libaom 源码分析：twopass_encoder.c 文件

libaom

libaom 是 AOMedia（开放媒体联盟）开发的一个开源视频编解码器库，它是 AV1 视频压缩格式的参考实现，并被广泛用于多种生产系统中。libaom 支持多种功能，包括可扩展视频编码（SVC）、实时通信（RTC）优化等，并定期进行更新以提高压缩效率和编码速度 。

libaom 的一些关键特性包括：

1. 多空间层和时间层编码：通过 aom_svc_layer_id_t 结构体支持空间层和时间层的ID标识，允许视频在不同的分辨率和帧率下进行编码 。
2. 编码参数配置：通过 aom_svc_params_t 结构体等配置编码参数，如空间层数量、时间层数量、量化器、缩放因子等 。
3. 基础编码参数：aom_codec_enc_cfg_t 结构体用于配置编码器的基础参数，如使用方式、时间基准、编码通道、帧重采样等 。
4. 多遍编码模式：支持多遍编码模式，包括单遍、双遍和多遍编码，以优化编码效率和质量 。
5. 帧超分采样：支持帧超分辨率模式，通过 rc_superres_mode 枚举值控制放大过程 。
6. 关键帧放置：支持关键帧放置模式，通过 kf_mode 枚举值决定是否自动放置关键帧 。
7. SVC 编码参数：支持 SVC 编码的参数类型配置，如层数量、量化器、缩放因子等 。

libaom 的更新通常每三个月进行一次，最近的更新包括对 SVC 丢帧模式的支持、新的构建配置以减小二进制文件大小、以及对 RTC 屏幕内容压缩效率的显著提升 。此外，libaom 还提供了对 AV1 视频压缩格式的支持，包括实时编码模式和对不同质量控制策略的优化 。

twopass_encoder.c 介绍

• 功能：两遍编码循环的 demo 输入yv12 格式，输出 ivf 格式。
• 文件位置：libaom/examples/twopass_encoder.c

函数关系

结构体

1. FILE：文件结构体
2. aom_codec_ctx_t：编解码上下文结构体
3. aom_codec_enc_cfg_t：编码器配置结构体
4. aom_image_t：输入图像结构体
5. aom_codec_err_t：算法返回编码状态码结构体
6. aom_fixed_buf_t：产生固定大小 buffer 结构体
7. aom_codec_iface_t：编解码接口结构体
8. AvxVideoInfo：av1 编码视频信息结构体
9. AvxVideoWriter：视频信息写入结构体

2pass 编码原理

1. 数据流转图
2. 核心原理：第一遍编码产生的aom_fixed_buf_t数据赋值给aom_codec_enc_cfg_t中的 rc_twopass_stats_in(aom_fixed_buf_t) 供第二遍编码使用；
3. aom_fixed_buf_t 结构体：

/*!\brief Generic fixed size buffer structure** This structure is able to hold a reference to any fixed size buffer.*/
typedef struct aom_fixed_buf {void *buf;       /**< Pointer to the data. Does NOT own the data! */size_t sz;       /**< Length of the buffer, in chars */
} aom_fixed_buf_t; /**< alias for struct aom_fixed_buf */

4. 在 pass0 函数中的get_frame_stats函数对aom_fixed_buf_t结构体进行赋值；

static int get_frame_stats(aom_codec_ctx_t *ctx, const aom_image_t *img,aom_codec_pts_t pts, unsigned int duration,aom_enc_frame_flags_t flags,aom_fixed_buf_t *stats) {int got_pkts = 0;aom_codec_iter_t iter = NULL;const aom_codec_cx_pkt_t *pkt = NULL;const aom_codec_err_t res = aom_codec_encode(ctx, img, pts, duration, flags);if (res != AOM_CODEC_OK) die_codec(ctx, "Failed to get frame stats.");while ((pkt = aom_codec_get_cx_data(ctx, &iter)) != NULL) {got_pkts = 1;if (pkt->kind == AOM_CODEC_STATS_PKT) {const uint8_t *const pkt_buf = pkt->data.twopass_stats.buf;const size_t pkt_size = pkt->data.twopass_stats.sz;stats->buf = realloc(stats->buf, stats->sz + pkt_size);if (!stats->buf) die("Failed to allocate frame stats buffer.");memcpy((uint8_t *)stats->buf + stats->sz, pkt_buf, pkt_size);stats->sz += pkt_size;}}return got_pkts;
}

5. 在函数 set_encoder_config 中对aom_codec_enc_cfg_t中的 rc_twopass_stats_in(aom_fixed_buf_t) 进行应用；根据aom_fixed_buf_t的 sz 大小除以每个包的状态大小FIRSTPASS_STATS，作为输入配置中的 limit 变量的值；

  if (cfg->g_pass >= AOM_RC_SECOND_PASS) {const size_t packet_sz = sizeof(FIRSTPASS_STATS);const int n_packets = (int)(cfg->rc_twopass_stats_in.sz / packet_sz);input_cfg->limit = n_packets - 1;} else {input_cfg->limit = cfg->g_limit;}

6. 在 validate_config 函数中对FIRSTPASS_STATS进行赋值，用来访问第一遍编码的统计信息。

if (cfg->g_pass >= AOM_RC_SECOND_PASS) {const size_t packet_sz = sizeof(FIRSTPASS_STATS);const int n_packets = (int)(cfg->rc_twopass_stats_in.sz / packet_sz);const FIRSTPASS_STATS *stats;if (cfg->rc_twopass_stats_in.buf == NULL)ERROR("rc_twopass_stats_in.buf not set.");if (cfg->rc_twopass_stats_in.sz % packet_sz)ERROR("rc_twopass_stats_in.sz indicates truncated packet.");if (cfg->rc_twopass_stats_in.sz < 2 * packet_sz)ERROR("rc_twopass_stats_in requires at least two packets.");stats =(const FIRSTPASS_STATS *)cfg->rc_twopass_stats_in.buf + n_packets - 1;if ((int)(stats->count + 0.5) != n_packets - 1)ERROR("rc_twopass_stats_in missing EOS stats packet");}

7. FIRSTPASS_STATS的定义如下，这个结构体用于在视频编码的第一遍（分析遍）中累积帧统计信息。这些统计数据有助于在第二遍（编码遍）中优化码率分配和提高编码质量，该结构体包含了帧、权重、mv 相关、帧编码信息等等变量。

/*!* \brief The stucture of acummulated frame stats in the first pass.** Errors (coded_error, intra_error, etc.) and counters (new_mv_count) are* normalized to each MB. MV related stats (MVc, MVr, etc.) are normalized to* the frame width and height. See function normalize_firstpass_stats.*/
typedef struct FIRSTPASS_STATS {/*!* Frame number in display order, if stats are for a single frame.* No real meaning for a collection of frames.*/double frame;/*!* Weight assigned to this frame (or total weight for the collection of* frames) currently based on intra factor and brightness factor. This is used* to distribute bits betweeen easier and harder frames.*/double weight;/*!* Intra prediction error.*/double intra_error;/*!* Average wavelet energy computed using Discrete Wavelet Transform (DWT).*/double frame_avg_wavelet_energy;/*!* Best of intra pred error and inter pred error using last frame as ref.*/double coded_error;/*!* Best of intra pred error and inter pred error using golden frame as ref.*/double sr_coded_error;/*!* Percentage of blocks with inter pred error < intra pred error.*/double pcnt_inter;/*!* Percentage of blocks using (inter prediction and) non-zero motion vectors.*/double pcnt_motion;/*!* Percentage of blocks where golden frame was better than last or intra:* inter pred error using golden frame < inter pred error using last frame and* inter pred error using golden frame < intra pred error*/double pcnt_second_ref;/*!* Percentage of blocks where intra and inter prediction errors were very* close. Note that this is a 'weighted count', that is, the so blocks may be* weighted by how close the two errors were.*/double pcnt_neutral;/*!* Percentage of blocks that have almost no intra error residual* (i.e. are in effect completely flat and untextured in the intra* domain). In natural videos this is uncommon, but it is much more* common in animations, graphics and screen content, so may be used* as a signal to detect these types of content.*/double intra_skip_pct;/*!* Image mask rows top and bottom.*/double inactive_zone_rows;/*!* Image mask columns at left and right edges.*/double inactive_zone_cols;/*!* Average of row motion vectors.*/double MVr;/*!* Mean of absolute value of row motion vectors.*/double mvr_abs;/*!* Mean of column motion vectors.*/double MVc;/*!* Mean of absolute value of column motion vectors.*/double mvc_abs;/*!* Variance of row motion vectors.*/double MVrv;/*!* Variance of column motion vectors.*/double MVcv;/*!* Value in range [-1,1] indicating fraction of row and column motion vectors* that point inwards (negative MV value) or outwards (positive MV value).* For example, value of 1 indicates, all row/column MVs are inwards.*/double mv_in_out_count;/*!* Count of unique non-zero motion vectors.*/double new_mv_count;/*!* Duration of the frame / collection of frames.*/double duration;/*!* 1.0 if stats are for a single frame, OR* Number of frames in this collection for which the stats are accumulated.*/double count;/*!* standard deviation for (0, 0) motion prediction error*/double raw_error_stdev;/*!* Whether the frame contains a flash*/int64_t is_flash;/*!* Estimated noise variance*/double noise_var;/*!* Correlation coefficient with the previous frame*/double cor_coeff;/*!* log of intra_error*/double log_intra_error;/*!* log of coded_error*/double log_coded_error;
} FIRSTPASS_STATS;