codec control function to set intended superblock size, unsigned int parameter.
Definition aomcx.h:651
@ AV1E_SET_TIER_MASK
Control to set bit mask that specifies which tier each of the 32 possible operating points conforms t...
Definition aomcx.h:1262
@ AV1E_SET_ENABLE_INTERINTRA_WEDGE
Codec control function to turn on / off interintra wedge compound, int parameter.
Definition aomcx.h:1018
@ AV1E_SET_NOISE_SENSITIVITY
Codec control function to set noise sensitivity, unsigned int parameter.
Definition aomcx.h:488
@ AV1E_SET_ENABLE_DIFF_WTD_COMP
Codec control function to turn on / off difference weighted compound, int parameter.
Definition aomcx.h:1002
@ AV1E_SET_QUANT_B_ADAPT
Control to use adaptive quantize_b, int parameter.
Definition aomcx.h:1206
@ AV1E_SET_ENABLE_FILTER_INTRA
Codec control function to turn on / off filter intra usage at sequence level, int parameter.
Definition aomcx.h:1059
@ AV1E_SET_ENABLE_PALETTE
Codec control function to turn on/off palette mode, int parameter.
Definition aomcx.h:1109
@ AV1E_SET_ENABLE_CFL_INTRA
Codec control function to turn on / off CFL uv intra mode usage, int parameter.
Definition aomcx.h:1088
@ AV1E_SET_ENABLE_KEYFRAME_FILTERING
Codec control function to enable temporal filtering on key frame, unsigned int parameter.
Definition aomcx.h:417
@ AV1E_SET_NUM_TG
Codec control function to set a maximum number of tile groups, unsigned int parameter.
Definition aomcx.h:789
@ AOME_SET_MAX_INTRA_BITRATE_PCT
Codec control function to set max data rate for intra frames, unsigned int parameter.
Definition aomcx.h:306
@ AV1E_SET_ERROR_RESILIENT_MODE
Codec control function to enable error_resilient_mode, int parameter.
Definition aomcx.h:442
@ AV1E_SET_ENABLE_SMOOTH_INTERINTRA
Codec control function to turn on / off smooth inter-intra mode for a sequence, int parameter.
Definition aomcx.h:994
@ AOME_SET_STATIC_THRESHOLD
Codec control function to set the threshold for MBs treated static, unsigned int parameter.
Definition aomcx.h:246
@ AV1E_SET_ENABLE_OBMC
Codec control function to predict with OBMC mode, unsigned int parameter.
Definition aomcx.h:697
@ AV1E_SET_PARTITION_INFO_PATH
Codec control to set the path for partition stats read and write. const char * parameter.
Definition aomcx.h:1363
@ AV1E_SET_MAX_PARTITION_SIZE
Codec control function to set max partition size, int parameter.
Definition aomcx.h:848
@ AV1E_SET_ENABLE_1TO4_PARTITIONS
Codec control function to enable/disable 1:4 and 4:1 partitions, int parameter.
Definition aomcx.h:826
@ AV1E_SET_DELTALF_MODE
Codec control function to turn on/off loopfilter modulation when delta q modulation is enabled,...
Definition aomcx.h:1139
@ AV1E_SET_ENABLE_TX64
Codec control function to turn on / off 64-length transforms, int parameter.
Definition aomcx.h:876
@ AOME_SET_TUNING
Codec control function to set visual tuning, aom_tune_metric (int) parameter.
Definition aomcx.h:282
@ AV1E_SET_TARGET_SEQ_LEVEL_IDX
Control to set target sequence level index for a certain operating point (OP), int parameter Possible...
Definition aomcx.h:636
@ AV1E_SET_CHROMA_SAMPLE_POSITION
Codec control function to set chroma 4:2:0 sample position info, aom_chroma_sample_position_t paramet...
Definition aomcx.h:580
@ AV1E_SET_REDUCED_TX_TYPE_SET
Control to use a reduced tx type set, int parameter.
Definition aomcx.h:1193
@ AV1E_SET_DELTAQ_STRENGTH
Set –deltaq-mode strength.
Definition aomcx.h:1398
@ AV1E_SET_INTER_DCT_ONLY
Control to use dct only for inter modes, int parameter.
Definition aomcx.h:1199
@ AV1E_SET_LOOPFILTER_CONTROL
Codec control to control loop filter.
Definition aomcx.h:1407
@ AOME_SET_ENABLEAUTOALTREF
Codec control function to enable automatic set and use alf frames, unsigned int parameter.
Definition aomcx.h:228
@ AV1E_ENABLE_RATE_GUIDE_DELTAQ
Codec control to enable the rate distribution guided delta quantization in all intra mode,...
Definition aomcx.h:1502
@ AV1E_SET_TILE_COLUMNS
Codec control function to set number of tile columns. unsigned int parameter.
Definition aomcx.h:380
@ AV1E_SET_ENABLE_ORDER_HINT
Codec control function to turn on / off frame order hint (int parameter). Affects: joint compound mod...
Definition aomcx.h:865
@ AV1E_SET_DELTAQ_MODE
Codec control function to set the delta q mode, unsigned int parameter.
Definition aomcx.h:1131
@ AV1E_SET_ENABLE_GLOBAL_MOTION
Codec control function to turn on / off global motion usage for a sequence, int parameter.
Definition aomcx.h:1028
@ AV1E_SET_FILM_GRAIN_TABLE
Codec control function to set the path to the film grain parameters, const char* parameter.
Definition aomcx.h:1178
@ AV1E_SET_QM_MAX
Codec control function to set the max quant matrix flatness, unsigned int parameter.
Definition aomcx.h:743
@ AV1E_SET_MAX_REFERENCE_FRAMES
Control to select maximum reference frames allowed per frame, int parameter.
Definition aomcx.h:1220
@ AOME_SET_CPUUSED
Codec control function to set encoder internal speed settings, int parameter.
Definition aomcx.h:220
@ AV1E_SET_GF_CBR_BOOST_PCT
Boost percentage for Golden Frame in CBR mode, unsigned int parameter.
Definition aomcx.h:339
@ AV1E_SET_ENABLE_ONESIDED_COMP
Codec control function to turn on / off one sided compound usage for a sequence, int parameter.
Definition aomcx.h:978
@ AV1E_SET_DENOISE_BLOCK_SIZE
Sets the denoisers block size, unsigned int parameter.
Definition aomcx.h:1184
@ AV1E_SET_VMAF_MODEL_PATH
Codec control function to set the path to the VMAF model used when tuning the encoder for VMAF,...
Definition aomcx.h:1292
@ AV1E_SET_QM_MIN
Codec control function to set the min quant matrix flatness, unsigned int parameter.
Definition aomcx.h:731
@ AV1E_SET_ENABLE_QM
Codec control function to encode with quantisation matrices, unsigned int parameter.
Definition aomcx.h:718
@ AV1E_SET_ENABLE_OVERLAY
Codec control function to turn on / off overlay frames for filtered ALTREF frames,...
Definition aomcx.h:1106
@ AV1E_SET_ENABLE_RECT_PARTITIONS
Codec control function to enable/disable rectangular partitions, int parameter.
Definition aomcx.h:810
@ AV1E_SET_COLOR_PRIMARIES
Codec control function to set color space info, int parameter.
Definition aomcx.h:527
@ AOME_SET_CQ_LEVEL
Codec control function to set constrained / constant quality level, unsigned int parameter.
Definition aomcx.h:292
@ AV1E_SET_ENABLE_TX_SIZE_SEARCH
Control to turn on / off transform size search. Note: it can not work with non RD pick mode in real-t...
Definition aomcx.h:1387
@ AV1E_SET_MODE_COST_UPD_FREQ
Control to set frequency of the cost updates for mode, unsigned int parameter.
Definition aomcx.h:1244
@ AV1E_SET_MIN_CR
Control to set minimum compression ratio, unsigned int parameter Take integer values....
Definition aomcx.h:1269
@ AV1E_SET_LOSSLESS
Codec control function to set lossless encoding mode, unsigned int parameter.
Definition aomcx.h:353
@ AOME_SET_ARNR_STRENGTH
Codec control function to set the filter strength for the arf, unsigned int parameter.
Definition aomcx.h:273
@ AV1_GET_NEW_FRAME_IMAGE
Codec control function to get a pointer to the new frame.
Definition aom.h:70
const char * aom_codec_iface_name(aom_codec_iface_t *iface)
Return the name for a given interface.
aom_codec_err_t aom_codec_control(aom_codec_ctx_t *ctx, int ctrl_id,...)
Algorithm Control.
const struct aom_codec_iface aom_codec_iface_t
Codec interface structure.
Definition aom_codec.h:254
const char * aom_codec_version_str(void)
Return the version information (as a string)
aom_codec_err_t aom_codec_set_option(aom_codec_ctx_t *ctx, const char *name, const char *value)
Key & Value API.
const char * aom_codec_error(const aom_codec_ctx_t *ctx)
Retrieve error synopsis for codec context.
int64_t aom_codec_pts_t
Time Stamp Type.
Definition aom_codec.h:235
aom_codec_err_t aom_codec_destroy(aom_codec_ctx_t *ctx)
Destroy a codec instance.
const char * aom_codec_err_to_string(aom_codec_err_t err)
Convert error number to printable string.
aom_codec_err_t
Algorithm return codes.
Definition aom_codec.h:155
#define AOM_CODEC_CONTROL_TYPECHECKED(ctx, id, data)
aom_codec_control wrapper macro (adds type-checking, less flexible)
Definition aom_codec.h:525
const char * aom_codec_error_detail(const aom_codec_ctx_t *ctx)
Retrieve detailed error information for codec context.
const void * aom_codec_iter_t
Iterator.
Definition aom_codec.h:288
@ AOM_BITS_8
Definition aom_codec.h:319
aom_codec_err_t aom_codec_decode(aom_codec_ctx_t *ctx, const uint8_t *data, size_t data_sz, void *user_priv)
Decode data.
#define aom_codec_dec_init(ctx, iface, cfg, flags)
Convenience macro for aom_codec_dec_init_ver()
Definition aom_decoder.h:129
#define AOM_USAGE_GOOD_QUALITY
usage parameter analogous to AV1 GOOD QUALITY mode.
Definition aom_encoder.h:1009
#define AOM_USAGE_ALL_INTRA
usage parameter analogous to AV1 all intra mode.
Definition aom_encoder.h:1013
const aom_codec_cx_pkt_t * aom_codec_get_cx_data(aom_codec_ctx_t *ctx, aom_codec_iter_t *iter)
Encoded data iterator.
aom_codec_err_t aom_codec_encode(aom_codec_ctx_t *ctx, const aom_image_t *img, aom_codec_pts_t pts, unsigned long duration, aom_enc_frame_flags_t flags)
Encode a frame.
#define aom_codec_enc_init(ctx, iface, cfg, flags)
Convenience macro for aom_codec_enc_init_ver()
Definition aom_encoder.h:938
aom_codec_err_t aom_codec_enc_config_default(aom_codec_iface_t *iface, aom_codec_enc_cfg_t *cfg, unsigned int usage)
Get the default configuration for a usage.
#define AOM_USAGE_REALTIME
usage parameter analogous to AV1 REALTIME mode.
Definition aom_encoder.h:1011
#define AOM_CODEC_USE_HIGHBITDEPTH
Definition aom_encoder.h:80
#define AOM_CODEC_USE_PSNR
Initialization-time Feature Enabling.
Definition aom_encoder.h:79
@ AOM_RC_ONE_PASS
Definition aom_encoder.h:175
@ AOM_RC_SECOND_PASS
Definition aom_encoder.h:177
@ AOM_RC_THIRD_PASS
Definition aom_encoder.h:178
@ AOM_RC_FIRST_PASS
Definition aom_encoder.h:176
@ AOM_KF_DISABLED
Definition aom_encoder.h:201
@ AOM_CODEC_PSNR_PKT
Definition aom_encoder.h:111
@ AOM_CODEC_CX_FRAME_PKT
Definition aom_encoder.h:108
@ AOM_CODEC_STATS_PKT
Definition aom_encoder.h:109
Codec context structure.
Definition aom_codec.h:298
Encoder output packet.
Definition aom_encoder.h:120
size_t sz
Definition aom_encoder.h:125
enum aom_codec_cx_pkt_kind kind
Definition aom_encoder.h:121
double psnr[4]
Definition aom_encoder.h:143
aom_fixed_buf_t twopass_stats
Definition aom_encoder.h:138
aom_fixed_buf_t raw
Definition aom_encoder.h:154
union aom_codec_cx_pkt::@1 data
aom_codec_pts_t pts
time stamp to show frame (in timebase units)
Definition aom_encoder.h:127
struct aom_codec_cx_pkt::@1::@2 frame
int partition_id
the partition id defines the decoding order of the partitions. Only applicable when "output partition...
Definition aom_encoder.h:134
void * buf
Definition aom_encoder.h:124
Initialization Configurations.
Definition aom_decoder.h:91
Encoder configuration structure.
Definition aom_encoder.h:385
struct aom_rational g_timebase
Stream timebase units.
Definition aom_encoder.h:487
unsigned int g_h
Height of the frame.
Definition aom_encoder.h:433
unsigned int monochrome
Monochrome mode.
Definition aom_encoder.h:820
unsigned int g_w
Width of the frame.
Definition aom_encoder.h:424
enum aom_enc_pass g_pass
Multi-pass Encoding Mode.
Definition aom_encoder.h:502
size_t sz
Definition aom_encoder.h:88
void * buf
Definition aom_encoder.h:87
Image Descriptor.
Definition aom_image.h:180
aom_chroma_sample_position_t csp
Definition aom_image.h:186
unsigned int y_chroma_shift
Definition aom_image.h:204
aom_img_fmt_t fmt
Definition aom_image.h:181
int stride[3]
Definition aom_image.h:214
unsigned char * img_data
Definition aom_image.h:228
unsigned int x_chroma_shift
Definition aom_image.h:203
unsigned int d_w
Definition aom_image.h:195
int bps
Definition aom_image.h:217
int monochrome
Definition aom_image.h:185
unsigned int d_h
Definition aom_image.h:196
unsigned char * planes[3]
Definition aom_image.h:213
int img_data_owner
Definition aom_image.h:229
int self_allocd
Definition aom_image.h:230
size_t sz
Definition aom_image.h:215
Rational Number.
Definition aom_encoder.h:162
int num
Definition aom_encoder.h:163
int den
Definition aom_encoder.h:164
Encoder Config Options.
Definition aom_encoder.h:225
unsigned int min_partition_size
min partition size 8, 16, 32, 64, 128
Definition aom_encoder.h:241
unsigned int max_partition_size
max partition size 8, 16, 32, 64, 128
Definition aom_encoder.h:237
unsigned int disable_trellis_quant
disable trellis quantization
Definition aom_encoder.h:353
unsigned int super_block_size
Superblock size 0, 64 or 128.
Definition aom_encoder.h:233