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 "third_party/libyuv/include/libyuv/scale.h"
51 #endif
52
53 /* Swallow warnings about unused results of fread/fwrite */
wrap_fread(void * ptr,size_t size,size_t nmemb,FILE * stream)54 static 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
wrap_fwrite(const void * ptr,size_t size,size_t nmemb,FILE * stream)59 static 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
65 static const char *exec_name;
66
warn_or_exit_on_errorv(aom_codec_ctx_t * ctx,int fatal,const char * s,va_list ap)67 static void warn_or_exit_on_errorv(aom_codec_ctx_t *ctx, int fatal,
68 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) exit(EXIT_FAILURE);
78 }
79 }
80
ctx_exit_on_error(aom_codec_ctx_t * ctx,const char * s,...)81 static void ctx_exit_on_error(aom_codec_ctx_t *ctx, const char *s, ...) {
82 va_list ap;
83
84 va_start(ap, s);
85 warn_or_exit_on_errorv(ctx, 1, s, ap);
86 va_end(ap);
87 }
88
warn_or_exit_on_error(aom_codec_ctx_t * ctx,int fatal,const char * s,...)89 static void warn_or_exit_on_error(aom_codec_ctx_t *ctx, int fatal,
90 const char *s, ...) {
91 va_list ap;
92
93 va_start(ap, s);
94 warn_or_exit_on_errorv(ctx, fatal, s, ap);
95 va_end(ap);
96 }
97
read_frame(struct AvxInputContext * input_ctx,aom_image_t * img)98 static int read_frame(struct AvxInputContext *input_ctx, aom_image_t *img) {
99 FILE *f = input_ctx->file;
100 y4m_input *y4m = &input_ctx->y4m;
101 int shortread = 0;
102
103 if (input_ctx->file_type == FILE_TYPE_Y4M) {
104 if (y4m_input_fetch_frame(y4m, f, img) < 1) return 0;
105 } else {
106 shortread = read_yuv_frame(input_ctx, img);
107 }
108
109 return !shortread;
110 }
111
file_is_y4m(const char detect[4])112 static int file_is_y4m(const char detect[4]) {
113 if (memcmp(detect, "YUV4", 4) == 0) {
114 return 1;
115 }
116 return 0;
117 }
118
fourcc_is_ivf(const char detect[4])119 static int fourcc_is_ivf(const char detect[4]) {
120 if (memcmp(detect, "DKIF", 4) == 0) {
121 return 1;
122 }
123 return 0;
124 }
125
126 static const arg_def_t help =
127 ARG_DEF(NULL, "help", 0, "Show usage options and exit");
128 static const arg_def_t debugmode =
129 ARG_DEF("D", "debug", 0, "Debug mode (makes output deterministic)");
130 static const arg_def_t outputfile =
131 ARG_DEF("o", "output", 1, "Output filename");
132 static const arg_def_t use_yv12 =
133 ARG_DEF(NULL, "yv12", 0, "Input file is YV12 ");
134 static const arg_def_t use_i420 =
135 ARG_DEF(NULL, "i420", 0, "Input file is I420 (default)");
136 static const arg_def_t use_i422 =
137 ARG_DEF(NULL, "i422", 0, "Input file is I422");
138 static const arg_def_t use_i444 =
139 ARG_DEF(NULL, "i444", 0, "Input file is I444");
140 static const arg_def_t codecarg = ARG_DEF(NULL, "codec", 1, "Codec to use");
141 static const arg_def_t passes =
142 ARG_DEF("p", "passes", 1, "Number of passes (1/2)");
143 static const arg_def_t pass_arg =
144 ARG_DEF(NULL, "pass", 1, "Pass to execute (1/2)");
145 static const arg_def_t fpf_name =
146 ARG_DEF(NULL, "fpf", 1, "First pass statistics file name");
147 static const arg_def_t limit =
148 ARG_DEF(NULL, "limit", 1, "Stop encoding after n input frames");
149 static const arg_def_t skip =
150 ARG_DEF(NULL, "skip", 1, "Skip the first n input frames");
151 static const arg_def_t good_dl =
152 ARG_DEF(NULL, "good", 0, "Use Good Quality Deadline");
153 static const arg_def_t rt_dl =
154 ARG_DEF(NULL, "rt", 0, "Use Realtime Quality Deadline");
155 static const arg_def_t quietarg =
156 ARG_DEF("q", "quiet", 0, "Do not print encode progress");
157 static const arg_def_t verbosearg =
158 ARG_DEF("v", "verbose", 0, "Show encoder parameters");
159 static const arg_def_t psnrarg =
160 ARG_DEF(NULL, "psnr", 0, "Show PSNR in status line");
161 static const arg_def_t use_cfg = ARG_DEF("c", "cfg", 1, "Config file to use");
162
163 static const struct arg_enum_list test_decode_enum[] = {
164 { "off", TEST_DECODE_OFF },
165 { "fatal", TEST_DECODE_FATAL },
166 { "warn", TEST_DECODE_WARN },
167 { NULL, 0 }
168 };
169 static const arg_def_t recontest = ARG_DEF_ENUM(
170 NULL, "test-decode", 1, "Test encode/decode mismatch", test_decode_enum);
171 static const arg_def_t framerate =
172 ARG_DEF(NULL, "fps", 1, "Stream frame rate (rate/scale)");
173 static const arg_def_t use_webm =
174 ARG_DEF(NULL, "webm", 0, "Output WebM (default when WebM IO is enabled)");
175 static const arg_def_t use_ivf = ARG_DEF(NULL, "ivf", 0, "Output IVF");
176 static const arg_def_t use_obu = ARG_DEF(NULL, "obu", 0, "Output OBU");
177 static const arg_def_t q_hist_n =
178 ARG_DEF(NULL, "q-hist", 1, "Show quantizer histogram (n-buckets)");
179 static const arg_def_t rate_hist_n =
180 ARG_DEF(NULL, "rate-hist", 1, "Show rate histogram (n-buckets)");
181 static const arg_def_t disable_warnings =
182 ARG_DEF(NULL, "disable-warnings", 0,
183 "Disable warnings about potentially incorrect encode settings.");
184 static const arg_def_t disable_warning_prompt =
185 ARG_DEF("y", "disable-warning-prompt", 0,
186 "Display warnings, but do not prompt user to continue.");
187 static const struct arg_enum_list bitdepth_enum[] = {
188 { "8", AOM_BITS_8 }, { "10", AOM_BITS_10 }, { "12", AOM_BITS_12 }, { NULL, 0 }
189 };
190
191 static const arg_def_t bitdeptharg = ARG_DEF_ENUM(
192 "b", "bit-depth", 1,
193 "Bit depth for codec (8 for version <=1, 10 or 12 for version 2)",
194 bitdepth_enum);
195 static const arg_def_t inbitdeptharg =
196 ARG_DEF(NULL, "input-bit-depth", 1, "Bit depth of input");
197
198 static const arg_def_t input_chroma_subsampling_x = ARG_DEF(
199 NULL, "input-chroma-subsampling-x", 1, "chroma subsampling x value.");
200 static const arg_def_t input_chroma_subsampling_y = ARG_DEF(
201 NULL, "input-chroma-subsampling-y", 1, "chroma subsampling y value.");
202
203 static const arg_def_t *main_args[] = { &help,
204 &use_cfg,
205 &debugmode,
206 &outputfile,
207 &codecarg,
208 &passes,
209 &pass_arg,
210 &fpf_name,
211 &limit,
212 &skip,
213 &good_dl,
214 &rt_dl,
215 &quietarg,
216 &verbosearg,
217 &psnrarg,
218 &use_webm,
219 &use_ivf,
220 &use_obu,
221 &q_hist_n,
222 &rate_hist_n,
223 &disable_warnings,
224 &disable_warning_prompt,
225 &recontest,
226 NULL };
227
228 static const arg_def_t usage =
229 ARG_DEF("u", "usage", 1, "Usage profile number to use");
230 static const arg_def_t threads =
231 ARG_DEF("t", "threads", 1, "Max number of threads to use");
232 static const arg_def_t profile =
233 ARG_DEF(NULL, "profile", 1, "Bitstream profile number to use");
234 static const arg_def_t width = ARG_DEF("w", "width", 1, "Frame width");
235 static const arg_def_t height = ARG_DEF("h", "height", 1, "Frame height");
236 static const arg_def_t forced_max_frame_width = ARG_DEF(
237 NULL, "forced_max_frame_width", 1, "Maximum frame width value to force");
238 static const arg_def_t forced_max_frame_height = ARG_DEF(
239 NULL, "forced_max_frame_height", 1, "Maximum frame height value to force");
240 #if CONFIG_WEBM_IO
241 static const struct arg_enum_list stereo_mode_enum[] = {
242 { "mono", STEREO_FORMAT_MONO },
243 { "left-right", STEREO_FORMAT_LEFT_RIGHT },
244 { "bottom-top", STEREO_FORMAT_BOTTOM_TOP },
245 { "top-bottom", STEREO_FORMAT_TOP_BOTTOM },
246 { "right-left", STEREO_FORMAT_RIGHT_LEFT },
247 { NULL, 0 }
248 };
249 static const arg_def_t stereo_mode = ARG_DEF_ENUM(
250 NULL, "stereo-mode", 1, "Stereo 3D video format", stereo_mode_enum);
251 #endif
252 static const arg_def_t timebase = ARG_DEF(
253 NULL, "timebase", 1, "Output timestamp precision (fractional seconds)");
254 static const arg_def_t global_error_resilient =
255 ARG_DEF(NULL, "global-error-resilient", 1,
256 "Enable global error resiliency features");
257 static const arg_def_t lag_in_frames =
258 ARG_DEF(NULL, "lag-in-frames", 1, "Max number of frames to lag");
259 static const arg_def_t large_scale_tile = ARG_DEF(
260 NULL, "large-scale-tile", 1,
261 "Large scale tile coding (0: off (default), 1: on (ivf output only))");
262 static const arg_def_t monochrome =
263 ARG_DEF(NULL, "monochrome", 0, "Monochrome video (no chroma planes)");
264 static const arg_def_t full_still_picture_hdr = ARG_DEF(
265 NULL, "full-still-picture-hdr", 0, "Use full header for still picture");
266
267 static const arg_def_t *global_args[] = { &use_yv12,
268 &use_i420,
269 &use_i422,
270 &use_i444,
271 &usage,
272 &threads,
273 &profile,
274 &width,
275 &height,
276 &forced_max_frame_width,
277 &forced_max_frame_height,
278 #if CONFIG_WEBM_IO
279 &stereo_mode,
280 #endif
281 &timebase,
282 &framerate,
283 &global_error_resilient,
284 &bitdeptharg,
285 &lag_in_frames,
286 &large_scale_tile,
287 &monochrome,
288 &full_still_picture_hdr,
289 NULL };
290
291 static const arg_def_t dropframe_thresh =
292 ARG_DEF(NULL, "drop-frame", 1, "Temporal resampling threshold (buf %)");
293 static const arg_def_t resize_mode =
294 ARG_DEF(NULL, "resize-mode", 1, "Frame resize mode");
295 static const arg_def_t resize_denominator =
296 ARG_DEF(NULL, "resize-denominator", 1, "Frame resize denominator");
297 static const arg_def_t resize_kf_denominator = ARG_DEF(
298 NULL, "resize-kf-denominator", 1, "Frame resize keyframe denominator");
299 static const arg_def_t superres_mode =
300 ARG_DEF(NULL, "superres-mode", 1, "Frame super-resolution mode");
301 static const arg_def_t superres_denominator = ARG_DEF(
302 NULL, "superres-denominator", 1, "Frame super-resolution denominator");
303 static const arg_def_t superres_kf_denominator =
304 ARG_DEF(NULL, "superres-kf-denominator", 1,
305 "Frame super-resolution keyframe denominator");
306 static const arg_def_t superres_qthresh = ARG_DEF(
307 NULL, "superres-qthresh", 1, "Frame super-resolution qindex threshold");
308 static const arg_def_t superres_kf_qthresh =
309 ARG_DEF(NULL, "superres-kf-qthresh", 1,
310 "Frame super-resolution keyframe qindex threshold");
311 static const struct arg_enum_list end_usage_enum[] = { { "vbr", AOM_VBR },
312 { "cbr", AOM_CBR },
313 { "cq", AOM_CQ },
314 { "q", AOM_Q },
315 { NULL, 0 } };
316 static const arg_def_t end_usage =
317 ARG_DEF_ENUM(NULL, "end-usage", 1, "Rate control mode", end_usage_enum);
318 static const arg_def_t target_bitrate =
319 ARG_DEF(NULL, "target-bitrate", 1, "Bitrate (kbps)");
320 static const arg_def_t min_quantizer =
321 ARG_DEF(NULL, "min-q", 1, "Minimum (best) quantizer");
322 static const arg_def_t max_quantizer =
323 ARG_DEF(NULL, "max-q", 1, "Maximum (worst) quantizer");
324 static const arg_def_t undershoot_pct =
325 ARG_DEF(NULL, "undershoot-pct", 1, "Datarate undershoot (min) target (%)");
326 static const arg_def_t overshoot_pct =
327 ARG_DEF(NULL, "overshoot-pct", 1, "Datarate overshoot (max) target (%)");
328 static const arg_def_t buf_sz =
329 ARG_DEF(NULL, "buf-sz", 1, "Client buffer size (ms)");
330 static const arg_def_t buf_initial_sz =
331 ARG_DEF(NULL, "buf-initial-sz", 1, "Client initial buffer size (ms)");
332 static const arg_def_t buf_optimal_sz =
333 ARG_DEF(NULL, "buf-optimal-sz", 1, "Client optimal buffer size (ms)");
334 static const arg_def_t *rc_args[] = { &dropframe_thresh,
335 &resize_mode,
336 &resize_denominator,
337 &resize_kf_denominator,
338 &superres_mode,
339 &superres_denominator,
340 &superres_kf_denominator,
341 &superres_qthresh,
342 &superres_kf_qthresh,
343 &end_usage,
344 &target_bitrate,
345 &min_quantizer,
346 &max_quantizer,
347 &undershoot_pct,
348 &overshoot_pct,
349 &buf_sz,
350 &buf_initial_sz,
351 &buf_optimal_sz,
352 NULL };
353
354 static const arg_def_t bias_pct =
355 ARG_DEF(NULL, "bias-pct", 1, "CBR/VBR bias (0=CBR, 100=VBR)");
356 static const arg_def_t minsection_pct =
357 ARG_DEF(NULL, "minsection-pct", 1, "GOP min bitrate (% of target)");
358 static const arg_def_t maxsection_pct =
359 ARG_DEF(NULL, "maxsection-pct", 1, "GOP max bitrate (% of target)");
360 static const arg_def_t *rc_twopass_args[] = { &bias_pct, &minsection_pct,
361 &maxsection_pct, NULL };
362 static const arg_def_t fwd_kf_enabled =
363 ARG_DEF(NULL, "enable-fwd-kf", 1, "Enable forward reference keyframes");
364 static const arg_def_t kf_min_dist =
365 ARG_DEF(NULL, "kf-min-dist", 1, "Minimum keyframe interval (frames)");
366 static const arg_def_t kf_max_dist =
367 ARG_DEF(NULL, "kf-max-dist", 1, "Maximum keyframe interval (frames)");
368 static const arg_def_t kf_disabled =
369 ARG_DEF(NULL, "disable-kf", 0, "Disable keyframe placement");
370 static const arg_def_t *kf_args[] = { &fwd_kf_enabled, &kf_min_dist,
371 &kf_max_dist, &kf_disabled, NULL };
372 static const arg_def_t sframe_dist =
373 ARG_DEF(NULL, "sframe-dist", 1, "S-Frame interval (frames)");
374 static const arg_def_t sframe_mode =
375 ARG_DEF(NULL, "sframe-mode", 1, "S-Frame insertion mode (1..2)");
376 static const arg_def_t save_as_annexb =
377 ARG_DEF(NULL, "annexb", 1, "Save as Annex-B");
378 static const arg_def_t noise_sens =
379 ARG_DEF(NULL, "noise-sensitivity", 1, "Noise sensitivity (frames to blur)");
380 static const arg_def_t sharpness =
381 ARG_DEF(NULL, "sharpness", 1, "Loop filter sharpness (0..7)");
382 static const arg_def_t static_thresh =
383 ARG_DEF(NULL, "static-thresh", 1, "Motion detection threshold");
384 static const arg_def_t auto_altref =
385 ARG_DEF(NULL, "auto-alt-ref", 1, "Enable automatic alt reference frames");
386 static const arg_def_t arnr_maxframes =
387 ARG_DEF(NULL, "arnr-maxframes", 1, "AltRef max frames (0..15)");
388 static const arg_def_t arnr_strength =
389 ARG_DEF(NULL, "arnr-strength", 1, "AltRef filter strength (0..6)");
390 static const struct arg_enum_list tuning_enum[] = {
391 { "psnr", AOM_TUNE_PSNR },
392 { "ssim", AOM_TUNE_SSIM },
393 { "vmaf_with_preprocessing", AOM_TUNE_VMAF_WITH_PREPROCESSING },
394 { "vmaf_without_preprocessing", AOM_TUNE_VMAF_WITHOUT_PREPROCESSING },
395 { "vmaf", AOM_TUNE_VMAF_MAX_GAIN },
396 { NULL, 0 }
397 };
398 static const arg_def_t tune_metric =
399 ARG_DEF_ENUM(NULL, "tune", 1, "Distortion metric tuned with", tuning_enum);
400 static const arg_def_t cq_level =
401 ARG_DEF(NULL, "cq-level", 1, "Constant/Constrained Quality level");
402 static const arg_def_t max_intra_rate_pct =
403 ARG_DEF(NULL, "max-intra-rate", 1, "Max I-frame bitrate (pct)");
404
405 #if CONFIG_AV1_ENCODER
406 static const arg_def_t cpu_used_av1 =
407 ARG_DEF(NULL, "cpu-used", 1,
408 "Speed setting (0..6 in good mode, 6..8 in realtime mode)");
409 static const arg_def_t rowmtarg =
410 ARG_DEF(NULL, "row-mt", 1,
411 "Enable row based multi-threading (0: off, 1: on (default))");
412 static const arg_def_t tile_cols =
413 ARG_DEF(NULL, "tile-columns", 1, "Number of tile columns to use, log2");
414 static const arg_def_t tile_rows =
415 ARG_DEF(NULL, "tile-rows", 1, "Number of tile rows to use, log2");
416 static const arg_def_t enable_tpl_model =
417 ARG_DEF(NULL, "enable-tpl-model", 1,
418 "RDO based on frame temporal dependency "
419 "(0: off, 1: backward source based). "
420 "This is required for deltaq mode.");
421 static const arg_def_t enable_keyframe_filtering =
422 ARG_DEF(NULL, "enable-keyframe-filtering", 1,
423 "Apply temporal filtering on key frame "
424 "(0: false, 1: true (default)");
425 static const arg_def_t tile_width =
426 ARG_DEF(NULL, "tile-width", 1, "Tile widths (comma separated)");
427 static const arg_def_t tile_height =
428 ARG_DEF(NULL, "tile-height", 1, "Tile heights (command separated)");
429 static const arg_def_t lossless =
430 ARG_DEF(NULL, "lossless", 1, "Lossless mode (0: false (default), 1: true)");
431 static const arg_def_t enable_cdef =
432 ARG_DEF(NULL, "enable-cdef", 1,
433 "Enable the constrained directional enhancement filter (0: false, "
434 "1: true (default))");
435 static const arg_def_t enable_restoration = ARG_DEF(
436 NULL, "enable-restoration", 1,
437 "Enable the loop restoration filter (0: false (default in Realtime mode), "
438 "1: true (default in Non-realtime mode))");
439 static const arg_def_t enable_rect_partitions =
440 ARG_DEF(NULL, "enable-rect-partitions", 1,
441 "Enable rectangular partitions "
442 "(0: false, 1: true (default))");
443 static const arg_def_t enable_ab_partitions =
444 ARG_DEF(NULL, "enable-ab-partitions", 1,
445 "Enable ab partitions (0: false, 1: true (default))");
446 static const arg_def_t enable_1to4_partitions =
447 ARG_DEF(NULL, "enable-1to4-partitions", 1,
448 "Enable 1:4 and 4:1 partitions "
449 "(0: false, 1: true (default))");
450 static const arg_def_t min_partition_size =
451 ARG_DEF(NULL, "min-partition-size", 4,
452 "Set min partition size "
453 "(4:4x4, 8:8x8, 16:16x16, 32:32x32, 64:64x64, 128:128x128). "
454 "On frame with 4k+ resolutions or higher speed settings, the min "
455 "partition size will have a minimum of 8.");
456 static const arg_def_t max_partition_size =
457 ARG_DEF(NULL, "max-partition-size", 128,
458 "Set max partition size "
459 "(4:4x4, 8:8x8, 16:16x16, 32:32x32, 64:64x64, 128:128x128)");
460 static const arg_def_t enable_dual_filter =
461 ARG_DEF(NULL, "enable-dual-filter", 1,
462 "Enable dual filter "
463 "(0: false, 1: true (default))");
464 static const arg_def_t enable_chroma_deltaq =
465 ARG_DEF(NULL, "enable-chroma-deltaq", 1,
466 "Enable chroma delta quant "
467 "(0: false (default), 1: true)");
468 static const arg_def_t enable_intra_edge_filter =
469 ARG_DEF(NULL, "enable-intra-edge-filter", 1,
470 "Enable intra edge filtering "
471 "(0: false, 1: true (default))");
472 static const arg_def_t enable_order_hint =
473 ARG_DEF(NULL, "enable-order-hint", 1,
474 "Enable order hint "
475 "(0: false, 1: true (default))");
476 static const arg_def_t enable_tx64 =
477 ARG_DEF(NULL, "enable-tx64", 1,
478 "Enable 64-pt transform (0: false, 1: true (default))");
479 static const arg_def_t enable_flip_idtx =
480 ARG_DEF(NULL, "enable-flip-idtx", 1,
481 "Enable extended transform type (0: false, 1: true (default)) "
482 "including FLIPADST_DCT, DCT_FLIPADST, FLIPADST_FLIPADST, "
483 "ADST_FLIPADST, FLIPADST_ADST, IDTX, V_DCT, H_DCT, V_ADST, "
484 "H_ADST, V_FLIPADST, H_FLIPADST");
485 static const arg_def_t enable_dist_wtd_comp =
486 ARG_DEF(NULL, "enable-dist-wtd-comp", 1,
487 "Enable distance-weighted compound "
488 "(0: false, 1: true (default))");
489 static const arg_def_t enable_masked_comp =
490 ARG_DEF(NULL, "enable-masked-comp", 1,
491 "Enable masked (wedge/diff-wtd) compound "
492 "(0: false, 1: true (default))");
493 static const arg_def_t enable_onesided_comp =
494 ARG_DEF(NULL, "enable-onesided-comp", 1,
495 "Enable one sided compound "
496 "(0: false, 1: true (default))");
497 static const arg_def_t enable_interintra_comp =
498 ARG_DEF(NULL, "enable-interintra-comp", 1,
499 "Enable interintra compound "
500 "(0: false, 1: true (default))");
501 static const arg_def_t enable_smooth_interintra =
502 ARG_DEF(NULL, "enable-smooth-interintra", 1,
503 "Enable smooth interintra mode "
504 "(0: false, 1: true (default))");
505 static const arg_def_t enable_diff_wtd_comp =
506 ARG_DEF(NULL, "enable-diff-wtd-comp", 1,
507 "Enable difference-weighted compound "
508 "(0: false, 1: true (default))");
509 static const arg_def_t enable_interinter_wedge =
510 ARG_DEF(NULL, "enable-interinter-wedge", 1,
511 "Enable interinter wedge compound "
512 "(0: false, 1: true (default))");
513 static const arg_def_t enable_interintra_wedge =
514 ARG_DEF(NULL, "enable-interintra-wedge", 1,
515 "Enable interintra wedge compound "
516 "(0: false, 1: true (default))");
517 static const arg_def_t enable_global_motion =
518 ARG_DEF(NULL, "enable-global-motion", 1,
519 "Enable global motion "
520 "(0: false, 1: true (default))");
521 static const arg_def_t enable_warped_motion =
522 ARG_DEF(NULL, "enable-warped-motion", 1,
523 "Enable local warped motion "
524 "(0: false, 1: true (default))");
525 static const arg_def_t enable_filter_intra =
526 ARG_DEF(NULL, "enable-filter-intra", 1,
527 "Enable filter intra prediction mode "
528 "(0: false, 1: true (default))");
529 static const arg_def_t enable_smooth_intra =
530 ARG_DEF(NULL, "enable-smooth-intra", 1,
531 "Enable smooth intra prediction modes "
532 "(0: false, 1: true (default))");
533 static const arg_def_t enable_paeth_intra =
534 ARG_DEF(NULL, "enable-paeth-intra", 1,
535 "Enable Paeth intra prediction mode (0: false, 1: true (default))");
536 static const arg_def_t enable_cfl_intra =
537 ARG_DEF(NULL, "enable-cfl-intra", 1,
538 "Enable chroma from luma intra prediction mode "
539 "(0: false, 1: true (default))");
540 static const arg_def_t force_video_mode =
541 ARG_DEF(NULL, "force-video-mode", 1,
542 "Force video mode (0: false, 1: true (default))");
543 static const arg_def_t enable_obmc = ARG_DEF(
544 NULL, "enable-obmc", 1, "Enable OBMC (0: false, 1: true (default))");
545 static const arg_def_t enable_overlay =
546 ARG_DEF(NULL, "enable-overlay", 1,
547 "Enable coding overlay frames (0: false, 1: true (default))");
548 static const arg_def_t enable_palette =
549 ARG_DEF(NULL, "enable-palette", 1,
550 "Enable palette prediction mode (0: false, 1: true (default))");
551 static const arg_def_t enable_intrabc =
552 ARG_DEF(NULL, "enable-intrabc", 1,
553 "Enable intra block copy prediction mode "
554 "(0: false, 1: true (default))");
555 static const arg_def_t enable_angle_delta =
556 ARG_DEF(NULL, "enable-angle-delta", 1,
557 "Enable intra angle delta (0: false, 1: true (default))");
558 static const arg_def_t disable_trellis_quant =
559 ARG_DEF(NULL, "disable-trellis-quant", 1,
560 "Disable trellis optimization of quantized coefficients (0: false "
561 "1: true 2: true for rd search 3: true for estimate yrd serch "
562 "(default))");
563 static const arg_def_t enable_qm =
564 ARG_DEF(NULL, "enable-qm", 1,
565 "Enable quantisation matrices (0: false (default), 1: true)");
566 static const arg_def_t qm_min = ARG_DEF(
567 NULL, "qm-min", 1, "Min quant matrix flatness (0..15), default is 8");
568 static const arg_def_t qm_max = ARG_DEF(
569 NULL, "qm-max", 1, "Max quant matrix flatness (0..15), default is 15");
570 static const arg_def_t reduced_tx_type_set = ARG_DEF(
571 NULL, "reduced-tx-type-set", 1, "Use reduced set of transform types");
572 static const arg_def_t use_intra_dct_only =
573 ARG_DEF(NULL, "use-intra-dct-only", 1, "Use DCT only for INTRA modes");
574 static const arg_def_t use_inter_dct_only =
575 ARG_DEF(NULL, "use-inter-dct-only", 1, "Use DCT only for INTER modes");
576 static const arg_def_t use_intra_default_tx_only =
577 ARG_DEF(NULL, "use-intra-default-tx-only", 1,
578 "Use Default-transform only for INTRA modes");
579 static const arg_def_t quant_b_adapt =
580 ARG_DEF(NULL, "quant-b-adapt", 1, "Use adaptive quantize_b");
581 static const arg_def_t coeff_cost_upd_freq =
582 ARG_DEF(NULL, "coeff-cost-upd-freq", 1,
583 "Update freq for coeff costs"
584 "0: SB, 1: SB Row per Tile, 2: Tile");
585 static const arg_def_t mode_cost_upd_freq =
586 ARG_DEF(NULL, "mode-cost-upd-freq", 1,
587 "Update freq for mode costs"
588 "0: SB, 1: SB Row per Tile, 2: Tile");
589 static const arg_def_t mv_cost_upd_freq =
590 ARG_DEF(NULL, "mv-cost-upd-freq", 1,
591 "Update freq for mv costs"
592 "0: SB, 1: SB Row per Tile, 2: Tile, 3: Off");
593 static const arg_def_t num_tg = ARG_DEF(
594 NULL, "num-tile-groups", 1, "Maximum number of tile groups, default is 1");
595 static const arg_def_t mtu_size =
596 ARG_DEF(NULL, "mtu-size", 1,
597 "MTU size for a tile group, default is 0 (no MTU targeting), "
598 "overrides maximum number of tile groups");
599 static const struct arg_enum_list timing_info_enum[] = {
600 { "unspecified", AOM_TIMING_UNSPECIFIED },
601 { "constant", AOM_TIMING_EQUAL },
602 { "model", AOM_TIMING_DEC_MODEL },
603 { NULL, 0 }
604 };
605 static const arg_def_t timing_info =
606 ARG_DEF_ENUM(NULL, "timing-info", 1,
607 "Signal timing info in the bitstream (model unly works for no "
608 "hidden frames, no super-res yet):",
609 timing_info_enum);
610 #if CONFIG_TUNE_VMAF
611 static const arg_def_t vmaf_model_path =
612 ARG_DEF(NULL, "vmaf-model-path", 1, "Path to the VMAF model file");
613 #endif
614 static const arg_def_t film_grain_test =
615 ARG_DEF(NULL, "film-grain-test", 1,
616 "Film grain test vectors (0: none (default), 1: test-1 2: test-2, "
617 "... 16: test-16)");
618 static const arg_def_t film_grain_table =
619 ARG_DEF(NULL, "film-grain-table", 1,
620 "Path to file containing film grain parameters");
621 #if CONFIG_DENOISE
622 static const arg_def_t denoise_noise_level =
623 ARG_DEF(NULL, "denoise-noise-level", 1,
624 "Amount of noise (from 0 = don't denoise, to 50)");
625 static const arg_def_t denoise_block_size =
626 ARG_DEF(NULL, "denoise-block-size", 1, "Denoise block size (default = 32)");
627 #endif
628 static const arg_def_t enable_ref_frame_mvs =
629 ARG_DEF(NULL, "enable-ref-frame-mvs", 1,
630 "Enable temporal mv prediction (default is 1)");
631 static const arg_def_t frame_parallel_decoding =
632 ARG_DEF(NULL, "frame-parallel", 1,
633 "Enable frame parallel decodability features "
634 "(0: false (default), 1: true)");
635 static const arg_def_t error_resilient_mode =
636 ARG_DEF(NULL, "error-resilient", 1,
637 "Enable error resilient features "
638 "(0: false (default), 1: true)");
639 static const arg_def_t aq_mode = ARG_DEF(
640 NULL, "aq-mode", 1,
641 "Adaptive quantization mode (0: off (default), 1: variance 2: complexity, "
642 "3: cyclic refresh)");
643 static const arg_def_t deltaq_mode =
644 ARG_DEF(NULL, "deltaq-mode", 1,
645 "Delta qindex mode (0: off, 1: deltaq objective (default), "
646 "2: deltaq perceptual). "
647 "Currently this requires enable-tpl-model as a prerequisite.");
648 static const arg_def_t deltalf_mode = ARG_DEF(
649 NULL, "delta-lf-mode", 1, "Enable delta-lf-mode (0: off (default), 1: on)");
650 static const arg_def_t frame_periodic_boost =
651 ARG_DEF(NULL, "frame-boost", 1,
652 "Enable frame periodic boost (0: off (default), 1: on)");
653 static const arg_def_t gf_cbr_boost_pct = ARG_DEF(
654 NULL, "gf-cbr-boost", 1, "Boost for Golden Frame in CBR mode (pct)");
655 static const arg_def_t max_inter_rate_pct =
656 ARG_DEF(NULL, "max-inter-rate", 1, "Max P-frame bitrate (pct)");
657 static const arg_def_t min_gf_interval = ARG_DEF(
658 NULL, "min-gf-interval", 1,
659 "min gf/arf frame interval (default 0, indicating in-built behavior)");
660 static const arg_def_t max_gf_interval = ARG_DEF(
661 NULL, "max-gf-interval", 1,
662 "max gf/arf frame interval (default 0, indicating in-built behavior)");
663 static const arg_def_t gf_min_pyr_height =
664 ARG_DEF(NULL, "gf-min-pyr-height", 1,
665 "Min height for GF group pyramid structure (0 (default) to 5)");
666 static const arg_def_t gf_max_pyr_height =
667 ARG_DEF(NULL, "gf-max-pyr-height", 1,
668 "maximum height for GF group pyramid structure (0 to 5 (default))");
669 static const arg_def_t max_reference_frames = ARG_DEF(
670 NULL, "max-reference-frames", 1,
671 "maximum number of reference frames allowed per frame (3 to 7 (default))");
672 static const arg_def_t reduced_reference_set =
673 ARG_DEF(NULL, "reduced-reference-set", 1,
674 "Use reduced set of single and compound references (0: off "
675 "(default), 1: on)");
676 static const arg_def_t target_seq_level_idx =
677 ARG_DEF(NULL, "target-seq-level-idx", 1,
678 "Target sequence level index. "
679 "Possible values are in the form of \"ABxy\"(pad leading zeros if "
680 "less than 4 digits). "
681 "AB: Operating point(OP) index; "
682 "xy: Target level index for the OP. "
683 "E.g. \"0\" means target level index 0 for the 0th OP; "
684 "\"1021\" means target level index 21 for the 10th OP.");
685 static const arg_def_t set_min_cr =
686 ARG_DEF(NULL, "min-cr", 1,
687 "Set minimum compression ratio. Take integer values. Default is 0. "
688 "If non-zero, encoder will try to keep the compression ratio of "
689 "each frame to be higher than the given value divided by 100.");
690
691 static const struct arg_enum_list color_primaries_enum[] = {
692 { "bt709", AOM_CICP_CP_BT_709 },
693 { "unspecified", AOM_CICP_CP_UNSPECIFIED },
694 { "bt601", AOM_CICP_CP_BT_601 },
695 { "bt470m", AOM_CICP_CP_BT_470_M },
696 { "bt470bg", AOM_CICP_CP_BT_470_B_G },
697 { "smpte240", AOM_CICP_CP_SMPTE_240 },
698 { "film", AOM_CICP_CP_GENERIC_FILM },
699 { "bt2020", AOM_CICP_CP_BT_2020 },
700 { "xyz", AOM_CICP_CP_XYZ },
701 { "smpte431", AOM_CICP_CP_SMPTE_431 },
702 { "smpte432", AOM_CICP_CP_SMPTE_432 },
703 { "ebu3213", AOM_CICP_CP_EBU_3213 },
704 { NULL, 0 }
705 };
706
707 static const arg_def_t input_color_primaries = ARG_DEF_ENUM(
708 NULL, "color-primaries", 1,
709 "Color primaries (CICP) of input content:", color_primaries_enum);
710
711 static const struct arg_enum_list transfer_characteristics_enum[] = {
712 { "unspecified", AOM_CICP_CP_UNSPECIFIED },
713 { "bt709", AOM_CICP_TC_BT_709 },
714 { "bt470m", AOM_CICP_TC_BT_470_M },
715 { "bt470bg", AOM_CICP_TC_BT_470_B_G },
716 { "bt601", AOM_CICP_TC_BT_601 },
717 { "smpte240", AOM_CICP_TC_SMPTE_240 },
718 { "lin", AOM_CICP_TC_LINEAR },
719 { "log100", AOM_CICP_TC_LOG_100 },
720 { "log100sq10", AOM_CICP_TC_LOG_100_SQRT10 },
721 { "iec61966", AOM_CICP_TC_IEC_61966 },
722 { "bt1361", AOM_CICP_TC_BT_1361 },
723 { "srgb", AOM_CICP_TC_SRGB },
724 { "bt2020-10bit", AOM_CICP_TC_BT_2020_10_BIT },
725 { "bt2020-12bit", AOM_CICP_TC_BT_2020_12_BIT },
726 { "smpte2084", AOM_CICP_TC_SMPTE_2084 },
727 { "hlg", AOM_CICP_TC_HLG },
728 { "smpte428", AOM_CICP_TC_SMPTE_428 },
729 { NULL, 0 }
730 };
731
732 static const arg_def_t input_transfer_characteristics =
733 ARG_DEF_ENUM(NULL, "transfer-characteristics", 1,
734 "Transfer characteristics (CICP) of input content:",
735 transfer_characteristics_enum);
736
737 static const struct arg_enum_list matrix_coefficients_enum[] = {
738 { "identity", AOM_CICP_MC_IDENTITY },
739 { "bt709", AOM_CICP_MC_BT_709 },
740 { "unspecified", AOM_CICP_MC_UNSPECIFIED },
741 { "fcc73", AOM_CICP_MC_FCC },
742 { "bt470bg", AOM_CICP_MC_BT_470_B_G },
743 { "bt601", AOM_CICP_MC_BT_601 },
744 { "smpte240", AOM_CICP_CP_SMPTE_240 },
745 { "ycgco", AOM_CICP_MC_SMPTE_YCGCO },
746 { "bt2020ncl", AOM_CICP_MC_BT_2020_NCL },
747 { "bt2020cl", AOM_CICP_MC_BT_2020_CL },
748 { "smpte2085", AOM_CICP_MC_SMPTE_2085 },
749 { "chromncl", AOM_CICP_MC_CHROMAT_NCL },
750 { "chromcl", AOM_CICP_MC_CHROMAT_CL },
751 { "ictcp", AOM_CICP_MC_ICTCP },
752 { NULL, 0 }
753 };
754
755 static const arg_def_t input_matrix_coefficients = ARG_DEF_ENUM(
756 NULL, "matrix-coefficients", 1,
757 "Matrix coefficients (CICP) of input content:", matrix_coefficients_enum);
758
759 static const struct arg_enum_list chroma_sample_position_enum[] = {
760 { "unknown", AOM_CSP_UNKNOWN },
761 { "vertical", AOM_CSP_VERTICAL },
762 { "colocated", AOM_CSP_COLOCATED },
763 { NULL, 0 }
764 };
765
766 static const arg_def_t input_chroma_sample_position =
767 ARG_DEF_ENUM(NULL, "chroma-sample-position", 1,
768 "The chroma sample position when chroma 4:2:0 is signaled:",
769 chroma_sample_position_enum);
770
771 static const struct arg_enum_list tune_content_enum[] = {
772 { "default", AOM_CONTENT_DEFAULT },
773 { "screen", AOM_CONTENT_SCREEN },
774 { NULL, 0 }
775 };
776
777 static const arg_def_t tune_content = ARG_DEF_ENUM(
778 NULL, "tune-content", 1, "Tune content type", tune_content_enum);
779
780 static const arg_def_t cdf_update_mode =
781 ARG_DEF(NULL, "cdf-update-mode", 1,
782 "CDF update mode for entropy coding "
783 "(0: no CDF update; 1: update CDF on all frames(default); "
784 "2: selectively update CDF on some frames");
785
786 static const struct arg_enum_list superblock_size_enum[] = {
787 { "dynamic", AOM_SUPERBLOCK_SIZE_DYNAMIC },
788 { "64", AOM_SUPERBLOCK_SIZE_64X64 },
789 { "128", AOM_SUPERBLOCK_SIZE_128X128 },
790 { NULL, 0 }
791 };
792 static const arg_def_t superblock_size = ARG_DEF_ENUM(
793 NULL, "sb-size", 1, "Superblock size to use", superblock_size_enum);
794
795 static const arg_def_t set_tier_mask =
796 ARG_DEF(NULL, "set-tier-mask", 1,
797 "Set bit mask to specify which tier each of the 32 possible "
798 "operating points conforms to. "
799 "Bit value 0(defualt): Main Tier; 1: High Tier.");
800
801 static const arg_def_t use_fixed_qp_offsets =
802 ARG_DEF(NULL, "use-fixed-qp-offsets", 1,
803 "Enable fixed QP offsets for frames at different levels of the "
804 "pyramid. Selected automatically from --cq-level if "
805 "--fixed-qp-offsets is not provided. If this option is not "
806 "specified (default), offsets are adaptively chosen by the "
807 "encoder.");
808
809 static const arg_def_t fixed_qp_offsets =
810 ARG_DEF(NULL, "fixed-qp-offsets", 1,
811 "Set fixed QP offsets for frames at different levels of the "
812 "pyramid. Comma-separated list of 5 offsets for keyframe, ALTREF, "
813 "and 3 levels of internal alt-refs. If this option is not "
814 "specified (default), offsets are adaptively chosen by the "
815 "encoder.");
816
817 static const arg_def_t *av1_args[] = { &cpu_used_av1,
818 &auto_altref,
819 &sharpness,
820 &static_thresh,
821 &rowmtarg,
822 &tile_cols,
823 &tile_rows,
824 &enable_tpl_model,
825 &enable_keyframe_filtering,
826 &arnr_maxframes,
827 &arnr_strength,
828 &tune_metric,
829 &cq_level,
830 &max_intra_rate_pct,
831 &max_inter_rate_pct,
832 &gf_cbr_boost_pct,
833 &lossless,
834 &enable_cdef,
835 &enable_restoration,
836 &enable_rect_partitions,
837 &enable_ab_partitions,
838 &enable_1to4_partitions,
839 &min_partition_size,
840 &max_partition_size,
841 &enable_dual_filter,
842 &enable_chroma_deltaq,
843 &enable_intra_edge_filter,
844 &enable_order_hint,
845 &enable_tx64,
846 &enable_flip_idtx,
847 &enable_dist_wtd_comp,
848 &enable_masked_comp,
849 &enable_onesided_comp,
850 &enable_interintra_comp,
851 &enable_smooth_interintra,
852 &enable_diff_wtd_comp,
853 &enable_interinter_wedge,
854 &enable_interintra_wedge,
855 &enable_global_motion,
856 &enable_warped_motion,
857 &enable_filter_intra,
858 &enable_smooth_intra,
859 &enable_paeth_intra,
860 &enable_cfl_intra,
861 &force_video_mode,
862 &enable_obmc,
863 &enable_overlay,
864 &enable_palette,
865 &enable_intrabc,
866 &enable_angle_delta,
867 &disable_trellis_quant,
868 &enable_qm,
869 &qm_min,
870 &qm_max,
871 &reduced_tx_type_set,
872 &use_intra_dct_only,
873 &use_inter_dct_only,
874 &use_intra_default_tx_only,
875 &quant_b_adapt,
876 &coeff_cost_upd_freq,
877 &mode_cost_upd_freq,
878 &mv_cost_upd_freq,
879 &frame_parallel_decoding,
880 &error_resilient_mode,
881 &aq_mode,
882 &deltaq_mode,
883 &deltalf_mode,
884 &frame_periodic_boost,
885 &noise_sens,
886 &tune_content,
887 &cdf_update_mode,
888 &input_color_primaries,
889 &input_transfer_characteristics,
890 &input_matrix_coefficients,
891 &input_chroma_sample_position,
892 &min_gf_interval,
893 &max_gf_interval,
894 &gf_min_pyr_height,
895 &gf_max_pyr_height,
896 &superblock_size,
897 &num_tg,
898 &mtu_size,
899 &timing_info,
900 &film_grain_test,
901 &film_grain_table,
902 #if CONFIG_DENOISE
903 &denoise_noise_level,
904 &denoise_block_size,
905 #endif // CONFIG_DENOISE
906 &max_reference_frames,
907 &reduced_reference_set,
908 &enable_ref_frame_mvs,
909 &target_seq_level_idx,
910 &set_tier_mask,
911 &set_min_cr,
912 &bitdeptharg,
913 &inbitdeptharg,
914 &input_chroma_subsampling_x,
915 &input_chroma_subsampling_y,
916 &sframe_dist,
917 &sframe_mode,
918 &save_as_annexb,
919 #if CONFIG_TUNE_VMAF
920 &vmaf_model_path,
921 #endif
922 NULL };
923 static const int av1_arg_ctrl_map[] = { AOME_SET_CPUUSED,
924 AOME_SET_ENABLEAUTOALTREF,
925 AOME_SET_SHARPNESS,
926 AOME_SET_STATIC_THRESHOLD,
927 AV1E_SET_ROW_MT,
928 AV1E_SET_TILE_COLUMNS,
929 AV1E_SET_TILE_ROWS,
930 AV1E_SET_ENABLE_TPL_MODEL,
931 AV1E_SET_ENABLE_KEYFRAME_FILTERING,
932 AOME_SET_ARNR_MAXFRAMES,
933 AOME_SET_ARNR_STRENGTH,
934 AOME_SET_TUNING,
935 AOME_SET_CQ_LEVEL,
936 AOME_SET_MAX_INTRA_BITRATE_PCT,
937 AV1E_SET_MAX_INTER_BITRATE_PCT,
938 AV1E_SET_GF_CBR_BOOST_PCT,
939 AV1E_SET_LOSSLESS,
940 AV1E_SET_ENABLE_CDEF,
941 AV1E_SET_ENABLE_RESTORATION,
942 AV1E_SET_ENABLE_RECT_PARTITIONS,
943 AV1E_SET_ENABLE_AB_PARTITIONS,
944 AV1E_SET_ENABLE_1TO4_PARTITIONS,
945 AV1E_SET_MIN_PARTITION_SIZE,
946 AV1E_SET_MAX_PARTITION_SIZE,
947 AV1E_SET_ENABLE_DUAL_FILTER,
948 AV1E_SET_ENABLE_CHROMA_DELTAQ,
949 AV1E_SET_ENABLE_INTRA_EDGE_FILTER,
950 AV1E_SET_ENABLE_ORDER_HINT,
951 AV1E_SET_ENABLE_TX64,
952 AV1E_SET_ENABLE_FLIP_IDTX,
953 AV1E_SET_ENABLE_DIST_WTD_COMP,
954 AV1E_SET_ENABLE_MASKED_COMP,
955 AV1E_SET_ENABLE_ONESIDED_COMP,
956 AV1E_SET_ENABLE_INTERINTRA_COMP,
957 AV1E_SET_ENABLE_SMOOTH_INTERINTRA,
958 AV1E_SET_ENABLE_DIFF_WTD_COMP,
959 AV1E_SET_ENABLE_INTERINTER_WEDGE,
960 AV1E_SET_ENABLE_INTERINTRA_WEDGE,
961 AV1E_SET_ENABLE_GLOBAL_MOTION,
962 AV1E_SET_ENABLE_WARPED_MOTION,
963 AV1E_SET_ENABLE_FILTER_INTRA,
964 AV1E_SET_ENABLE_SMOOTH_INTRA,
965 AV1E_SET_ENABLE_PAETH_INTRA,
966 AV1E_SET_ENABLE_CFL_INTRA,
967 AV1E_SET_FORCE_VIDEO_MODE,
968 AV1E_SET_ENABLE_OBMC,
969 AV1E_SET_ENABLE_OVERLAY,
970 AV1E_SET_ENABLE_PALETTE,
971 AV1E_SET_ENABLE_INTRABC,
972 AV1E_SET_ENABLE_ANGLE_DELTA,
973 AV1E_SET_DISABLE_TRELLIS_QUANT,
974 AV1E_SET_ENABLE_QM,
975 AV1E_SET_QM_MIN,
976 AV1E_SET_QM_MAX,
977 AV1E_SET_REDUCED_TX_TYPE_SET,
978 AV1E_SET_INTRA_DCT_ONLY,
979 AV1E_SET_INTER_DCT_ONLY,
980 AV1E_SET_INTRA_DEFAULT_TX_ONLY,
981 AV1E_SET_QUANT_B_ADAPT,
982 AV1E_SET_COEFF_COST_UPD_FREQ,
983 AV1E_SET_MODE_COST_UPD_FREQ,
984 AV1E_SET_MV_COST_UPD_FREQ,
985 AV1E_SET_FRAME_PARALLEL_DECODING,
986 AV1E_SET_ERROR_RESILIENT_MODE,
987 AV1E_SET_AQ_MODE,
988 AV1E_SET_DELTAQ_MODE,
989 AV1E_SET_DELTALF_MODE,
990 AV1E_SET_FRAME_PERIODIC_BOOST,
991 AV1E_SET_NOISE_SENSITIVITY,
992 AV1E_SET_TUNE_CONTENT,
993 AV1E_SET_CDF_UPDATE_MODE,
994 AV1E_SET_COLOR_PRIMARIES,
995 AV1E_SET_TRANSFER_CHARACTERISTICS,
996 AV1E_SET_MATRIX_COEFFICIENTS,
997 AV1E_SET_CHROMA_SAMPLE_POSITION,
998 AV1E_SET_MIN_GF_INTERVAL,
999 AV1E_SET_MAX_GF_INTERVAL,
1000 AV1E_SET_GF_MIN_PYRAMID_HEIGHT,
1001 AV1E_SET_GF_MAX_PYRAMID_HEIGHT,
1002 AV1E_SET_SUPERBLOCK_SIZE,
1003 AV1E_SET_NUM_TG,
1004 AV1E_SET_MTU,
1005 AV1E_SET_TIMING_INFO_TYPE,
1006 AV1E_SET_FILM_GRAIN_TEST_VECTOR,
1007 AV1E_SET_FILM_GRAIN_TABLE,
1008 #if CONFIG_DENOISE
1009 AV1E_SET_DENOISE_NOISE_LEVEL,
1010 AV1E_SET_DENOISE_BLOCK_SIZE,
1011 #endif // CONFIG_DENOISE
1012 AV1E_SET_MAX_REFERENCE_FRAMES,
1013 AV1E_SET_REDUCED_REFERENCE_SET,
1014 AV1E_SET_ENABLE_REF_FRAME_MVS,
1015 AV1E_SET_TARGET_SEQ_LEVEL_IDX,
1016 AV1E_SET_TIER_MASK,
1017 AV1E_SET_MIN_CR,
1018 #if CONFIG_TUNE_VMAF
1019 AV1E_SET_VMAF_MODEL_PATH,
1020 #endif
1021 0 };
1022 #endif // CONFIG_AV1_ENCODER
1023
1024 static const arg_def_t *no_args[] = { NULL };
1025
show_help(FILE * fout,int shorthelp)1026 static void show_help(FILE *fout, int shorthelp) {
1027 fprintf(fout, "Usage: %s <options> -o dst_filename src_filename \n",
1028 exec_name);
1029
1030 if (shorthelp) {
1031 fprintf(fout, "Use --help to see the full list of options.\n");
1032 return;
1033 }
1034
1035 fprintf(fout, "\nOptions:\n");
1036 arg_show_usage(fout, main_args);
1037 fprintf(fout, "\nEncoder Global Options:\n");
1038 arg_show_usage(fout, global_args);
1039 fprintf(fout, "\nRate Control Options:\n");
1040 arg_show_usage(fout, rc_args);
1041 fprintf(fout, "\nTwopass Rate Control Options:\n");
1042 arg_show_usage(fout, rc_twopass_args);
1043 fprintf(fout, "\nKeyframe Placement Options:\n");
1044 arg_show_usage(fout, kf_args);
1045 #if CONFIG_AV1_ENCODER
1046 fprintf(fout, "\nAV1 Specific Options:\n");
1047 arg_show_usage(fout, av1_args);
1048 #endif
1049 fprintf(fout,
1050 "\nStream timebase (--timebase):\n"
1051 " The desired precision of timestamps in the output, expressed\n"
1052 " in fractional seconds. Default is 1/1000.\n");
1053 fprintf(fout, "\nIncluded encoders:\n\n");
1054
1055 const int num_encoder = get_aom_encoder_count();
1056 for (int i = 0; i < num_encoder; ++i) {
1057 const AvxInterface *const encoder = get_aom_encoder_by_index(i);
1058 const char *defstr = (i == (num_encoder - 1)) ? "(default)" : "";
1059 fprintf(fout, " %-6s - %s %s\n", encoder->name,
1060 aom_codec_iface_name(encoder->codec_interface()), defstr);
1061 }
1062 fprintf(fout, "\n ");
1063 fprintf(fout, "Use --codec to switch to a non-default encoder.\n\n");
1064 }
1065
usage_exit(void)1066 void usage_exit(void) {
1067 show_help(stderr, 1);
1068 exit(EXIT_FAILURE);
1069 }
1070
1071 #if CONFIG_AV1_ENCODER
1072 #define ARG_CTRL_CNT_MAX NELEMENTS(av1_arg_ctrl_map)
1073 #endif
1074
1075 #if !CONFIG_WEBM_IO
1076 typedef int stereo_format_t;
1077 struct WebmOutputContext {
1078 int debug;
1079 };
1080 #endif
1081
1082 /* Per-stream configuration */
1083 struct stream_config {
1084 struct aom_codec_enc_cfg cfg;
1085 const char *out_fn;
1086 const char *stats_fn;
1087 stereo_format_t stereo_fmt;
1088 int arg_ctrls[ARG_CTRL_CNT_MAX][2];
1089 int arg_ctrl_cnt;
1090 int write_webm;
1091 const char *film_grain_filename;
1092 int write_ivf;
1093 // whether to use 16bit internal buffers
1094 int use_16bit_internal;
1095 #if CONFIG_TUNE_VMAF
1096 const char *vmaf_model_path;
1097 #endif
1098 };
1099
1100 struct stream_state {
1101 int index;
1102 struct stream_state *next;
1103 struct stream_config config;
1104 FILE *file;
1105 struct rate_hist *rate_hist;
1106 struct WebmOutputContext webm_ctx;
1107 uint64_t psnr_sse_total;
1108 uint64_t psnr_samples_total;
1109 double psnr_totals[4];
1110 int psnr_count;
1111 int counts[64];
1112 aom_codec_ctx_t encoder;
1113 unsigned int frames_out;
1114 uint64_t cx_time;
1115 size_t nbytes;
1116 stats_io_t stats;
1117 struct aom_image *img;
1118 aom_codec_ctx_t decoder;
1119 int mismatch_seen;
1120 unsigned int chroma_subsampling_x;
1121 unsigned int chroma_subsampling_y;
1122 };
1123
validate_positive_rational(const char * msg,struct aom_rational * rat)1124 static void validate_positive_rational(const char *msg,
1125 struct aom_rational *rat) {
1126 if (rat->den < 0) {
1127 rat->num *= -1;
1128 rat->den *= -1;
1129 }
1130
1131 if (rat->num < 0) die("Error: %s must be positive\n", msg);
1132
1133 if (!rat->den) die("Error: %s has zero denominator\n", msg);
1134 }
1135
init_config(cfg_options_t * config)1136 static void init_config(cfg_options_t *config) {
1137 memset(config, 0, sizeof(cfg_options_t));
1138 config->super_block_size = 0; // Dynamic
1139 config->max_partition_size = 128;
1140 config->min_partition_size = 4;
1141 config->disable_trellis_quant = 3;
1142 }
1143
1144 /* Parses global config arguments into the AvxEncoderConfig. Note that
1145 * argv is modified and overwrites all parsed arguments.
1146 */
parse_global_config(struct AvxEncoderConfig * global,char *** argv)1147 static void parse_global_config(struct AvxEncoderConfig *global, char ***argv) {
1148 char **argi, **argj;
1149 struct arg arg;
1150 const int num_encoder = get_aom_encoder_count();
1151 char **argv_local = (char **)*argv;
1152 if (num_encoder < 1) die("Error: no valid encoder available\n");
1153
1154 /* Initialize default parameters */
1155 memset(global, 0, sizeof(*global));
1156 global->codec = get_aom_encoder_by_index(num_encoder - 1);
1157 global->passes = 0;
1158 global->color_type = I420;
1159 global->csp = AOM_CSP_UNKNOWN;
1160
1161 int cfg_included = 0;
1162 init_config(&global->encoder_config);
1163
1164 for (argi = argj = argv_local; (*argj = *argi); argi += arg.argv_step) {
1165 arg.argv_step = 1;
1166
1167 if (arg_match(&arg, &use_cfg, argi)) {
1168 if (cfg_included) continue;
1169 parse_cfg(arg.val, &global->encoder_config);
1170 cfg_included = 1;
1171 continue;
1172 }
1173 if (arg_match(&arg, &help, argi)) {
1174 show_help(stdout, 0);
1175 exit(EXIT_SUCCESS);
1176 } else if (arg_match(&arg, &codecarg, argi)) {
1177 global->codec = get_aom_encoder_by_name(arg.val);
1178 if (!global->codec)
1179 die("Error: Unrecognized argument (%s) to --codec\n", arg.val);
1180 } else if (arg_match(&arg, &passes, argi)) {
1181 global->passes = arg_parse_uint(&arg);
1182
1183 if (global->passes < 1 || global->passes > 2)
1184 die("Error: Invalid number of passes (%d)\n", global->passes);
1185 } else if (arg_match(&arg, &pass_arg, argi)) {
1186 global->pass = arg_parse_uint(&arg);
1187
1188 if (global->pass < 1 || global->pass > 2)
1189 die("Error: Invalid pass selected (%d)\n", global->pass);
1190 } else if (arg_match(&arg, &input_chroma_sample_position, argi)) {
1191 global->csp = arg_parse_enum(&arg);
1192 /* Flag is used by later code as well, preserve it. */
1193 argj++;
1194 } else if (arg_match(&arg, &usage, argi))
1195 global->usage = arg_parse_uint(&arg);
1196 else if (arg_match(&arg, &good_dl, argi))
1197 global->usage = AOM_USAGE_GOOD_QUALITY; // Good quality usage
1198 else if (arg_match(&arg, &rt_dl, argi))
1199 global->usage = AOM_USAGE_REALTIME; // Real-time usage
1200 else if (arg_match(&arg, &use_yv12, argi))
1201 global->color_type = YV12;
1202 else if (arg_match(&arg, &use_i420, argi))
1203 global->color_type = I420;
1204 else if (arg_match(&arg, &use_i422, argi))
1205 global->color_type = I422;
1206 else if (arg_match(&arg, &use_i444, argi))
1207 global->color_type = I444;
1208 else if (arg_match(&arg, &quietarg, argi))
1209 global->quiet = 1;
1210 else if (arg_match(&arg, &verbosearg, argi))
1211 global->verbose = 1;
1212 else if (arg_match(&arg, &limit, argi))
1213 global->limit = arg_parse_uint(&arg);
1214 else if (arg_match(&arg, &skip, argi))
1215 global->skip_frames = arg_parse_uint(&arg);
1216 else if (arg_match(&arg, &psnrarg, argi))
1217 global->show_psnr = 1;
1218 else if (arg_match(&arg, &recontest, argi))
1219 global->test_decode = arg_parse_enum_or_int(&arg);
1220 else if (arg_match(&arg, &framerate, argi)) {
1221 global->framerate = arg_parse_rational(&arg);
1222 validate_positive_rational(arg.name, &global->framerate);
1223 global->have_framerate = 1;
1224 } else if (arg_match(&arg, &debugmode, argi))
1225 global->debug = 1;
1226 else if (arg_match(&arg, &q_hist_n, argi))
1227 global->show_q_hist_buckets = arg_parse_uint(&arg);
1228 else if (arg_match(&arg, &rate_hist_n, argi))
1229 global->show_rate_hist_buckets = arg_parse_uint(&arg);
1230 else if (arg_match(&arg, &disable_warnings, argi))
1231 global->disable_warnings = 1;
1232 else if (arg_match(&arg, &disable_warning_prompt, argi))
1233 global->disable_warning_prompt = 1;
1234 else
1235 argj++;
1236 }
1237
1238 if (global->pass) {
1239 /* DWIM: Assume the user meant passes=2 if pass=2 is specified */
1240 if (global->pass > global->passes) {
1241 warn("Assuming --pass=%d implies --passes=%d\n", global->pass,
1242 global->pass);
1243 global->passes = global->pass;
1244 }
1245 }
1246 /* Validate global config */
1247 if (global->passes == 0) {
1248 #if CONFIG_AV1_ENCODER
1249 // Make default AV1 passes = 2 until there is a better quality 1-pass
1250 // encoder
1251 if (global->codec != NULL && global->codec->name != NULL)
1252 global->passes = (strcmp(global->codec->name, "av1") == 0 &&
1253 global->usage != AOM_USAGE_REALTIME)
1254 ? 2
1255 : 1;
1256 #else
1257 global->passes = 1;
1258 #endif
1259 }
1260
1261 if (global->usage == AOM_USAGE_REALTIME && global->passes > 1) {
1262 warn("Enforcing one-pass encoding in realtime mode\n");
1263 global->passes = 1;
1264 }
1265 }
1266
open_input_file(struct AvxInputContext * input,aom_chroma_sample_position_t csp)1267 static void open_input_file(struct AvxInputContext *input,
1268 aom_chroma_sample_position_t csp) {
1269 /* Parse certain options from the input file, if possible */
1270 input->file = strcmp(input->filename, "-") ? fopen(input->filename, "rb")
1271 : set_binary_mode(stdin);
1272
1273 if (!input->file) fatal("Failed to open input file");
1274
1275 if (!fseeko(input->file, 0, SEEK_END)) {
1276 /* Input file is seekable. Figure out how long it is, so we can get
1277 * progress info.
1278 */
1279 input->length = ftello(input->file);
1280 rewind(input->file);
1281 }
1282
1283 /* Default to 1:1 pixel aspect ratio. */
1284 input->pixel_aspect_ratio.numerator = 1;
1285 input->pixel_aspect_ratio.denominator = 1;
1286
1287 /* For RAW input sources, these bytes will applied on the first frame
1288 * in read_frame().
1289 */
1290 input->detect.buf_read = fread(input->detect.buf, 1, 4, input->file);
1291 input->detect.position = 0;
1292
1293 if (input->detect.buf_read == 4 && file_is_y4m(input->detect.buf)) {
1294 if (y4m_input_open(&input->y4m, input->file, input->detect.buf, 4, csp,
1295 input->only_i420) >= 0) {
1296 input->file_type = FILE_TYPE_Y4M;
1297 input->width = input->y4m.pic_w;
1298 input->height = input->y4m.pic_h;
1299 input->pixel_aspect_ratio.numerator = input->y4m.par_n;
1300 input->pixel_aspect_ratio.denominator = input->y4m.par_d;
1301 input->framerate.numerator = input->y4m.fps_n;
1302 input->framerate.denominator = input->y4m.fps_d;
1303 input->fmt = input->y4m.aom_fmt;
1304 input->bit_depth = input->y4m.bit_depth;
1305 } else
1306 fatal("Unsupported Y4M stream.");
1307 } else if (input->detect.buf_read == 4 && fourcc_is_ivf(input->detect.buf)) {
1308 fatal("IVF is not supported as input.");
1309 } else {
1310 input->file_type = FILE_TYPE_RAW;
1311 }
1312 }
1313
close_input_file(struct AvxInputContext * input)1314 static void close_input_file(struct AvxInputContext *input) {
1315 fclose(input->file);
1316 if (input->file_type == FILE_TYPE_Y4M) y4m_input_close(&input->y4m);
1317 }
1318
new_stream(struct AvxEncoderConfig * global,struct stream_state * prev)1319 static struct stream_state *new_stream(struct AvxEncoderConfig *global,
1320 struct stream_state *prev) {
1321 struct stream_state *stream;
1322
1323 stream = calloc(1, sizeof(*stream));
1324 if (stream == NULL) {
1325 fatal("Failed to allocate new stream.");
1326 }
1327
1328 if (prev) {
1329 memcpy(stream, prev, sizeof(*stream));
1330 stream->index++;
1331 prev->next = stream;
1332 } else {
1333 aom_codec_err_t res;
1334
1335 /* Populate encoder configuration */
1336 res = aom_codec_enc_config_default(global->codec->codec_interface(),
1337 &stream->config.cfg, global->usage);
1338 if (res) fatal("Failed to get config: %s\n", aom_codec_err_to_string(res));
1339
1340 /* Change the default timebase to a high enough value so that the
1341 * encoder will always create strictly increasing timestamps.
1342 */
1343 stream->config.cfg.g_timebase.den = 1000;
1344
1345 /* Never use the library's default resolution, require it be parsed
1346 * from the file or set on the command line.
1347 */
1348 stream->config.cfg.g_w = 0;
1349 stream->config.cfg.g_h = 0;
1350
1351 /* Initialize remaining stream parameters */
1352 stream->config.write_webm = 1;
1353 stream->config.write_ivf = 0;
1354
1355 #if CONFIG_WEBM_IO
1356 stream->config.stereo_fmt = STEREO_FORMAT_MONO;
1357 stream->webm_ctx.last_pts_ns = -1;
1358 stream->webm_ctx.writer = NULL;
1359 stream->webm_ctx.segment = NULL;
1360 #endif
1361
1362 /* Allows removal of the application version from the EBML tags */
1363 stream->webm_ctx.debug = global->debug;
1364 memcpy(&stream->config.cfg.encoder_cfg, &global->encoder_config,
1365 sizeof(stream->config.cfg.encoder_cfg));
1366 }
1367
1368 /* Output files must be specified for each stream */
1369 stream->config.out_fn = NULL;
1370
1371 stream->next = NULL;
1372 return stream;
1373 }
1374
set_config_arg_ctrls(struct stream_config * config,int key,const struct arg * arg)1375 static void set_config_arg_ctrls(struct stream_config *config, int key,
1376 const struct arg *arg) {
1377 int j;
1378 if (key == AV1E_SET_FILM_GRAIN_TABLE) {
1379 config->film_grain_filename = arg->val;
1380 return;
1381 }
1382
1383 // For target level, the settings should accumulate rather than overwrite,
1384 // so we simply append it.
1385 if (key == AV1E_SET_TARGET_SEQ_LEVEL_IDX) {
1386 j = config->arg_ctrl_cnt;
1387 assert(j < (int)ARG_CTRL_CNT_MAX);
1388 config->arg_ctrls[j][0] = key;
1389 config->arg_ctrls[j][1] = arg_parse_enum_or_int(arg);
1390 ++config->arg_ctrl_cnt;
1391 return;
1392 }
1393
1394 /* Point either to the next free element or the first instance of this
1395 * control.
1396 */
1397 for (j = 0; j < config->arg_ctrl_cnt; j++)
1398 if (config->arg_ctrls[j][0] == key) break;
1399
1400 /* Update/insert */
1401 assert(j < (int)ARG_CTRL_CNT_MAX);
1402 config->arg_ctrls[j][0] = key;
1403 config->arg_ctrls[j][1] = arg_parse_enum_or_int(arg);
1404
1405 if (key == AOME_SET_ENABLEAUTOALTREF && config->arg_ctrls[j][1] > 1) {
1406 warn("auto-alt-ref > 1 is deprecated... setting auto-alt-ref=1\n");
1407 config->arg_ctrls[j][1] = 1;
1408 }
1409 if (j == config->arg_ctrl_cnt) config->arg_ctrl_cnt++;
1410 }
1411
parse_stream_params(struct AvxEncoderConfig * global,struct stream_state * stream,char ** argv)1412 static int parse_stream_params(struct AvxEncoderConfig *global,
1413 struct stream_state *stream, char **argv) {
1414 char **argi, **argj;
1415 struct arg arg;
1416 static const arg_def_t **ctrl_args = no_args;
1417 static const int *ctrl_args_map = NULL;
1418 struct stream_config *config = &stream->config;
1419 int eos_mark_found = 0;
1420 int webm_forced = 0;
1421
1422 // Handle codec specific options
1423 if (0) {
1424 #if CONFIG_AV1_ENCODER
1425 } else if (strcmp(global->codec->name, "av1") == 0) {
1426 // TODO(jingning): Reuse AV1 specific encoder configuration parameters.
1427 // Consider to expand this set for AV1 encoder control.
1428 ctrl_args = av1_args;
1429 ctrl_args_map = av1_arg_ctrl_map;
1430 #endif
1431 }
1432
1433 for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step) {
1434 arg.argv_step = 1;
1435
1436 /* Once we've found an end-of-stream marker (--) we want to continue
1437 * shifting arguments but not consuming them.
1438 */
1439 if (eos_mark_found) {
1440 argj++;
1441 continue;
1442 } else if (!strcmp(*argj, "--")) {
1443 eos_mark_found = 1;
1444 continue;
1445 }
1446
1447 if (arg_match(&arg, &outputfile, argi)) {
1448 config->out_fn = arg.val;
1449 if (!webm_forced) {
1450 const size_t out_fn_len = strlen(config->out_fn);
1451 if (out_fn_len >= 4 &&
1452 !strcmp(config->out_fn + out_fn_len - 4, ".ivf")) {
1453 config->write_webm = 0;
1454 config->write_ivf = 1;
1455 } else if (out_fn_len >= 4 &&
1456 !strcmp(config->out_fn + out_fn_len - 4, ".obu")) {
1457 config->write_webm = 0;
1458 config->write_ivf = 0;
1459 }
1460 }
1461 } else if (arg_match(&arg, &fpf_name, argi)) {
1462 config->stats_fn = arg.val;
1463 } else if (arg_match(&arg, &use_webm, argi)) {
1464 #if CONFIG_WEBM_IO
1465 config->write_webm = 1;
1466 webm_forced = 1;
1467 #else
1468 die("Error: --webm specified but webm is disabled.");
1469 #endif
1470 } else if (arg_match(&arg, &use_ivf, argi)) {
1471 config->write_webm = 0;
1472 config->write_ivf = 1;
1473 } else if (arg_match(&arg, &use_obu, argi)) {
1474 config->write_webm = 0;
1475 config->write_ivf = 0;
1476 } else if (arg_match(&arg, &threads, argi)) {
1477 config->cfg.g_threads = arg_parse_uint(&arg);
1478 } else if (arg_match(&arg, &profile, argi)) {
1479 config->cfg.g_profile = arg_parse_uint(&arg);
1480 } else if (arg_match(&arg, &width, argi)) {
1481 config->cfg.g_w = arg_parse_uint(&arg);
1482 } else if (arg_match(&arg, &height, argi)) {
1483 config->cfg.g_h = arg_parse_uint(&arg);
1484 } else if (arg_match(&arg, &forced_max_frame_width, argi)) {
1485 config->cfg.g_forced_max_frame_width = arg_parse_uint(&arg);
1486 } else if (arg_match(&arg, &forced_max_frame_height, argi)) {
1487 config->cfg.g_forced_max_frame_height = arg_parse_uint(&arg);
1488 } else if (arg_match(&arg, &bitdeptharg, argi)) {
1489 config->cfg.g_bit_depth = arg_parse_enum_or_int(&arg);
1490 } else if (arg_match(&arg, &inbitdeptharg, argi)) {
1491 config->cfg.g_input_bit_depth = arg_parse_uint(&arg);
1492 } else if (arg_match(&arg, &input_chroma_subsampling_x, argi)) {
1493 stream->chroma_subsampling_x = arg_parse_uint(&arg);
1494 } else if (arg_match(&arg, &input_chroma_subsampling_y, argi)) {
1495 stream->chroma_subsampling_y = arg_parse_uint(&arg);
1496 #if CONFIG_WEBM_IO
1497 } else if (arg_match(&arg, &stereo_mode, argi)) {
1498 config->stereo_fmt = arg_parse_enum_or_int(&arg);
1499 #endif
1500 } else if (arg_match(&arg, &timebase, argi)) {
1501 config->cfg.g_timebase = arg_parse_rational(&arg);
1502 validate_positive_rational(arg.name, &config->cfg.g_timebase);
1503 } else if (arg_match(&arg, &global_error_resilient, argi)) {
1504 config->cfg.g_error_resilient = arg_parse_uint(&arg);
1505 } else if (arg_match(&arg, &lag_in_frames, argi)) {
1506 config->cfg.g_lag_in_frames = arg_parse_uint(&arg);
1507 if (global->usage == AOM_USAGE_REALTIME &&
1508 config->cfg.rc_end_usage == AOM_CBR &&
1509 config->cfg.g_lag_in_frames != 0) {
1510 warn("non-zero %s option ignored in realtime CBR mode.\n", arg.name);
1511 config->cfg.g_lag_in_frames = 0;
1512 }
1513 } else if (arg_match(&arg, &large_scale_tile, argi)) {
1514 config->cfg.large_scale_tile = arg_parse_uint(&arg);
1515 if (config->cfg.large_scale_tile) global->codec = get_aom_lst_encoder();
1516 } else if (arg_match(&arg, &monochrome, argi)) {
1517 config->cfg.monochrome = 1;
1518 } else if (arg_match(&arg, &full_still_picture_hdr, argi)) {
1519 config->cfg.full_still_picture_hdr = 1;
1520 } else if (arg_match(&arg, &dropframe_thresh, argi)) {
1521 config->cfg.rc_dropframe_thresh = arg_parse_uint(&arg);
1522 } else if (arg_match(&arg, &resize_mode, argi)) {
1523 config->cfg.rc_resize_mode = arg_parse_uint(&arg);
1524 } else if (arg_match(&arg, &resize_denominator, argi)) {
1525 config->cfg.rc_resize_denominator = arg_parse_uint(&arg);
1526 } else if (arg_match(&arg, &resize_kf_denominator, argi)) {
1527 config->cfg.rc_resize_kf_denominator = arg_parse_uint(&arg);
1528 } else if (arg_match(&arg, &superres_mode, argi)) {
1529 config->cfg.rc_superres_mode = arg_parse_uint(&arg);
1530 } else if (arg_match(&arg, &superres_denominator, argi)) {
1531 config->cfg.rc_superres_denominator = arg_parse_uint(&arg);
1532 } else if (arg_match(&arg, &superres_kf_denominator, argi)) {
1533 config->cfg.rc_superres_kf_denominator = arg_parse_uint(&arg);
1534 } else if (arg_match(&arg, &superres_qthresh, argi)) {
1535 config->cfg.rc_superres_qthresh = arg_parse_uint(&arg);
1536 } else if (arg_match(&arg, &superres_kf_qthresh, argi)) {
1537 config->cfg.rc_superres_kf_qthresh = arg_parse_uint(&arg);
1538 } else if (arg_match(&arg, &end_usage, argi)) {
1539 config->cfg.rc_end_usage = arg_parse_enum_or_int(&arg);
1540 } else if (arg_match(&arg, &target_bitrate, argi)) {
1541 config->cfg.rc_target_bitrate = arg_parse_uint(&arg);
1542 } else if (arg_match(&arg, &min_quantizer, argi)) {
1543 config->cfg.rc_min_quantizer = arg_parse_uint(&arg);
1544 } else if (arg_match(&arg, &max_quantizer, argi)) {
1545 config->cfg.rc_max_quantizer = arg_parse_uint(&arg);
1546 } else if (arg_match(&arg, &undershoot_pct, argi)) {
1547 config->cfg.rc_undershoot_pct = arg_parse_uint(&arg);
1548 } else if (arg_match(&arg, &overshoot_pct, argi)) {
1549 config->cfg.rc_overshoot_pct = arg_parse_uint(&arg);
1550 } else if (arg_match(&arg, &buf_sz, argi)) {
1551 config->cfg.rc_buf_sz = arg_parse_uint(&arg);
1552 } else if (arg_match(&arg, &buf_initial_sz, argi)) {
1553 config->cfg.rc_buf_initial_sz = arg_parse_uint(&arg);
1554 } else if (arg_match(&arg, &buf_optimal_sz, argi)) {
1555 config->cfg.rc_buf_optimal_sz = arg_parse_uint(&arg);
1556 } else if (arg_match(&arg, &bias_pct, argi)) {
1557 config->cfg.rc_2pass_vbr_bias_pct = arg_parse_uint(&arg);
1558 if (global->passes < 2)
1559 warn("option %s ignored in one-pass mode.\n", arg.name);
1560 } else if (arg_match(&arg, &minsection_pct, argi)) {
1561 config->cfg.rc_2pass_vbr_minsection_pct = arg_parse_uint(&arg);
1562
1563 if (global->passes < 2)
1564 warn("option %s ignored in one-pass mode.\n", arg.name);
1565 } else if (arg_match(&arg, &maxsection_pct, argi)) {
1566 config->cfg.rc_2pass_vbr_maxsection_pct = arg_parse_uint(&arg);
1567
1568 if (global->passes < 2)
1569 warn("option %s ignored in one-pass mode.\n", arg.name);
1570 } else if (arg_match(&arg, &fwd_kf_enabled, argi)) {
1571 config->cfg.fwd_kf_enabled = arg_parse_uint(&arg);
1572 } else if (arg_match(&arg, &kf_min_dist, argi)) {
1573 config->cfg.kf_min_dist = arg_parse_uint(&arg);
1574 } else if (arg_match(&arg, &kf_max_dist, argi)) {
1575 config->cfg.kf_max_dist = arg_parse_uint(&arg);
1576 } else if (arg_match(&arg, &kf_disabled, argi)) {
1577 config->cfg.kf_mode = AOM_KF_DISABLED;
1578 } else if (arg_match(&arg, &sframe_dist, argi)) {
1579 config->cfg.sframe_dist = arg_parse_uint(&arg);
1580 } else if (arg_match(&arg, &sframe_mode, argi)) {
1581 config->cfg.sframe_mode = arg_parse_uint(&arg);
1582 } else if (arg_match(&arg, &save_as_annexb, argi)) {
1583 config->cfg.save_as_annexb = arg_parse_uint(&arg);
1584 } else if (arg_match(&arg, &tile_width, argi)) {
1585 config->cfg.tile_width_count =
1586 arg_parse_list(&arg, config->cfg.tile_widths, MAX_TILE_WIDTHS);
1587 } else if (arg_match(&arg, &tile_height, argi)) {
1588 config->cfg.tile_height_count =
1589 arg_parse_list(&arg, config->cfg.tile_heights, MAX_TILE_HEIGHTS);
1590 #if CONFIG_TUNE_VMAF
1591 } else if (arg_match(&arg, &vmaf_model_path, argi)) {
1592 config->vmaf_model_path = arg.val;
1593 #endif
1594 } else if (arg_match(&arg, &use_fixed_qp_offsets, argi)) {
1595 config->cfg.use_fixed_qp_offsets = arg_parse_uint(&arg);
1596 } else if (arg_match(&arg, &fixed_qp_offsets, argi)) {
1597 const int fixed_qp_offset_count = arg_parse_list(
1598 &arg, config->cfg.fixed_qp_offsets, FIXED_QP_OFFSET_COUNT);
1599 if (fixed_qp_offset_count < FIXED_QP_OFFSET_COUNT) {
1600 die("Option --fixed_qp_offsets requires %d comma-separated values, but "
1601 "only %d values were provided.\n",
1602 FIXED_QP_OFFSET_COUNT, fixed_qp_offset_count);
1603 }
1604 config->cfg.use_fixed_qp_offsets = 1;
1605 } else if (global->usage == AOM_USAGE_REALTIME &&
1606 arg_match(&arg, &enable_restoration, argi)) {
1607 if (arg_parse_uint(&arg) == 1) {
1608 warn("non-zero %s option ignored in realtime mode.\n", arg.name);
1609 }
1610 } else {
1611 int i, match = 0;
1612 for (i = 0; ctrl_args[i]; i++) {
1613 if (arg_match(&arg, ctrl_args[i], argi)) {
1614 match = 1;
1615 if (ctrl_args_map) {
1616 set_config_arg_ctrls(config, ctrl_args_map[i], &arg);
1617 }
1618 }
1619 }
1620 if (!match) argj++;
1621 }
1622 }
1623 config->use_16bit_internal =
1624 config->cfg.g_bit_depth > AOM_BITS_8 || FORCE_HIGHBITDEPTH_DECODING;
1625 return eos_mark_found;
1626 }
1627
1628 #define FOREACH_STREAM(iterator, list) \
1629 for (struct stream_state *iterator = list; iterator; \
1630 iterator = iterator->next)
1631
validate_stream_config(const struct stream_state * stream,const struct AvxEncoderConfig * global)1632 static void validate_stream_config(const struct stream_state *stream,
1633 const struct AvxEncoderConfig *global) {
1634 const struct stream_state *streami;
1635 (void)global;
1636
1637 if (!stream->config.cfg.g_w || !stream->config.cfg.g_h)
1638 fatal(
1639 "Stream %d: Specify stream dimensions with --width (-w) "
1640 " and --height (-h)",
1641 stream->index);
1642
1643 /* Even if bit depth is set on the command line flag to be lower,
1644 * it is upgraded to at least match the input bit depth.
1645 */
1646 assert(stream->config.cfg.g_input_bit_depth <=
1647 (unsigned int)stream->config.cfg.g_bit_depth);
1648
1649 for (streami = stream; streami; streami = streami->next) {
1650 /* All streams require output files */
1651 if (!streami->config.out_fn)
1652 fatal("Stream %d: Output file is required (specify with -o)",
1653 streami->index);
1654
1655 /* Check for two streams outputting to the same file */
1656 if (streami != stream) {
1657 const char *a = stream->config.out_fn;
1658 const char *b = streami->config.out_fn;
1659 if (!strcmp(a, b) && strcmp(a, "/dev/null") && strcmp(a, ":nul"))
1660 fatal("Stream %d: duplicate output file (from stream %d)",
1661 streami->index, stream->index);
1662 }
1663
1664 /* Check for two streams sharing a stats file. */
1665 if (streami != stream) {
1666 const char *a = stream->config.stats_fn;
1667 const char *b = streami->config.stats_fn;
1668 if (a && b && !strcmp(a, b))
1669 fatal("Stream %d: duplicate stats file (from stream %d)",
1670 streami->index, stream->index);
1671 }
1672 }
1673 }
1674
set_stream_dimensions(struct stream_state * stream,unsigned int w,unsigned int h)1675 static void set_stream_dimensions(struct stream_state *stream, unsigned int w,
1676 unsigned int h) {
1677 if (!stream->config.cfg.g_w) {
1678 if (!stream->config.cfg.g_h)
1679 stream->config.cfg.g_w = w;
1680 else
1681 stream->config.cfg.g_w = w * stream->config.cfg.g_h / h;
1682 }
1683 if (!stream->config.cfg.g_h) {
1684 stream->config.cfg.g_h = h * stream->config.cfg.g_w / w;
1685 }
1686 }
1687
file_type_to_string(enum VideoFileType t)1688 static const char *file_type_to_string(enum VideoFileType t) {
1689 switch (t) {
1690 case FILE_TYPE_RAW: return "RAW";
1691 case FILE_TYPE_Y4M: return "Y4M";
1692 default: return "Other";
1693 }
1694 }
1695
image_format_to_string(aom_img_fmt_t f)1696 static const char *image_format_to_string(aom_img_fmt_t f) {
1697 switch (f) {
1698 case AOM_IMG_FMT_I420: return "I420";
1699 case AOM_IMG_FMT_I422: return "I422";
1700 case AOM_IMG_FMT_I444: return "I444";
1701 case AOM_IMG_FMT_YV12: return "YV12";
1702 case AOM_IMG_FMT_YV1216: return "YV1216";
1703 case AOM_IMG_FMT_I42016: return "I42016";
1704 case AOM_IMG_FMT_I42216: return "I42216";
1705 case AOM_IMG_FMT_I44416: return "I44416";
1706 default: return "Other";
1707 }
1708 }
1709
show_stream_config(struct stream_state * stream,struct AvxEncoderConfig * global,struct AvxInputContext * input)1710 static void show_stream_config(struct stream_state *stream,
1711 struct AvxEncoderConfig *global,
1712 struct AvxInputContext *input) {
1713 #define SHOW(field) \
1714 fprintf(stderr, " %-28s = %d\n", #field, stream->config.cfg.field)
1715
1716 if (stream->index == 0) {
1717 fprintf(stderr, "Codec: %s\n",
1718 aom_codec_iface_name(global->codec->codec_interface()));
1719 fprintf(stderr, "Source file: %s File Type: %s Format: %s\n",
1720 input->filename, file_type_to_string(input->file_type),
1721 image_format_to_string(input->fmt));
1722 }
1723 if (stream->next || stream->index)
1724 fprintf(stderr, "\nStream Index: %d\n", stream->index);
1725 fprintf(stderr, "Destination file: %s\n", stream->config.out_fn);
1726 fprintf(stderr, "Coding path: %s\n",
1727 stream->config.use_16bit_internal ? "HBD" : "LBD");
1728 fprintf(stderr, "Encoder parameters:\n");
1729
1730 SHOW(g_usage);
1731 SHOW(g_threads);
1732 SHOW(g_profile);
1733 SHOW(g_w);
1734 SHOW(g_h);
1735 SHOW(g_bit_depth);
1736 SHOW(g_input_bit_depth);
1737 SHOW(g_timebase.num);
1738 SHOW(g_timebase.den);
1739 SHOW(g_error_resilient);
1740 SHOW(g_pass);
1741 SHOW(g_lag_in_frames);
1742 SHOW(large_scale_tile);
1743 SHOW(rc_dropframe_thresh);
1744 SHOW(rc_resize_mode);
1745 SHOW(rc_resize_denominator);
1746 SHOW(rc_resize_kf_denominator);
1747 SHOW(rc_superres_mode);
1748 SHOW(rc_superres_denominator);
1749 SHOW(rc_superres_kf_denominator);
1750 SHOW(rc_superres_qthresh);
1751 SHOW(rc_superres_kf_qthresh);
1752 SHOW(rc_end_usage);
1753 SHOW(rc_target_bitrate);
1754 SHOW(rc_min_quantizer);
1755 SHOW(rc_max_quantizer);
1756 SHOW(rc_undershoot_pct);
1757 SHOW(rc_overshoot_pct);
1758 SHOW(rc_buf_sz);
1759 SHOW(rc_buf_initial_sz);
1760 SHOW(rc_buf_optimal_sz);
1761 SHOW(rc_2pass_vbr_bias_pct);
1762 SHOW(rc_2pass_vbr_minsection_pct);
1763 SHOW(rc_2pass_vbr_maxsection_pct);
1764 SHOW(fwd_kf_enabled);
1765 SHOW(kf_mode);
1766 SHOW(kf_min_dist);
1767 SHOW(kf_max_dist);
1768
1769 #define SHOW_PARAMS(field) \
1770 fprintf(stderr, " %-28s = %d\n", #field, \
1771 stream->config.cfg.encoder_cfg.field)
1772 SHOW_PARAMS(super_block_size);
1773 SHOW_PARAMS(max_partition_size);
1774 SHOW_PARAMS(min_partition_size);
1775 SHOW_PARAMS(disable_ab_partition_type);
1776 SHOW_PARAMS(disable_rect_partition_type);
1777 SHOW_PARAMS(disable_1to4_partition_type);
1778 SHOW_PARAMS(disable_flip_idtx);
1779 SHOW_PARAMS(disable_cdef);
1780 SHOW_PARAMS(disable_lr);
1781 SHOW_PARAMS(disable_obmc);
1782 SHOW_PARAMS(disable_warp_motion);
1783 SHOW_PARAMS(disable_global_motion);
1784 SHOW_PARAMS(disable_dist_wtd_comp);
1785 SHOW_PARAMS(disable_diff_wtd_comp);
1786 SHOW_PARAMS(disable_inter_intra_comp);
1787 SHOW_PARAMS(disable_masked_comp);
1788 SHOW_PARAMS(disable_one_sided_comp);
1789 SHOW_PARAMS(disable_palette);
1790 SHOW_PARAMS(disable_intrabc);
1791 SHOW_PARAMS(disable_cfl);
1792 SHOW_PARAMS(disable_smooth_intra);
1793 SHOW_PARAMS(disable_filter_intra);
1794 SHOW_PARAMS(disable_dual_filter);
1795 SHOW_PARAMS(disable_intra_angle_delta);
1796 SHOW_PARAMS(disable_intra_edge_filter);
1797 SHOW_PARAMS(disable_tx_64x64);
1798 SHOW_PARAMS(disable_smooth_inter_intra);
1799 SHOW_PARAMS(disable_inter_inter_wedge);
1800 SHOW_PARAMS(disable_inter_intra_wedge);
1801 SHOW_PARAMS(disable_paeth_intra);
1802 SHOW_PARAMS(disable_trellis_quant);
1803 SHOW_PARAMS(disable_ref_frame_mv);
1804 SHOW_PARAMS(reduced_reference_set);
1805 SHOW_PARAMS(reduced_tx_type_set);
1806 }
1807
open_output_file(struct stream_state * stream,struct AvxEncoderConfig * global,const struct AvxRational * pixel_aspect_ratio)1808 static void open_output_file(struct stream_state *stream,
1809 struct AvxEncoderConfig *global,
1810 const struct AvxRational *pixel_aspect_ratio) {
1811 const char *fn = stream->config.out_fn;
1812 const struct aom_codec_enc_cfg *const cfg = &stream->config.cfg;
1813
1814 if (cfg->g_pass == AOM_RC_FIRST_PASS) return;
1815
1816 stream->file = strcmp(fn, "-") ? fopen(fn, "wb") : set_binary_mode(stdout);
1817
1818 if (!stream->file) fatal("Failed to open output file");
1819
1820 if (stream->config.write_webm && fseek(stream->file, 0, SEEK_CUR))
1821 fatal("WebM output to pipes not supported.");
1822
1823 #if CONFIG_WEBM_IO
1824 if (stream->config.write_webm) {
1825 stream->webm_ctx.stream = stream->file;
1826 if (write_webm_file_header(&stream->webm_ctx, &stream->encoder, cfg,
1827 stream->config.stereo_fmt, global->codec->fourcc,
1828 pixel_aspect_ratio) != 0) {
1829 fatal("WebM writer initialization failed.");
1830 }
1831 }
1832 #else
1833 (void)pixel_aspect_ratio;
1834 #endif
1835
1836 if (!stream->config.write_webm && stream->config.write_ivf) {
1837 ivf_write_file_header(stream->file, cfg, global->codec->fourcc, 0);
1838 }
1839 }
1840
close_output_file(struct stream_state * stream,unsigned int fourcc)1841 static void close_output_file(struct stream_state *stream,
1842 unsigned int fourcc) {
1843 const struct aom_codec_enc_cfg *const cfg = &stream->config.cfg;
1844
1845 if (cfg->g_pass == AOM_RC_FIRST_PASS) return;
1846
1847 #if CONFIG_WEBM_IO
1848 if (stream->config.write_webm) {
1849 if (write_webm_file_footer(&stream->webm_ctx) != 0) {
1850 fatal("WebM writer finalization failed.");
1851 }
1852 }
1853 #endif
1854
1855 if (!stream->config.write_webm && stream->config.write_ivf) {
1856 if (!fseek(stream->file, 0, SEEK_SET))
1857 ivf_write_file_header(stream->file, &stream->config.cfg, fourcc,
1858 stream->frames_out);
1859 }
1860
1861 fclose(stream->file);
1862 }
1863
setup_pass(struct stream_state * stream,struct AvxEncoderConfig * global,int pass)1864 static void setup_pass(struct stream_state *stream,
1865 struct AvxEncoderConfig *global, int pass) {
1866 if (stream->config.stats_fn) {
1867 if (!stats_open_file(&stream->stats, stream->config.stats_fn, pass))
1868 fatal("Failed to open statistics store");
1869 } else {
1870 if (!stats_open_mem(&stream->stats, pass))
1871 fatal("Failed to open statistics store");
1872 }
1873
1874 stream->config.cfg.g_pass = global->passes == 2
1875 ? pass ? AOM_RC_LAST_PASS : AOM_RC_FIRST_PASS
1876 : AOM_RC_ONE_PASS;
1877 if (pass) {
1878 stream->config.cfg.rc_twopass_stats_in = stats_get(&stream->stats);
1879 }
1880
1881 stream->cx_time = 0;
1882 stream->nbytes = 0;
1883 stream->frames_out = 0;
1884 }
1885
initialize_encoder(struct stream_state * stream,struct AvxEncoderConfig * global)1886 static void initialize_encoder(struct stream_state *stream,
1887 struct AvxEncoderConfig *global) {
1888 int i;
1889 int flags = 0;
1890
1891 flags |= global->show_psnr ? AOM_CODEC_USE_PSNR : 0;
1892 flags |= stream->config.use_16bit_internal ? AOM_CODEC_USE_HIGHBITDEPTH : 0;
1893
1894 /* Construct Encoder Context */
1895 aom_codec_enc_init(&stream->encoder, global->codec->codec_interface(),
1896 &stream->config.cfg, flags);
1897 ctx_exit_on_error(&stream->encoder, "Failed to initialize encoder");
1898
1899 for (i = 0; i < stream->config.arg_ctrl_cnt; i++) {
1900 int ctrl = stream->config.arg_ctrls[i][0];
1901 int value = stream->config.arg_ctrls[i][1];
1902 if (aom_codec_control(&stream->encoder, ctrl, value))
1903 fprintf(stderr, "Error: Tried to set control %d = %d\n", ctrl, value);
1904
1905 ctx_exit_on_error(&stream->encoder, "Failed to control codec");
1906 }
1907
1908 #if CONFIG_TUNE_VMAF
1909 if (stream->config.vmaf_model_path) {
1910 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder, AV1E_SET_VMAF_MODEL_PATH,
1911 stream->config.vmaf_model_path);
1912 }
1913 #endif
1914
1915 if (stream->config.film_grain_filename) {
1916 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder, AV1E_SET_FILM_GRAIN_TABLE,
1917 stream->config.film_grain_filename);
1918 }
1919
1920 #if CONFIG_AV1_DECODER
1921 if (global->test_decode != TEST_DECODE_OFF) {
1922 const AvxInterface *decoder = get_aom_decoder_by_name(global->codec->name);
1923 aom_codec_dec_cfg_t cfg = { 0, 0, 0, !FORCE_HIGHBITDEPTH_DECODING };
1924 aom_codec_dec_init(&stream->decoder, decoder->codec_interface(), &cfg, 0);
1925
1926 if (strcmp(global->codec->name, "av1") == 0) {
1927 AOM_CODEC_CONTROL_TYPECHECKED(&stream->decoder, AV1_SET_TILE_MODE,
1928 stream->config.cfg.large_scale_tile);
1929 ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_mode");
1930
1931 AOM_CODEC_CONTROL_TYPECHECKED(&stream->decoder, AV1D_SET_IS_ANNEXB,
1932 stream->config.cfg.save_as_annexb);
1933 ctx_exit_on_error(&stream->decoder, "Failed to set is_annexb");
1934
1935 AOM_CODEC_CONTROL_TYPECHECKED(&stream->decoder, AV1_SET_DECODE_TILE_ROW,
1936 -1);
1937 ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_row");
1938
1939 AOM_CODEC_CONTROL_TYPECHECKED(&stream->decoder, AV1_SET_DECODE_TILE_COL,
1940 -1);
1941 ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_col");
1942 }
1943 }
1944 #endif
1945 }
1946
encode_frame(struct stream_state * stream,struct AvxEncoderConfig * global,struct aom_image * img,unsigned int frames_in)1947 static void encode_frame(struct stream_state *stream,
1948 struct AvxEncoderConfig *global, struct aom_image *img,
1949 unsigned int frames_in) {
1950 aom_codec_pts_t frame_start, next_frame_start;
1951 struct aom_codec_enc_cfg *cfg = &stream->config.cfg;
1952 struct aom_usec_timer timer;
1953
1954 frame_start =
1955 (cfg->g_timebase.den * (int64_t)(frames_in - 1) * global->framerate.den) /
1956 cfg->g_timebase.num / global->framerate.num;
1957 next_frame_start =
1958 (cfg->g_timebase.den * (int64_t)(frames_in)*global->framerate.den) /
1959 cfg->g_timebase.num / global->framerate.num;
1960
1961 /* Scale if necessary */
1962 if (img) {
1963 if ((img->fmt & AOM_IMG_FMT_HIGHBITDEPTH) &&
1964 (img->d_w != cfg->g_w || img->d_h != cfg->g_h)) {
1965 if (img->fmt != AOM_IMG_FMT_I42016) {
1966 fprintf(stderr, "%s can only scale 4:2:0 inputs\n", exec_name);
1967 exit(EXIT_FAILURE);
1968 }
1969 #if CONFIG_LIBYUV
1970 if (!stream->img) {
1971 stream->img =
1972 aom_img_alloc(NULL, AOM_IMG_FMT_I42016, cfg->g_w, cfg->g_h, 16);
1973 }
1974 I420Scale_16(
1975 (uint16_t *)img->planes[AOM_PLANE_Y], img->stride[AOM_PLANE_Y] / 2,
1976 (uint16_t *)img->planes[AOM_PLANE_U], img->stride[AOM_PLANE_U] / 2,
1977 (uint16_t *)img->planes[AOM_PLANE_V], img->stride[AOM_PLANE_V] / 2,
1978 img->d_w, img->d_h, (uint16_t *)stream->img->planes[AOM_PLANE_Y],
1979 stream->img->stride[AOM_PLANE_Y] / 2,
1980 (uint16_t *)stream->img->planes[AOM_PLANE_U],
1981 stream->img->stride[AOM_PLANE_U] / 2,
1982 (uint16_t *)stream->img->planes[AOM_PLANE_V],
1983 stream->img->stride[AOM_PLANE_V] / 2, stream->img->d_w,
1984 stream->img->d_h, kFilterBox);
1985 img = stream->img;
1986 #else
1987 stream->encoder.err = 1;
1988 ctx_exit_on_error(&stream->encoder,
1989 "Stream %d: Failed to encode frame.\n"
1990 "libyuv is required for scaling but is currently "
1991 "disabled.\n"
1992 "Be sure to specify -DCONFIG_LIBYUV=1 when running "
1993 "cmake.\n",
1994 stream->index);
1995 #endif
1996 }
1997 }
1998 if (img && (img->d_w != cfg->g_w || img->d_h != cfg->g_h)) {
1999 if (img->fmt != AOM_IMG_FMT_I420 && img->fmt != AOM_IMG_FMT_YV12) {
2000 fprintf(stderr, "%s can only scale 4:2:0 8bpp inputs\n", exec_name);
2001 exit(EXIT_FAILURE);
2002 }
2003 #if CONFIG_LIBYUV
2004 if (!stream->img)
2005 stream->img =
2006 aom_img_alloc(NULL, AOM_IMG_FMT_I420, cfg->g_w, cfg->g_h, 16);
2007 I420Scale(
2008 img->planes[AOM_PLANE_Y], img->stride[AOM_PLANE_Y],
2009 img->planes[AOM_PLANE_U], img->stride[AOM_PLANE_U],
2010 img->planes[AOM_PLANE_V], img->stride[AOM_PLANE_V], img->d_w, img->d_h,
2011 stream->img->planes[AOM_PLANE_Y], stream->img->stride[AOM_PLANE_Y],
2012 stream->img->planes[AOM_PLANE_U], stream->img->stride[AOM_PLANE_U],
2013 stream->img->planes[AOM_PLANE_V], stream->img->stride[AOM_PLANE_V],
2014 stream->img->d_w, stream->img->d_h, kFilterBox);
2015 img = stream->img;
2016 #else
2017 stream->encoder.err = 1;
2018 ctx_exit_on_error(&stream->encoder,
2019 "Stream %d: Failed to encode frame.\n"
2020 "Scaling disabled in this configuration. \n"
2021 "To enable, configure with --enable-libyuv\n",
2022 stream->index);
2023 #endif
2024 }
2025
2026 aom_usec_timer_start(&timer);
2027 aom_codec_encode(&stream->encoder, img, frame_start,
2028 (uint32_t)(next_frame_start - frame_start), 0);
2029 aom_usec_timer_mark(&timer);
2030 stream->cx_time += aom_usec_timer_elapsed(&timer);
2031 ctx_exit_on_error(&stream->encoder, "Stream %d: Failed to encode frame",
2032 stream->index);
2033 }
2034
update_quantizer_histogram(struct stream_state * stream)2035 static void update_quantizer_histogram(struct stream_state *stream) {
2036 if (stream->config.cfg.g_pass != AOM_RC_FIRST_PASS) {
2037 int q;
2038
2039 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder, AOME_GET_LAST_QUANTIZER_64,
2040 &q);
2041 ctx_exit_on_error(&stream->encoder, "Failed to read quantizer");
2042 stream->counts[q]++;
2043 }
2044 }
2045
get_cx_data(struct stream_state * stream,struct AvxEncoderConfig * global,int * got_data)2046 static void get_cx_data(struct stream_state *stream,
2047 struct AvxEncoderConfig *global, int *got_data) {
2048 const aom_codec_cx_pkt_t *pkt;
2049 const struct aom_codec_enc_cfg *cfg = &stream->config.cfg;
2050 aom_codec_iter_t iter = NULL;
2051
2052 *got_data = 0;
2053 while ((pkt = aom_codec_get_cx_data(&stream->encoder, &iter))) {
2054 static size_t fsize = 0;
2055 static FileOffset ivf_header_pos = 0;
2056
2057 switch (pkt->kind) {
2058 case AOM_CODEC_CX_FRAME_PKT:
2059 ++stream->frames_out;
2060 if (!global->quiet)
2061 fprintf(stderr, " %6luF", (unsigned long)pkt->data.frame.sz);
2062
2063 update_rate_histogram(stream->rate_hist, cfg, pkt);
2064 #if CONFIG_WEBM_IO
2065 if (stream->config.write_webm) {
2066 if (write_webm_block(&stream->webm_ctx, cfg, pkt) != 0) {
2067 fatal("WebM writer failed.");
2068 }
2069 }
2070 #endif
2071 if (!stream->config.write_webm) {
2072 if (stream->config.write_ivf) {
2073 if (pkt->data.frame.partition_id <= 0) {
2074 ivf_header_pos = ftello(stream->file);
2075 fsize = pkt->data.frame.sz;
2076
2077 ivf_write_frame_header(stream->file, pkt->data.frame.pts, fsize);
2078 } else {
2079 fsize += pkt->data.frame.sz;
2080
2081 const FileOffset currpos = ftello(stream->file);
2082 fseeko(stream->file, ivf_header_pos, SEEK_SET);
2083 ivf_write_frame_size(stream->file, fsize);
2084 fseeko(stream->file, currpos, SEEK_SET);
2085 }
2086 }
2087
2088 (void)fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
2089 stream->file);
2090 }
2091 stream->nbytes += pkt->data.raw.sz;
2092
2093 *got_data = 1;
2094 #if CONFIG_AV1_DECODER
2095 if (global->test_decode != TEST_DECODE_OFF && !stream->mismatch_seen) {
2096 aom_codec_decode(&stream->decoder, pkt->data.frame.buf,
2097 pkt->data.frame.sz, NULL);
2098 if (stream->decoder.err) {
2099 warn_or_exit_on_error(&stream->decoder,
2100 global->test_decode == TEST_DECODE_FATAL,
2101 "Failed to decode frame %d in stream %d",
2102 stream->frames_out + 1, stream->index);
2103 stream->mismatch_seen = stream->frames_out + 1;
2104 }
2105 }
2106 #endif
2107 break;
2108 case AOM_CODEC_STATS_PKT:
2109 stream->frames_out++;
2110 stats_write(&stream->stats, pkt->data.twopass_stats.buf,
2111 pkt->data.twopass_stats.sz);
2112 stream->nbytes += pkt->data.raw.sz;
2113 break;
2114 case AOM_CODEC_PSNR_PKT:
2115
2116 if (global->show_psnr) {
2117 int i;
2118
2119 stream->psnr_sse_total += pkt->data.psnr.sse[0];
2120 stream->psnr_samples_total += pkt->data.psnr.samples[0];
2121 for (i = 0; i < 4; i++) {
2122 if (!global->quiet)
2123 fprintf(stderr, "%.3f ", pkt->data.psnr.psnr[i]);
2124 stream->psnr_totals[i] += pkt->data.psnr.psnr[i];
2125 }
2126 stream->psnr_count++;
2127 }
2128
2129 break;
2130 default: break;
2131 }
2132 }
2133 }
2134
show_psnr(struct stream_state * stream,double peak,int64_t bps)2135 static void show_psnr(struct stream_state *stream, double peak, int64_t bps) {
2136 int i;
2137 double ovpsnr;
2138
2139 if (!stream->psnr_count) return;
2140
2141 fprintf(stderr, "Stream %d PSNR (Overall/Avg/Y/U/V)", stream->index);
2142 ovpsnr = sse_to_psnr((double)stream->psnr_samples_total, peak,
2143 (double)stream->psnr_sse_total);
2144 fprintf(stderr, " %.3f", ovpsnr);
2145
2146 for (i = 0; i < 4; i++) {
2147 fprintf(stderr, " %.3f", stream->psnr_totals[i] / stream->psnr_count);
2148 }
2149 if (bps > 0) {
2150 fprintf(stderr, " %7" PRId64 " bps", bps);
2151 }
2152 fprintf(stderr, " %7" PRId64 " ms", stream->cx_time / 1000);
2153 fprintf(stderr, "\n");
2154 }
2155
usec_to_fps(uint64_t usec,unsigned int frames)2156 static float usec_to_fps(uint64_t usec, unsigned int frames) {
2157 return (float)(usec > 0 ? frames * 1000000.0 / (float)usec : 0);
2158 }
2159
test_decode(struct stream_state * stream,enum TestDecodeFatality fatal)2160 static void test_decode(struct stream_state *stream,
2161 enum TestDecodeFatality fatal) {
2162 aom_image_t enc_img, dec_img;
2163
2164 if (stream->mismatch_seen) return;
2165
2166 /* Get the internal reference frame */
2167 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder, AV1_GET_NEW_FRAME_IMAGE,
2168 &enc_img);
2169 AOM_CODEC_CONTROL_TYPECHECKED(&stream->decoder, AV1_GET_NEW_FRAME_IMAGE,
2170 &dec_img);
2171
2172 if ((enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) !=
2173 (dec_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH)) {
2174 if (enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
2175 aom_image_t enc_hbd_img;
2176 aom_img_alloc(&enc_hbd_img, enc_img.fmt - AOM_IMG_FMT_HIGHBITDEPTH,
2177 enc_img.d_w, enc_img.d_h, 16);
2178 aom_img_truncate_16_to_8(&enc_hbd_img, &enc_img);
2179 enc_img = enc_hbd_img;
2180 }
2181 if (dec_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
2182 aom_image_t dec_hbd_img;
2183 aom_img_alloc(&dec_hbd_img, dec_img.fmt - AOM_IMG_FMT_HIGHBITDEPTH,
2184 dec_img.d_w, dec_img.d_h, 16);
2185 aom_img_truncate_16_to_8(&dec_hbd_img, &dec_img);
2186 dec_img = dec_hbd_img;
2187 }
2188 }
2189
2190 ctx_exit_on_error(&stream->encoder, "Failed to get encoder reference frame");
2191 ctx_exit_on_error(&stream->decoder, "Failed to get decoder reference frame");
2192
2193 if (!aom_compare_img(&enc_img, &dec_img)) {
2194 int y[4], u[4], v[4];
2195 if (enc_img.fmt & AOM_IMG_FMT_HIGHBITDEPTH) {
2196 aom_find_mismatch_high(&enc_img, &dec_img, y, u, v);
2197 } else {
2198 aom_find_mismatch(&enc_img, &dec_img, y, u, v);
2199 }
2200 stream->decoder.err = 1;
2201 warn_or_exit_on_error(&stream->decoder, fatal == TEST_DECODE_FATAL,
2202 "Stream %d: Encode/decode mismatch on frame %d at"
2203 " Y[%d, %d] {%d/%d},"
2204 " U[%d, %d] {%d/%d},"
2205 " V[%d, %d] {%d/%d}",
2206 stream->index, stream->frames_out, y[0], y[1], y[2],
2207 y[3], u[0], u[1], u[2], u[3], v[0], v[1], v[2], v[3]);
2208 stream->mismatch_seen = stream->frames_out;
2209 }
2210
2211 aom_img_free(&enc_img);
2212 aom_img_free(&dec_img);
2213 }
2214
print_time(const char * label,int64_t etl)2215 static void print_time(const char *label, int64_t etl) {
2216 int64_t hours;
2217 int64_t mins;
2218 int64_t secs;
2219
2220 if (etl >= 0) {
2221 hours = etl / 3600;
2222 etl -= hours * 3600;
2223 mins = etl / 60;
2224 etl -= mins * 60;
2225 secs = etl;
2226
2227 fprintf(stderr, "[%3s %2" PRId64 ":%02" PRId64 ":%02" PRId64 "] ", label,
2228 hours, mins, secs);
2229 } else {
2230 fprintf(stderr, "[%3s unknown] ", label);
2231 }
2232 }
2233
main(int argc,const char ** argv_)2234 int main(int argc, const char **argv_) {
2235 int pass;
2236 aom_image_t raw;
2237 aom_image_t raw_shift;
2238 int allocated_raw_shift = 0;
2239 int use_16bit_internal = 0;
2240 int input_shift = 0;
2241 int frame_avail, got_data;
2242
2243 struct AvxInputContext input;
2244 struct AvxEncoderConfig global;
2245 struct stream_state *streams = NULL;
2246 char **argv, **argi;
2247 uint64_t cx_time = 0;
2248 int stream_cnt = 0;
2249 int res = 0;
2250 int profile_updated = 0;
2251
2252 memset(&input, 0, sizeof(input));
2253 exec_name = argv_[0];
2254
2255 /* Setup default input stream settings */
2256 input.framerate.numerator = 30;
2257 input.framerate.denominator = 1;
2258 input.only_i420 = 1;
2259 input.bit_depth = 0;
2260
2261 /* First parse the global configuration values, because we want to apply
2262 * other parameters on top of the default configuration provided by the
2263 * codec.
2264 */
2265 argv = argv_dup(argc - 1, argv_ + 1);
2266 parse_global_config(&global, &argv);
2267
2268 if (argc < 2) usage_exit();
2269
2270 switch (global.color_type) {
2271 case I420: input.fmt = AOM_IMG_FMT_I420; break;
2272 case I422: input.fmt = AOM_IMG_FMT_I422; break;
2273 case I444: input.fmt = AOM_IMG_FMT_I444; break;
2274 case YV12: input.fmt = AOM_IMG_FMT_YV12; break;
2275 }
2276
2277 {
2278 /* Now parse each stream's parameters. Using a local scope here
2279 * due to the use of 'stream' as loop variable in FOREACH_STREAM
2280 * loops
2281 */
2282 struct stream_state *stream = NULL;
2283
2284 do {
2285 stream = new_stream(&global, stream);
2286 stream_cnt++;
2287 if (!streams) streams = stream;
2288 } while (parse_stream_params(&global, stream, argv));
2289 }
2290
2291 /* Check for unrecognized options */
2292 for (argi = argv; *argi; argi++)
2293 if (argi[0][0] == '-' && argi[0][1])
2294 die("Error: Unrecognized option %s\n", *argi);
2295
2296 FOREACH_STREAM(stream, streams) {
2297 check_encoder_config(global.disable_warning_prompt, &global,
2298 &stream->config.cfg);
2299
2300 // If large_scale_tile = 1, only support to output to ivf format.
2301 if (stream->config.cfg.large_scale_tile && !stream->config.write_ivf)
2302 die("only support ivf output format while large-scale-tile=1\n");
2303 }
2304
2305 /* Handle non-option arguments */
2306 input.filename = argv[0];
2307
2308 if (!input.filename) {
2309 fprintf(stderr, "No input file specified!\n");
2310 usage_exit();
2311 }
2312
2313 /* Decide if other chroma subsamplings than 4:2:0 are supported */
2314 if (global.codec->fourcc == AV1_FOURCC) input.only_i420 = 0;
2315
2316 for (pass = global.pass ? global.pass - 1 : 0; pass < global.passes; pass++) {
2317 int frames_in = 0, seen_frames = 0;
2318 int64_t estimated_time_left = -1;
2319 int64_t average_rate = -1;
2320 int64_t lagged_count = 0;
2321
2322 open_input_file(&input, global.csp);
2323
2324 /* If the input file doesn't specify its w/h (raw files), try to get
2325 * the data from the first stream's configuration.
2326 */
2327 if (!input.width || !input.height) {
2328 FOREACH_STREAM(stream, streams) {
2329 if (stream->config.cfg.g_w && stream->config.cfg.g_h) {
2330 input.width = stream->config.cfg.g_w;
2331 input.height = stream->config.cfg.g_h;
2332 break;
2333 }
2334 };
2335 }
2336
2337 /* Update stream configurations from the input file's parameters */
2338 if (!input.width || !input.height)
2339 fatal(
2340 "Specify stream dimensions with --width (-w) "
2341 " and --height (-h)");
2342
2343 /* If input file does not specify bit-depth but input-bit-depth parameter
2344 * exists, assume that to be the input bit-depth. However, if the
2345 * input-bit-depth paramter does not exist, assume the input bit-depth
2346 * to be the same as the codec bit-depth.
2347 */
2348 if (!input.bit_depth) {
2349 FOREACH_STREAM(stream, streams) {
2350 if (stream->config.cfg.g_input_bit_depth)
2351 input.bit_depth = stream->config.cfg.g_input_bit_depth;
2352 else
2353 input.bit_depth = stream->config.cfg.g_input_bit_depth =
2354 (int)stream->config.cfg.g_bit_depth;
2355 }
2356 if (input.bit_depth > 8) input.fmt |= AOM_IMG_FMT_HIGHBITDEPTH;
2357 } else {
2358 FOREACH_STREAM(stream, streams) {
2359 stream->config.cfg.g_input_bit_depth = input.bit_depth;
2360 }
2361 }
2362
2363 FOREACH_STREAM(stream, streams) {
2364 if (input.fmt != AOM_IMG_FMT_I420 && input.fmt != AOM_IMG_FMT_I42016) {
2365 /* Automatically upgrade if input is non-4:2:0 but a 4:2:0 profile
2366 was selected. */
2367 switch (stream->config.cfg.g_profile) {
2368 case 0:
2369 if (input.bit_depth < 12 && (input.fmt == AOM_IMG_FMT_I444 ||
2370 input.fmt == AOM_IMG_FMT_I44416)) {
2371 if (!stream->config.cfg.monochrome) {
2372 stream->config.cfg.g_profile = 1;
2373 profile_updated = 1;
2374 }
2375 } else if (input.bit_depth == 12 || input.fmt == AOM_IMG_FMT_I422 ||
2376 input.fmt == AOM_IMG_FMT_I42216) {
2377 stream->config.cfg.g_profile = 2;
2378 profile_updated = 1;
2379 }
2380 break;
2381 case 1:
2382 if (input.bit_depth == 12 || input.fmt == AOM_IMG_FMT_I422 ||
2383 input.fmt == AOM_IMG_FMT_I42216) {
2384 stream->config.cfg.g_profile = 2;
2385 profile_updated = 1;
2386 } else if (input.bit_depth < 12 &&
2387 (input.fmt == AOM_IMG_FMT_I420 ||
2388 input.fmt == AOM_IMG_FMT_I42016)) {
2389 stream->config.cfg.g_profile = 0;
2390 profile_updated = 1;
2391 }
2392 break;
2393 case 2:
2394 if (input.bit_depth < 12 && (input.fmt == AOM_IMG_FMT_I444 ||
2395 input.fmt == AOM_IMG_FMT_I44416)) {
2396 stream->config.cfg.g_profile = 1;
2397 profile_updated = 1;
2398 } else if (input.bit_depth < 12 &&
2399 (input.fmt == AOM_IMG_FMT_I420 ||
2400 input.fmt == AOM_IMG_FMT_I42016)) {
2401 stream->config.cfg.g_profile = 0;
2402 profile_updated = 1;
2403 } else if (input.bit_depth == 12 &&
2404 input.file_type == FILE_TYPE_Y4M) {
2405 // Note that here the input file values for chroma subsampling
2406 // are used instead of those from the command line.
2407 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2408 AV1E_SET_CHROMA_SUBSAMPLING_X,
2409 input.y4m.dst_c_dec_h >> 1);
2410 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2411 AV1E_SET_CHROMA_SUBSAMPLING_Y,
2412 input.y4m.dst_c_dec_v >> 1);
2413 } else if (input.bit_depth == 12 &&
2414 input.file_type == FILE_TYPE_RAW) {
2415 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2416 AV1E_SET_CHROMA_SUBSAMPLING_X,
2417 stream->chroma_subsampling_x);
2418 AOM_CODEC_CONTROL_TYPECHECKED(&stream->encoder,
2419 AV1E_SET_CHROMA_SUBSAMPLING_Y,
2420 stream->chroma_subsampling_y);
2421 }
2422 break;
2423 default: break;
2424 }
2425 }
2426 /* Automatically set the codec bit depth to match the input bit depth.
2427 * Upgrade the profile if required. */
2428 if (stream->config.cfg.g_input_bit_depth >
2429 (unsigned int)stream->config.cfg.g_bit_depth) {
2430 stream->config.cfg.g_bit_depth = stream->config.cfg.g_input_bit_depth;
2431 if (!global.quiet) {
2432 fprintf(stderr,
2433 "Warning: automatically updating bit depth to %d to "
2434 "match input format.\n",
2435 stream->config.cfg.g_input_bit_depth);
2436 }
2437 }
2438 if (stream->config.cfg.g_bit_depth > 10) {
2439 switch (stream->config.cfg.g_profile) {
2440 case 0:
2441 case 1:
2442 stream->config.cfg.g_profile = 2;
2443 profile_updated = 1;
2444 break;
2445 default: break;
2446 }
2447 }
2448 if (stream->config.cfg.g_bit_depth > 8) {
2449 stream->config.use_16bit_internal = 1;
2450 }
2451 if (profile_updated && !global.quiet) {
2452 fprintf(stderr,
2453 "Warning: automatically updating to profile %d to "
2454 "match input format.\n",
2455 stream->config.cfg.g_profile);
2456 }
2457 /* Set limit */
2458 stream->config.cfg.g_limit = global.limit;
2459 }
2460
2461 FOREACH_STREAM(stream, streams) {
2462 set_stream_dimensions(stream, input.width, input.height);
2463 }
2464 FOREACH_STREAM(stream, streams) { validate_stream_config(stream, &global); }
2465
2466 /* Ensure that --passes and --pass are consistent. If --pass is set and
2467 * --passes=2, ensure --fpf was set.
2468 */
2469 if (global.pass && global.passes == 2) {
2470 FOREACH_STREAM(stream, streams) {
2471 if (!stream->config.stats_fn)
2472 die("Stream %d: Must specify --fpf when --pass=%d"
2473 " and --passes=2\n",
2474 stream->index, global.pass);
2475 }
2476 }
2477
2478 #if !CONFIG_WEBM_IO
2479 FOREACH_STREAM(stream, streams) {
2480 if (stream->config.write_webm) {
2481 stream->config.write_webm = 0;
2482 stream->config.write_ivf = 0;
2483 warn("aomenc compiled w/o WebM support. Writing OBU stream.");
2484 }
2485 }
2486 #endif
2487
2488 /* Use the frame rate from the file only if none was specified
2489 * on the command-line.
2490 */
2491 if (!global.have_framerate) {
2492 global.framerate.num = input.framerate.numerator;
2493 global.framerate.den = input.framerate.denominator;
2494 }
2495 FOREACH_STREAM(stream, streams) {
2496 stream->config.cfg.g_timebase.den = global.framerate.num;
2497 stream->config.cfg.g_timebase.num = global.framerate.den;
2498 }
2499 /* Show configuration */
2500 if (global.verbose && pass == 0) {
2501 FOREACH_STREAM(stream, streams) {
2502 show_stream_config(stream, &global, &input);
2503 }
2504 }
2505
2506 if (pass == (global.pass ? global.pass - 1 : 0)) {
2507 if (input.file_type == FILE_TYPE_Y4M)
2508 /*The Y4M reader does its own allocation.
2509 Just initialize this here to avoid problems if we never read any
2510 frames.*/
2511 memset(&raw, 0, sizeof(raw));
2512 else
2513 aom_img_alloc(&raw, input.fmt, input.width, input.height, 32);
2514
2515 FOREACH_STREAM(stream, streams) {
2516 stream->rate_hist =
2517 init_rate_histogram(&stream->config.cfg, &global.framerate);
2518 }
2519 }
2520
2521 FOREACH_STREAM(stream, streams) { setup_pass(stream, &global, pass); }
2522 FOREACH_STREAM(stream, streams) { initialize_encoder(stream, &global); }
2523 FOREACH_STREAM(stream, streams) {
2524 open_output_file(stream, &global, &input.pixel_aspect_ratio);
2525 }
2526
2527 if (strcmp(global.codec->name, "av1") == 0 ||
2528 strcmp(global.codec->name, "av1") == 0) {
2529 // Check to see if at least one stream uses 16 bit internal.
2530 // Currently assume that the bit_depths for all streams using
2531 // highbitdepth are the same.
2532 FOREACH_STREAM(stream, streams) {
2533 if (stream->config.use_16bit_internal) {
2534 use_16bit_internal = 1;
2535 }
2536 input_shift = (int)stream->config.cfg.g_bit_depth -
2537 stream->config.cfg.g_input_bit_depth;
2538 };
2539 }
2540
2541 frame_avail = 1;
2542 got_data = 0;
2543
2544 while (frame_avail || got_data) {
2545 struct aom_usec_timer timer;
2546
2547 if (!global.limit || frames_in < global.limit) {
2548 frame_avail = read_frame(&input, &raw);
2549
2550 if (frame_avail) frames_in++;
2551 seen_frames =
2552 frames_in > global.skip_frames ? frames_in - global.skip_frames : 0;
2553
2554 if (!global.quiet) {
2555 float fps = usec_to_fps(cx_time, seen_frames);
2556 fprintf(stderr, "\rPass %d/%d ", pass + 1, global.passes);
2557
2558 if (stream_cnt == 1)
2559 fprintf(stderr, "frame %4d/%-4d %7" PRId64 "B ", frames_in,
2560 streams->frames_out, (int64_t)streams->nbytes);
2561 else
2562 fprintf(stderr, "frame %4d ", frames_in);
2563
2564 fprintf(stderr, "%7" PRId64 " %s %.2f %s ",
2565 cx_time > 9999999 ? cx_time / 1000 : cx_time,
2566 cx_time > 9999999 ? "ms" : "us", fps >= 1.0 ? fps : fps * 60,
2567 fps >= 1.0 ? "fps" : "fpm");
2568 print_time("ETA", estimated_time_left);
2569 }
2570
2571 } else {
2572 frame_avail = 0;
2573 }
2574
2575 if (frames_in > global.skip_frames) {
2576 aom_image_t *frame_to_encode;
2577 if (input_shift || (use_16bit_internal && input.bit_depth == 8)) {
2578 assert(use_16bit_internal);
2579 // Input bit depth and stream bit depth do not match, so up
2580 // shift frame to stream bit depth
2581 if (!allocated_raw_shift) {
2582 aom_img_alloc(&raw_shift, raw.fmt | AOM_IMG_FMT_HIGHBITDEPTH,
2583 input.width, input.height, 32);
2584 allocated_raw_shift = 1;
2585 }
2586 aom_img_upshift(&raw_shift, &raw, input_shift);
2587 frame_to_encode = &raw_shift;
2588 } else {
2589 frame_to_encode = &raw;
2590 }
2591 aom_usec_timer_start(&timer);
2592 if (use_16bit_internal) {
2593 assert(frame_to_encode->fmt & AOM_IMG_FMT_HIGHBITDEPTH);
2594 FOREACH_STREAM(stream, streams) {
2595 if (stream->config.use_16bit_internal)
2596 encode_frame(stream, &global,
2597 frame_avail ? frame_to_encode : NULL, frames_in);
2598 else
2599 assert(0);
2600 };
2601 } else {
2602 assert((frame_to_encode->fmt & AOM_IMG_FMT_HIGHBITDEPTH) == 0);
2603 FOREACH_STREAM(stream, streams) {
2604 encode_frame(stream, &global, frame_avail ? frame_to_encode : NULL,
2605 frames_in);
2606 }
2607 }
2608 aom_usec_timer_mark(&timer);
2609 cx_time += aom_usec_timer_elapsed(&timer);
2610
2611 FOREACH_STREAM(stream, streams) { update_quantizer_histogram(stream); }
2612
2613 got_data = 0;
2614 FOREACH_STREAM(stream, streams) {
2615 get_cx_data(stream, &global, &got_data);
2616 }
2617
2618 if (!got_data && input.length && streams != NULL &&
2619 !streams->frames_out) {
2620 lagged_count = global.limit ? seen_frames : ftello(input.file);
2621 } else if (input.length) {
2622 int64_t remaining;
2623 int64_t rate;
2624
2625 if (global.limit) {
2626 const int64_t frame_in_lagged = (seen_frames - lagged_count) * 1000;
2627
2628 rate = cx_time ? frame_in_lagged * (int64_t)1000000 / cx_time : 0;
2629 remaining = 1000 * (global.limit - global.skip_frames -
2630 seen_frames + lagged_count);
2631 } else {
2632 const int64_t input_pos = ftello(input.file);
2633 const int64_t input_pos_lagged = input_pos - lagged_count;
2634 const int64_t input_limit = input.length;
2635
2636 rate = cx_time ? input_pos_lagged * (int64_t)1000000 / cx_time : 0;
2637 remaining = input_limit - input_pos + lagged_count;
2638 }
2639
2640 average_rate =
2641 (average_rate <= 0) ? rate : (average_rate * 7 + rate) / 8;
2642 estimated_time_left = average_rate ? remaining / average_rate : -1;
2643 }
2644
2645 if (got_data && global.test_decode != TEST_DECODE_OFF) {
2646 FOREACH_STREAM(stream, streams) {
2647 test_decode(stream, global.test_decode);
2648 }
2649 }
2650 }
2651
2652 fflush(stdout);
2653 if (!global.quiet) fprintf(stderr, "\033[K");
2654 }
2655
2656 if (stream_cnt > 1) fprintf(stderr, "\n");
2657
2658 if (!global.quiet) {
2659 FOREACH_STREAM(stream, streams) {
2660 const int64_t bpf =
2661 seen_frames ? (int64_t)(stream->nbytes * 8 / seen_frames) : 0;
2662 const int64_t bps = bpf * global.framerate.num / global.framerate.den;
2663 fprintf(stderr,
2664 "\rPass %d/%d frame %4d/%-4d %7" PRId64 "B %7" PRId64
2665 "b/f %7" PRId64
2666 "b/s"
2667 " %7" PRId64 " %s (%.2f fps)\033[K\n",
2668 pass + 1, global.passes, frames_in, stream->frames_out,
2669 (int64_t)stream->nbytes, bpf, bps,
2670 stream->cx_time > 9999999 ? stream->cx_time / 1000
2671 : stream->cx_time,
2672 stream->cx_time > 9999999 ? "ms" : "us",
2673 usec_to_fps(stream->cx_time, seen_frames));
2674 }
2675 }
2676
2677 if (global.show_psnr) {
2678 if (global.codec->fourcc == AV1_FOURCC) {
2679 FOREACH_STREAM(stream, streams) {
2680 int64_t bps = 0;
2681 if (stream->psnr_count && seen_frames && global.framerate.den) {
2682 bps = (int64_t)stream->nbytes * 8 * (int64_t)global.framerate.num /
2683 global.framerate.den / seen_frames;
2684 }
2685 show_psnr(stream, (1 << stream->config.cfg.g_input_bit_depth) - 1,
2686 bps);
2687 }
2688 } else {
2689 FOREACH_STREAM(stream, streams) { show_psnr(stream, 255.0, 0); }
2690 }
2691 }
2692
2693 FOREACH_STREAM(stream, streams) { aom_codec_destroy(&stream->encoder); }
2694
2695 if (global.test_decode != TEST_DECODE_OFF) {
2696 FOREACH_STREAM(stream, streams) { aom_codec_destroy(&stream->decoder); }
2697 }
2698
2699 close_input_file(&input);
2700
2701 if (global.test_decode == TEST_DECODE_FATAL) {
2702 FOREACH_STREAM(stream, streams) { res |= stream->mismatch_seen; }
2703 }
2704 FOREACH_STREAM(stream, streams) {
2705 close_output_file(stream, global.codec->fourcc);
2706 }
2707
2708 FOREACH_STREAM(stream, streams) {
2709 stats_close(&stream->stats, global.passes - 1);
2710 }
2711
2712 if (global.pass) break;
2713 }
2714
2715 if (global.show_q_hist_buckets) {
2716 FOREACH_STREAM(stream, streams) {
2717 show_q_histogram(stream->counts, global.show_q_hist_buckets);
2718 }
2719 }
2720
2721 if (global.show_rate_hist_buckets) {
2722 FOREACH_STREAM(stream, streams) {
2723 show_rate_histogram(stream->rate_hist, &stream->config.cfg,
2724 global.show_rate_hist_buckets);
2725 }
2726 }
2727 FOREACH_STREAM(stream, streams) { destroy_rate_histogram(stream->rate_hist); }
2728
2729 #if CONFIG_INTERNAL_STATS
2730 /* TODO(jkoleszar): This doesn't belong in this executable. Do it for now,
2731 * to match some existing utilities.
2732 */
2733 if (!(global.pass == 1 && global.passes == 2)) {
2734 FOREACH_STREAM(stream, streams) {
2735 FILE *f = fopen("opsnr.stt", "a");
2736 if (stream->mismatch_seen) {
2737 fprintf(f, "First mismatch occurred in frame %d\n",
2738 stream->mismatch_seen);
2739 } else {
2740 fprintf(f, "No mismatch detected in recon buffers\n");
2741 }
2742 fclose(f);
2743 }
2744 }
2745 #endif
2746
2747 if (allocated_raw_shift) aom_img_free(&raw_shift);
2748 aom_img_free(&raw);
2749 free(argv);
2750 free(streams);
2751 return res ? EXIT_FAILURE : EXIT_SUCCESS;
2752 }
2753