aom_cx_set_ref.c 14.5 KB
Newer Older
1
/*
Yaowu Xu's avatar
Yaowu Xu committed
2
 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
3
 *
Yaowu Xu's avatar
Yaowu Xu committed
4 5 6 7 8 9
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
10 11
 */

Yaowu Xu's avatar
Yaowu Xu committed
12
// AV1 Set Reference Frame
13 14
// ============================
//
Yaowu Xu's avatar
Yaowu Xu committed
15
// This is an example demonstrating how to overwrite the AV1 encoder's
16 17 18 19 20 21 22 23 24 25 26 27
// internal reference frame. In the sample we set the last frame to the
// current frame. This technique could be used to bounce between two cameras.
//
// The decoder would also have to set the reference frame to the same value
// on the same frame, or the video will become corrupt. The 'test_decode'
// variable is set to 1 in this example that tests if the encoder and decoder
// results are matching.
//
// Usage
// -----
// This example encodes a raw video. And the last argument passed in specifies
// the frame number to update the reference frame on. For example, run
Yaowu Xu's avatar
Yaowu Xu committed
28
// examples/aom_cx_set_ref av1 352 288 in.yuv out.ivf 4 30
29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
// The parameter is parsed as follows:
//
//
// Extra Variables
// ---------------
// This example maintains the frame number passed on the command line
// in the `update_frame_num` variable.
//
//
// Configuration
// -------------
//
// The reference frame is updated on the frame specified on the command
// line.
//
// Observing The Effects
// ---------------------
// The encoder and decoder results should be matching when the same reference
// frame setting operation is done in both encoder and decoder. Otherwise,
// the encoder/decoder mismatch would be seen.

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

Yaowu Xu's avatar
Yaowu Xu committed
54 55 56
#include "aom/aomcx.h"
#include "aom/aom_decoder.h"
#include "aom/aom_encoder.h"
57 58 59 60 61 62 63

#include "./tools_common.h"
#include "./video_writer.h"

static const char *exec_name;

void usage_exit() {
clang-format's avatar
clang-format committed
64 65
  fprintf(stderr,
          "Usage: %s <codec> <width> <height> <infile> <outfile> "
66 67 68 69 70
          "<frame> <limit(optional)>\n",
          exec_name);
  exit(EXIT_FAILURE);
}

Yaowu Xu's avatar
Yaowu Xu committed
71 72
static int compare_img(const aom_image_t *const img1,
                       const aom_image_t *const img2) {
73
  uint32_t l_w = img1->d_w;
clang-format's avatar
clang-format committed
74
  uint32_t c_w = (img1->d_w + img1->x_chroma_shift) >> img1->x_chroma_shift;
75 76 77 78 79 80 81 82 83 84
  const uint32_t c_h =
      (img1->d_h + img1->y_chroma_shift) >> img1->y_chroma_shift;
  uint32_t i;
  int match = 1;

  match &= (img1->fmt == img2->fmt);
  match &= (img1->d_w == img2->d_w);
  match &= (img1->d_h == img2->d_h);

  for (i = 0; i < img1->d_h; ++i)
Yaowu Xu's avatar
Yaowu Xu committed
85 86
    match &= (memcmp(img1->planes[AOM_PLANE_Y] + i * img1->stride[AOM_PLANE_Y],
                     img2->planes[AOM_PLANE_Y] + i * img2->stride[AOM_PLANE_Y],
87 88 89
                     l_w) == 0);

  for (i = 0; i < c_h; ++i)
Yaowu Xu's avatar
Yaowu Xu committed
90 91
    match &= (memcmp(img1->planes[AOM_PLANE_U] + i * img1->stride[AOM_PLANE_U],
                     img2->planes[AOM_PLANE_U] + i * img2->stride[AOM_PLANE_U],
92 93 94
                     c_w) == 0);

  for (i = 0; i < c_h; ++i)
Yaowu Xu's avatar
Yaowu Xu committed
95 96
    match &= (memcmp(img1->planes[AOM_PLANE_V] + i * img1->stride[AOM_PLANE_V],
                     img2->planes[AOM_PLANE_V] + i * img2->stride[AOM_PLANE_V],
97 98 99 100 101
                     c_w) == 0);

  return match;
}

clang-format's avatar
clang-format committed
102
#define mmin(a, b) ((a) < (b) ? (a) : (b))
Yaowu Xu's avatar
Yaowu Xu committed
103 104
static void find_mismatch(const aom_image_t *const img1,
                          const aom_image_t *const img2, int yloc[4],
clang-format's avatar
clang-format committed
105
                          int uloc[4], int vloc[4]) {
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
  const uint32_t bsize = 64;
  const uint32_t bsizey = bsize >> img1->y_chroma_shift;
  const uint32_t bsizex = bsize >> img1->x_chroma_shift;
  const uint32_t c_w =
      (img1->d_w + img1->x_chroma_shift) >> img1->x_chroma_shift;
  const uint32_t c_h =
      (img1->d_h + img1->y_chroma_shift) >> img1->y_chroma_shift;
  int match = 1;
  uint32_t i, j;
  yloc[0] = yloc[1] = yloc[2] = yloc[3] = -1;
  for (i = 0, match = 1; match && i < img1->d_h; i += bsize) {
    for (j = 0; match && j < img1->d_w; j += bsize) {
      int k, l;
      const int si = mmin(i + bsize, img1->d_h) - i;
      const int sj = mmin(j + bsize, img1->d_w) - j;
      for (k = 0; match && k < si; ++k) {
        for (l = 0; match && l < sj; ++l) {
Yaowu Xu's avatar
Yaowu Xu committed
123 124 125 126
          if (*(img1->planes[AOM_PLANE_Y] +
                (i + k) * img1->stride[AOM_PLANE_Y] + j + l) !=
              *(img2->planes[AOM_PLANE_Y] +
                (i + k) * img2->stride[AOM_PLANE_Y] + j + l)) {
127 128
            yloc[0] = i + k;
            yloc[1] = j + l;
Yaowu Xu's avatar
Yaowu Xu committed
129 130 131 132
            yloc[2] = *(img1->planes[AOM_PLANE_Y] +
                        (i + k) * img1->stride[AOM_PLANE_Y] + j + l);
            yloc[3] = *(img2->planes[AOM_PLANE_Y] +
                        (i + k) * img2->stride[AOM_PLANE_Y] + j + l);
133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
            match = 0;
            break;
          }
        }
      }
    }
  }

  uloc[0] = uloc[1] = uloc[2] = uloc[3] = -1;
  for (i = 0, match = 1; match && i < c_h; i += bsizey) {
    for (j = 0; match && j < c_w; j += bsizex) {
      int k, l;
      const int si = mmin(i + bsizey, c_h - i);
      const int sj = mmin(j + bsizex, c_w - j);
      for (k = 0; match && k < si; ++k) {
        for (l = 0; match && l < sj; ++l) {
Yaowu Xu's avatar
Yaowu Xu committed
149 150 151 152
          if (*(img1->planes[AOM_PLANE_U] +
                (i + k) * img1->stride[AOM_PLANE_U] + j + l) !=
              *(img2->planes[AOM_PLANE_U] +
                (i + k) * img2->stride[AOM_PLANE_U] + j + l)) {
153 154
            uloc[0] = i + k;
            uloc[1] = j + l;
Yaowu Xu's avatar
Yaowu Xu committed
155 156 157 158
            uloc[2] = *(img1->planes[AOM_PLANE_U] +
                        (i + k) * img1->stride[AOM_PLANE_U] + j + l);
            uloc[3] = *(img2->planes[AOM_PLANE_U] +
                        (i + k) * img2->stride[AOM_PLANE_U] + j + l);
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173
            match = 0;
            break;
          }
        }
      }
    }
  }
  vloc[0] = vloc[1] = vloc[2] = vloc[3] = -1;
  for (i = 0, match = 1; match && i < c_h; i += bsizey) {
    for (j = 0; match && j < c_w; j += bsizex) {
      int k, l;
      const int si = mmin(i + bsizey, c_h - i);
      const int sj = mmin(j + bsizex, c_w - j);
      for (k = 0; match && k < si; ++k) {
        for (l = 0; match && l < sj; ++l) {
Yaowu Xu's avatar
Yaowu Xu committed
174 175 176 177
          if (*(img1->planes[AOM_PLANE_V] +
                (i + k) * img1->stride[AOM_PLANE_V] + j + l) !=
              *(img2->planes[AOM_PLANE_V] +
                (i + k) * img2->stride[AOM_PLANE_V] + j + l)) {
178 179
            vloc[0] = i + k;
            vloc[1] = j + l;
Yaowu Xu's avatar
Yaowu Xu committed
180 181 182 183
            vloc[2] = *(img1->planes[AOM_PLANE_V] +
                        (i + k) * img1->stride[AOM_PLANE_V] + j + l);
            vloc[3] = *(img2->planes[AOM_PLANE_V] +
                        (i + k) * img2->stride[AOM_PLANE_V] + j + l);
184 185 186 187 188 189 190 191 192
            match = 0;
            break;
          }
        }
      }
    }
  }
}

Yaowu Xu's avatar
Yaowu Xu committed
193
static void testing_decode(aom_codec_ctx_t *encoder, aom_codec_ctx_t *decoder,
194
                           unsigned int frame_out, int *mismatch_seen) {
Yaowu Xu's avatar
Yaowu Xu committed
195 196
  aom_image_t enc_img, dec_img;
  struct av1_ref_frame ref_enc, ref_dec;
197

clang-format's avatar
clang-format committed
198
  if (*mismatch_seen) return;
199 200 201

  ref_enc.idx = 0;
  ref_dec.idx = 0;
Yaowu Xu's avatar
Yaowu Xu committed
202
  if (aom_codec_control(encoder, AV1_GET_REFERENCE, &ref_enc))
clang-format's avatar
clang-format committed
203
    die_codec(encoder, "Failed to get encoder reference frame");
204
  enc_img = ref_enc.img;
Yaowu Xu's avatar
Yaowu Xu committed
205
  if (aom_codec_control(decoder, AV1_GET_REFERENCE, &ref_dec))
206 207 208 209 210 211 212 213 214
    die_codec(decoder, "Failed to get decoder reference frame");
  dec_img = ref_dec.img;

  if (!compare_img(&enc_img, &dec_img)) {
    int y[4], u[4], v[4];

    *mismatch_seen = 1;

    find_mismatch(&enc_img, &dec_img, y, u, v);
clang-format's avatar
clang-format committed
215 216 217 218 219 220 221
    printf(
        "Encode/decode mismatch on frame %d at"
        " Y[%d, %d] {%d/%d},"
        " U[%d, %d] {%d/%d},"
        " V[%d, %d] {%d/%d}",
        frame_out, y[0], y[1], y[2], y[3], u[0], u[1], u[2], u[3], v[0], v[1],
        v[2], v[3]);
222 223
  }

Yaowu Xu's avatar
Yaowu Xu committed
224 225
  aom_img_free(&enc_img);
  aom_img_free(&dec_img);
226 227
}

228 229 230 231
static int encode_frame(aom_codec_ctx_t *ecodec, aom_image_t *img,
                        unsigned int frame_in, AvxVideoWriter *writer,
                        int test_decode, aom_codec_ctx_t *dcodec,
                        unsigned int *frame_out, int *mismatch_seen) {
232
  int got_pkts = 0;
Yaowu Xu's avatar
Yaowu Xu committed
233 234
  aom_codec_iter_t iter = NULL;
  const aom_codec_cx_pkt_t *pkt = NULL;
235
  int got_data;
Yaowu Xu's avatar
Yaowu Xu committed
236 237 238
  const aom_codec_err_t res =
      aom_codec_encode(ecodec, img, frame_in, 1, 0, AOM_DL_GOOD_QUALITY);
  if (res != AOM_CODEC_OK) die_codec(ecodec, "Failed to encode frame");
239 240 241

  got_data = 0;

Yaowu Xu's avatar
Yaowu Xu committed
242
  while ((pkt = aom_codec_get_cx_data(ecodec, &iter)) != NULL) {
243 244
    got_pkts = 1;

Yaowu Xu's avatar
Yaowu Xu committed
245 246
    if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) {
      const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0;
247

Yaowu Xu's avatar
Yaowu Xu committed
248
      if (!(pkt->data.frame.flags & AOM_FRAME_IS_FRAGMENT)) {
clang-format's avatar
clang-format committed
249 250
        *frame_out += 1;
      }
251

Yaowu Xu's avatar
Yaowu Xu committed
252
      if (!aom_video_writer_write_frame(writer, pkt->data.frame.buf,
253 254 255 256 257 258 259 260 261 262
                                        pkt->data.frame.sz,
                                        pkt->data.frame.pts)) {
        die_codec(ecodec, "Failed to write compressed frame");
      }
      printf(keyframe ? "K" : ".");
      fflush(stdout);
      got_data = 1;

      // Decode 1 frame.
      if (test_decode) {
Yaowu Xu's avatar
Yaowu Xu committed
263
        if (aom_codec_decode(dcodec, pkt->data.frame.buf,
264 265 266 267 268 269 270 271
                             (unsigned int)pkt->data.frame.sz, NULL, 0))
          die_codec(dcodec, "Failed to decode frame.");
      }
    }
  }

  // Mismatch checking
  if (got_data && test_decode) {
272
    testing_decode(ecodec, dcodec, *frame_out, mismatch_seen);
273 274 275 276 277 278 279 280
  }

  return got_pkts;
}

int main(int argc, char **argv) {
  FILE *infile = NULL;
  // Encoder
281 282
  aom_codec_ctx_t ecodec;
  aom_codec_enc_cfg_t cfg;
283
  unsigned int frame_in = 0;
Yaowu Xu's avatar
Yaowu Xu committed
284 285
  aom_image_t raw;
  aom_codec_err_t res;
286
  AvxVideoInfo info;
Yaowu Xu's avatar
Yaowu Xu committed
287 288
  AvxVideoWriter *writer = NULL;
  const AvxInterface *encoder = NULL;
289 290 291 292

  // Test encoder/decoder mismatch.
  int test_decode = 1;
  // Decoder
Yaowu Xu's avatar
Yaowu Xu committed
293
  aom_codec_ctx_t dcodec;
294 295 296
  unsigned int frame_out = 0;

  // The frame number to set reference frame on
297
  unsigned int update_frame_num = 0;
298 299 300 301 302 303 304 305 306 307
  int mismatch_seen = 0;

  const int fps = 30;
  const int bitrate = 500;

  const char *codec_arg = NULL;
  const char *width_arg = NULL;
  const char *height_arg = NULL;
  const char *infile_arg = NULL;
  const char *outfile_arg = NULL;
308
  const char *update_frame_num_arg = NULL;
309
  unsigned int limit = 0;
310 311
  exec_name = argv[0];

312 313 314 315 316 317
  // Clear explicitly, as simply assigning "{ 0 }" generates
  // "missing-field-initializers" warning in some compilers.
  memset(&ecodec, 0, sizeof(ecodec));
  memset(&cfg, 0, sizeof(cfg));
  memset(&info, 0, sizeof(info));

clang-format's avatar
clang-format committed
318
  if (argc < 7) die("Invalid number of arguments");
319 320 321 322 323 324

  codec_arg = argv[1];
  width_arg = argv[2];
  height_arg = argv[3];
  infile_arg = argv[4];
  outfile_arg = argv[5];
325
  update_frame_num_arg = argv[6];
326

Yaowu Xu's avatar
Yaowu Xu committed
327
  encoder = get_aom_encoder_by_name(codec_arg);
clang-format's avatar
clang-format committed
328
  if (!encoder) die("Unsupported codec.");
329

330
  update_frame_num = (unsigned int)strtoul(update_frame_num_arg, NULL, 0);
Yaowu Xu's avatar
Yaowu Xu committed
331 332
  // In AV1, the reference buffers (cm->buffer_pool->frame_bufs[i].buf) are
  // allocated while calling aom_codec_encode(), thus, setting reference for
333
  // 1st frame isn't supported.
334 335 336
  if (update_frame_num <= 1) {
    die("Couldn't parse frame number '%s'\n", update_frame_num_arg);
  }
337 338

  if (argc > 7) {
339
    limit = (unsigned int)strtoul(argv[7], NULL, 0);
340 341 342 343 344 345 346 347 348 349
    if (update_frame_num > limit)
      die("Update frame number couldn't larger than limit\n");
  }

  info.codec_fourcc = encoder->fourcc;
  info.frame_width = strtol(width_arg, NULL, 0);
  info.frame_height = strtol(height_arg, NULL, 0);
  info.time_base.numerator = 1;
  info.time_base.denominator = fps;

clang-format's avatar
clang-format committed
350 351
  if (info.frame_width <= 0 || info.frame_height <= 0 ||
      (info.frame_width % 2) != 0 || (info.frame_height % 2) != 0) {
352 353 354
    die("Invalid frame size: %dx%d", info.frame_width, info.frame_height);
  }

Yaowu Xu's avatar
Yaowu Xu committed
355
  if (!aom_img_alloc(&raw, AOM_IMG_FMT_I420, info.frame_width,
clang-format's avatar
clang-format committed
356
                     info.frame_height, 1)) {
357 358 359
    die("Failed to allocate image.");
  }

Yaowu Xu's avatar
Yaowu Xu committed
360
  printf("Using %s\n", aom_codec_iface_name(encoder->codec_interface()));
361

Yaowu Xu's avatar
Yaowu Xu committed
362
  res = aom_codec_enc_config_default(encoder->codec_interface(), &cfg, 0);
clang-format's avatar
clang-format committed
363
  if (res) die_codec(&ecodec, "Failed to get default codec config.");
364 365 366 367 368 369 370 371

  cfg.g_w = info.frame_width;
  cfg.g_h = info.frame_height;
  cfg.g_timebase.num = info.time_base.numerator;
  cfg.g_timebase.den = info.time_base.denominator;
  cfg.rc_target_bitrate = bitrate;
  cfg.g_lag_in_frames = 3;

Yaowu Xu's avatar
Yaowu Xu committed
372
  writer = aom_video_writer_open(outfile_arg, kContainerIVF, &info);
clang-format's avatar
clang-format committed
373
  if (!writer) die("Failed to open %s for writing.", outfile_arg);
374 375 376 377

  if (!(infile = fopen(infile_arg, "rb")))
    die("Failed to open %s for reading.", infile_arg);

Yaowu Xu's avatar
Yaowu Xu committed
378
  if (aom_codec_enc_init(&ecodec, encoder->codec_interface(), &cfg, 0))
379 380 381
    die_codec(&ecodec, "Failed to initialize encoder");

  // Disable alt_ref.
Yaowu Xu's avatar
Yaowu Xu committed
382
  if (aom_codec_control(&ecodec, AOME_SET_ENABLEAUTOALTREF, 0))
383 384 385
    die_codec(&ecodec, "Failed to set enable auto alt ref");

  if (test_decode) {
Yaowu Xu's avatar
Yaowu Xu committed
386 387
    const AvxInterface *decoder = get_aom_decoder_by_name(codec_arg);
    if (aom_codec_dec_init(&dcodec, decoder->codec_interface(), NULL, 0))
clang-format's avatar
clang-format committed
388
      die_codec(&dcodec, "Failed to initialize decoder.");
389 390 391
  }

  // Encode frames.
Yaowu Xu's avatar
Yaowu Xu committed
392
  while (aom_img_read(&raw, infile)) {
clang-format's avatar
clang-format committed
393
    if (limit && frame_in >= limit) break;
394
    if (update_frame_num > 1 && frame_out + 1 == update_frame_num) {
Yaowu Xu's avatar
Yaowu Xu committed
395 396
      aom_ref_frame_t ref;
      ref.frame_type = AOM_LAST_FRAME;
397 398
      ref.img = raw;
      // Set reference frame in encoder.
Yaowu Xu's avatar
Yaowu Xu committed
399
      if (aom_codec_control(&ecodec, AOM_SET_REFERENCE, &ref))
400 401 402 403 404 405
        die_codec(&ecodec, "Failed to set reference frame");
      printf(" <SET_REF>");

      // If set_reference in decoder is commented out, the enc/dec mismatch
      // would be seen.
      if (test_decode) {
Yaowu Xu's avatar
Yaowu Xu committed
406
        if (aom_codec_control(&dcodec, AOM_SET_REFERENCE, &ref))
407 408 409 410
          die_codec(&dcodec, "Failed to set reference frame");
      }
    }

411
    encode_frame(&ecodec, &raw, frame_in, writer, test_decode, &dcodec,
clang-format's avatar
clang-format committed
412
                 &frame_out, &mismatch_seen);
413
    frame_in++;
clang-format's avatar
clang-format committed
414
    if (mismatch_seen) break;
415 416 417 418
  }

  // Flush encoder.
  if (!mismatch_seen)
419 420
    while (encode_frame(&ecodec, NULL, frame_in, writer, test_decode, &dcodec,
                        &frame_out, &mismatch_seen)) {
clang-format's avatar
clang-format committed
421
    }
422 423 424 425 426 427 428 429 430 431 432 433 434

  printf("\n");
  fclose(infile);
  printf("Processed %d frames.\n", frame_out);

  if (test_decode) {
    if (!mismatch_seen)
      printf("Encoder/decoder results are matching.\n");
    else
      printf("Encoder/decoder results are NOT matching.\n");
  }

  if (test_decode)
Yaowu Xu's avatar
Yaowu Xu committed
435
    if (aom_codec_destroy(&dcodec))
436 437
      die_codec(&dcodec, "Failed to destroy decoder");

Yaowu Xu's avatar
Yaowu Xu committed
438 439
  aom_img_free(&raw);
  if (aom_codec_destroy(&ecodec))
440 441
    die_codec(&ecodec, "Failed to destroy encoder.");

Yaowu Xu's avatar
Yaowu Xu committed
442
  aom_video_writer_close(writer);
443 444 445

  return EXIT_SUCCESS;
}