x264 码控-ABR
总体流程
rc初始化
在open encoder的时候会初始化，根据h->param.rc中的参数，初始化h->rc(x264_ratecontrol_t)
h->param.rc中几个重要的默认参数：
h->param.rc.i_qp_constant = -1;
h->param.rc.i_qp_min = 0;
h->param.rc.i_qp_max = 69;
h->param.rc.i_qp_step = 4; // 两帧间qp的最大变化幅度
h->param.rc.f_ip_factor = 1.4;
h->param.rc.f_pb_factor = 1.0; // 开启mb_tree后pb_factor强制为1.0
h->param.rc.f_rf_constant = 23.0;
h->param.rc.f_rate_tolerance = 1.0;
h->param.rc.f_vbv_buffer_init: 0.9;
h->param.rc.f_qcompress = 0.6;
h->param.rc.f_qblur = 0.5;
h->param.rc.f_complexity_blur = 20.0;
h->rc中几个重要的默认参数：
h->rc->qcompress = 1; // 开启mbtree后rc->qcompress强制为0
// 初始QP
#define ABR_INIT_QP 24
rc->accum_p_norm = .01;
rc->accum_p_qp = ABR_INIT_QP * rc->accum_p_norm;
rc->buffer_fill_final =
rc->buffer_fill_final_min = rc->buffer_size * h->param.rc.f_vbv_buffer_init;
rc->qp_constant[SLICE_TYPE_P] = h->param.rc.i_qp_constant;
rc->qp_constant[SLICE_TYPE_I] = h->param.rc.i_qp_constant - rc->ip_offset;
rc->qp_constant[SLICE_TYPE_B] = h->param.rc.i_qp_constant + rc->pb_offset;
h->mb.ip_offset = rc->ip_offset + 0.5;
rc->lstep = pow( 2, h->param.rc.i_qp_step / 6.0 );
rc->last_qscale = qp2qscale( 26 + QP_BD_OFFSET );
初始的predictor：
static const float pred_coeff_table[3] = { 1.0, 1.0, 1.5 };
for( int i = 0; i < 3; i++ )
{
rc->last_qscale_for[i] = qp2qscale( ABR_INIT_QP );
rc->lmin[i] = qp2qscale( h->param.rc.i_qp_min );
rc->lmax[i] = qp2qscale( h->param.rc.i_qp_max );
rc->pred[i].coeff_min = pred_coeff_table[i] / 2;
rc->pred[i].coeff = pred_coeff_table[i];
rc->pred[i].count = 1.0;
rc->pred[i].decay = 0.5;
rc->pred[i].offset = 0.0;
for( int j = 0; j < 2; j++ )
{
rc->row_preds[i][j].coeff_min = .25 / 4;
rc->row_preds[i][j].coeff = .25;
rc->row_preds[i][j].count = 1.0;
rc->row_preds[i][j].decay = 0.5;
rc->row_preds[i][j].offset = 0.0;
}
}
rc->pred_b_from_p->coeff_min = 0.5 / 2;
rc->pred_b_from_p->coeff = 0.5;
rc->pred_b_from_p->count = 1.0;
rc->pred_b_from_p->decay = 0.5;
rc->pred_b_from_p->offset = 0.0;
初始的ratefactor:
// 初始的cplxr_sum
rc->cplxr_sum = .01 * pow( 7.0e5, rc->qcompress ) * pow( h->mb.i_mb_count, 0.5 );
// 初始的每帧期望bit数
rc->wanted_bits_window = 1.0 * rc->bitrate / rc->fps;
初始的cplxr_sum被定义为：
$rc->cplxr-sum = 0.01\times(7\times10^{5})^{rc->qcompress}\times (h->mb.i\_mb\_count)^{0.5}$
ratefactor被定义为：
$ratefactor = \frac{rc-> wanted\_bits\_window}{rc->cplxr\_sum}$
编码前计算帧级QP
计算帧级QP的过程实际就是计算qscale
参考帧的qscale计算
-
初步的计算qscale
$recq = {(\frac{BASE_FRAME_DURATION}{\frac{frame->i\_duration}{2 \times fps}})}^{1-h->param.rc.f\_qcompress}$
$qscale = \frac{recq}{ratefactor}$
其中
-
$BASE_FRAME_DURATION$固定为0.04
-
$ratefactor = \frac{rc-> wanted\_bits\_window}{rc->cplxr\_sum}$
-
-
第一次调整qscale
计算当前已经编完了多少秒的视频：
double time_done = h->i_frame / rcc->fps;
计算截至目前预期的bits：
wanted_bits = time_done * rcc->bitrate;
调整qscale(首帧不调整)：
$overflow=1+\frac{predicted\_bits-wanted\_bits}{MAX(1, \sqrt{time\_done)}}$
$qscale=old\_qscale \times overflow$
其中：
- predicted_bits代表截至目前已经编码完成的帧的实际消耗bits
- clip qscale，防止qscale波动太大
double lmin = rcc->last_qscale_for[pict_type] / rcc->lstep;
double lmax = rcc->last_qscale_for[pict_type] * rcc->lstep;
if( overflow > 1.1 && h->i_frame > 3 )
lmax *= rcc->lstep;
else if( overflow < 0.9 )
lmin /= rcc->lstep;
q = x264_clip3f(q, lmin, lmax);
其中：
- last_qscale_for初始值为qp2qscale(ABR_INIT_QP)，在每次初步计算qscale后将其更小到last_qscale_for中
- lstep代表qscale的的最大变化幅度
- lmin和lmax表示在上一个qscale的基础上，当前帧的qscale能波动的范围
-
第二次调整qscale(VBV)
先列出此时的VBV相关的几个变量
rc->buffer_fill =
- 首先基于上面得到的qscale，预测当前帧消耗的bits 2.