scalable_encoder.c (10669B)
1 /* 2 * Copyright (c) 2018, 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 // Scalable Encoder 13 // ============== 14 // 15 // This is an example of a scalable encoder loop. It takes two input files in 16 // YV12 format, passes it through the encoder, and writes the compressed 17 // frames to disk in OBU format. 18 // 19 // Getting The Default Configuration 20 // --------------------------------- 21 // Encoders have the notion of "usage profiles." For example, an encoder 22 // may want to publish default configurations for both a video 23 // conferencing application and a best quality offline encoder. These 24 // obviously have very different default settings. Consult the 25 // documentation for your codec to see if it provides any default 26 // configurations. All codecs provide a default configuration, number 0, 27 // which is valid for material in the vacinity of QCIF/QVGA. 28 // 29 // Updating The Configuration 30 // --------------------------------- 31 // Almost all applications will want to update the default configuration 32 // with settings specific to their usage. Here we set the width and height 33 // of the video file to that specified on the command line. We also scale 34 // the default bitrate based on the ratio between the default resolution 35 // and the resolution specified on the command line. 36 // 37 // Encoding A Frame 38 // ---------------- 39 // The frame is read as a continuous block (size = width * height * 3 / 2) 40 // from the input file. If a frame was read (the input file has not hit 41 // EOF) then the frame is passed to the encoder. Otherwise, a NULL 42 // is passed, indicating the End-Of-Stream condition to the encoder. The 43 // `frame_cnt` is reused as the presentation time stamp (PTS) and each 44 // frame is shown for one frame-time in duration. The flags parameter is 45 // unused in this example. 46 47 // Forced Keyframes 48 // ---------------- 49 // Keyframes can be forced by setting the AOM_EFLAG_FORCE_KF bit of the 50 // flags passed to `aom_codec_control()`. In this example, we force a 51 // keyframe every <keyframe-interval> frames. Note, the output stream can 52 // contain additional keyframes beyond those that have been forced using the 53 // AOM_EFLAG_FORCE_KF flag because of automatic keyframe placement by the 54 // encoder. 55 // 56 // Processing The Encoded Data 57 // --------------------------- 58 // Each packet of type `AOM_CODEC_CX_FRAME_PKT` contains the encoded data 59 // for this frame. We write a IVF frame header, followed by the raw data. 60 // 61 // Cleanup 62 // ------- 63 // The `aom_codec_destroy` call frees any memory allocated by the codec. 64 // 65 // Error Handling 66 // -------------- 67 // This example does not special case any error return codes. If there was 68 // an error, a descriptive message is printed and the program exits. With 69 // few exeptions, aom_codec functions return an enumerated error status, 70 // with the value `0` indicating success. 71 72 #include <stdio.h> 73 #include <stdlib.h> 74 #include <string.h> 75 76 #include "aom/aom_encoder.h" 77 #include "aom/aomcx.h" 78 #include "av1/common/enums.h" 79 #include "common/tools_common.h" 80 #include "common/video_writer.h" 81 82 static const char *exec_name; 83 84 void usage_exit(void) { 85 fprintf(stderr, 86 "Usage: %s <codec> <width> <height> <infile0> <infile1> " 87 "<outfile> <frames to encode>\n" 88 "See comments in scalable_encoder.c for more information.\n", 89 exec_name); 90 exit(EXIT_FAILURE); 91 } 92 93 static int encode_frame(aom_codec_ctx_t *codec, aom_image_t *img, 94 int frame_index, int flags, FILE *outfile) { 95 int got_pkts = 0; 96 aom_codec_iter_t iter = NULL; 97 const aom_codec_cx_pkt_t *pkt = NULL; 98 const aom_codec_err_t res = 99 aom_codec_encode(codec, img, frame_index, 1, flags); 100 if (res != AOM_CODEC_OK) die_codec(codec, "Failed to encode frame"); 101 102 while ((pkt = aom_codec_get_cx_data(codec, &iter)) != NULL) { 103 got_pkts = 1; 104 105 if (pkt->kind == AOM_CODEC_CX_FRAME_PKT) { 106 const int keyframe = (pkt->data.frame.flags & AOM_FRAME_IS_KEY) != 0; 107 if (fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz, outfile) != 108 pkt->data.frame.sz) { 109 die_codec(codec, "Failed to write compressed frame"); 110 } 111 printf(keyframe ? "K" : "."); 112 printf(" %6d\n", (int)pkt->data.frame.sz); 113 fflush(stdout); 114 } 115 } 116 117 return got_pkts; 118 } 119 120 int main(int argc, char **argv) { 121 FILE *infile0 = NULL; 122 FILE *infile1 = NULL; 123 aom_codec_enc_cfg_t cfg; 124 int frame_count = 0; 125 aom_image_t raw0, raw1; 126 aom_codec_err_t res; 127 AvxVideoInfo info; 128 const int fps = 30; 129 const int bitrate = 200; 130 int keyframe_interval = 0; 131 int max_frames = 0; 132 int frames_encoded = 0; 133 const char *codec_arg = NULL; 134 const char *width_arg = NULL; 135 const char *height_arg = NULL; 136 const char *infile0_arg = NULL; 137 const char *infile1_arg = NULL; 138 const char *outfile_arg = NULL; 139 // const char *keyframe_interval_arg = NULL; 140 FILE *outfile = NULL; 141 142 exec_name = argv[0]; 143 144 // Clear explicitly, as simply assigning "{ 0 }" generates 145 // "missing-field-initializers" warning in some compilers. 146 memset(&info, 0, sizeof(info)); 147 148 if (argc != 8) die("Invalid number of arguments"); 149 150 codec_arg = argv[1]; 151 width_arg = argv[2]; 152 height_arg = argv[3]; 153 infile0_arg = argv[4]; 154 infile1_arg = argv[5]; 155 outfile_arg = argv[6]; 156 max_frames = (int)strtol(argv[7], NULL, 0); 157 158 aom_codec_iface_t *encoder = get_aom_encoder_by_short_name(codec_arg); 159 if (!encoder) die("Unsupported codec."); 160 161 info.codec_fourcc = get_fourcc_by_aom_encoder(encoder); 162 info.frame_width = (int)strtol(width_arg, NULL, 0); 163 info.frame_height = (int)strtol(height_arg, NULL, 0); 164 info.time_base.numerator = 1; 165 info.time_base.denominator = fps; 166 167 if (info.frame_width <= 0 || info.frame_height <= 0 || 168 (info.frame_width % 2) != 0 || (info.frame_height % 2) != 0) { 169 die("Invalid frame size: %dx%d", info.frame_width, info.frame_height); 170 } 171 172 if (!aom_img_alloc(&raw0, AOM_IMG_FMT_I420, info.frame_width, 173 info.frame_height, 1)) { 174 die("Failed to allocate image for layer 0."); 175 } 176 if (!aom_img_alloc(&raw1, AOM_IMG_FMT_I420, info.frame_width, 177 info.frame_height, 1)) { 178 die("Failed to allocate image for layer 1."); 179 } 180 181 // keyframe_interval = (int)strtol(keyframe_interval_arg, NULL, 0); 182 keyframe_interval = 100; 183 if (keyframe_interval < 0) die("Invalid keyframe interval value."); 184 185 printf("Using %s\n", aom_codec_iface_name(encoder)); 186 187 aom_codec_ctx_t codec; 188 res = aom_codec_enc_config_default(encoder, &cfg, 0); 189 if (res) die_codec(&codec, "Failed to get default codec config."); 190 191 cfg.g_w = info.frame_width; 192 cfg.g_h = info.frame_height; 193 cfg.g_timebase.num = info.time_base.numerator; 194 cfg.g_timebase.den = info.time_base.denominator; 195 cfg.rc_target_bitrate = bitrate; 196 cfg.g_error_resilient = 0; 197 cfg.g_lag_in_frames = 0; 198 cfg.rc_end_usage = AOM_Q; 199 cfg.save_as_annexb = 0; 200 201 outfile = fopen(outfile_arg, "wb"); 202 if (!outfile) die("Failed to open %s for writing.", outfile_arg); 203 204 if (!(infile0 = fopen(infile0_arg, "rb"))) 205 die("Failed to open %s for reading.", infile0_arg); 206 if (!(infile1 = fopen(infile1_arg, "rb"))) 207 die("Failed to open %s for reading.", infile0_arg); 208 209 if (aom_codec_enc_init(&codec, encoder, &cfg, 0)) 210 die("Failed to initialize encoder"); 211 if (aom_codec_control(&codec, AOME_SET_CPUUSED, 8)) 212 die_codec(&codec, "Failed to set cpu to 8"); 213 214 if (aom_codec_control(&codec, AV1E_SET_TILE_COLUMNS, 2)) 215 die_codec(&codec, "Failed to set tile columns to 2"); 216 if (aom_codec_control(&codec, AV1E_SET_NUM_TG, 3)) 217 die_codec(&codec, "Failed to set num of tile groups to 3"); 218 219 if (aom_codec_control(&codec, AOME_SET_NUMBER_SPATIAL_LAYERS, 2)) 220 die_codec(&codec, "Failed to set number of spatial layers to 2"); 221 222 // Encode frames. 223 while (aom_img_read(&raw0, infile0)) { 224 int flags = 0; 225 226 // configure and encode base layer 227 228 if (keyframe_interval > 0 && frames_encoded % keyframe_interval == 0) 229 flags |= AOM_EFLAG_FORCE_KF; 230 else 231 // use previous base layer (LAST) as sole reference 232 // save this frame as LAST to be used as reference by enhanmcent layer 233 // and next base layer 234 flags |= AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 | 235 AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF | 236 AOM_EFLAG_NO_REF_BWD | AOM_EFLAG_NO_REF_ARF2 | 237 AOM_EFLAG_NO_UPD_GF | AOM_EFLAG_NO_UPD_ARF | 238 AOM_EFLAG_NO_UPD_ENTROPY; 239 cfg.g_w = info.frame_width; 240 cfg.g_h = info.frame_height; 241 if (aom_codec_enc_config_set(&codec, &cfg)) 242 die_codec(&codec, "Failed to set enc cfg for layer 0"); 243 if (aom_codec_control(&codec, AOME_SET_SPATIAL_LAYER_ID, 0)) 244 die_codec(&codec, "Failed to set layer id to 0"); 245 if (aom_codec_control(&codec, AOME_SET_CQ_LEVEL, 62)) 246 die_codec(&codec, "Failed to set cq level"); 247 encode_frame(&codec, &raw0, frame_count++, flags, outfile); 248 249 // configure and encode enhancement layer 250 251 // use LAST (base layer) as sole reference 252 flags = AOM_EFLAG_NO_REF_LAST2 | AOM_EFLAG_NO_REF_LAST3 | 253 AOM_EFLAG_NO_REF_GF | AOM_EFLAG_NO_REF_ARF | AOM_EFLAG_NO_REF_BWD | 254 AOM_EFLAG_NO_REF_ARF2 | AOM_EFLAG_NO_UPD_LAST | 255 AOM_EFLAG_NO_UPD_GF | AOM_EFLAG_NO_UPD_ARF | 256 AOM_EFLAG_NO_UPD_ENTROPY; 257 cfg.g_w = info.frame_width; 258 cfg.g_h = info.frame_height; 259 aom_img_read(&raw1, infile1); 260 if (aom_codec_enc_config_set(&codec, &cfg)) 261 die_codec(&codec, "Failed to set enc cfg for layer 1"); 262 if (aom_codec_control(&codec, AOME_SET_SPATIAL_LAYER_ID, 1)) 263 die_codec(&codec, "Failed to set layer id to 1"); 264 if (aom_codec_control(&codec, AOME_SET_CQ_LEVEL, 10)) 265 die_codec(&codec, "Failed to set cq level"); 266 encode_frame(&codec, &raw1, frame_count++, flags, outfile); 267 268 frames_encoded++; 269 270 if (max_frames > 0 && frames_encoded >= max_frames) break; 271 } 272 273 // Flush encoder. 274 while (encode_frame(&codec, NULL, -1, 0, outfile)) continue; 275 276 printf("\n"); 277 fclose(infile0); 278 fclose(infile1); 279 printf("Processed %d frames.\n", frame_count / 2); 280 281 aom_img_free(&raw0); 282 aom_img_free(&raw1); 283 if (aom_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec."); 284 285 fclose(outfile); 286 287 return EXIT_SUCCESS; 288 }