Remove unneeded constraints in tpl inference model
Remove the additional constraint that multiplies 7/8 to the block
qp offset in the tpl model inference. The new tpl kernel allows
more precise track of the information propagation flow, hence no
need to add in this heuristic. It improves the coding performance.
In speed 1 vbr 150 frames setting, the changes are
avg PSNR ovr PSNR SSIM
lowres -0.01% -0.02% -0.39%
midres -0.04% -0.06% -0.38%
ugc360p -0.01% 0.0% -0.16%
STATS_CHANGED
Change-Id: Ie951d50a15e640cff1996b3d8a844e3d0829ea01
diff --git a/av1/encoder/encodeframe.c b/av1/encoder/encodeframe.c
index be09a1f..05e6055 100644
--- a/av1/encoder/encodeframe.c
+++ b/av1/encoder/encodeframe.c
@@ -3732,8 +3732,7 @@
// analysis_type 2: Use cost reduction from intra to inter for best inter
// predictor chosen
static int get_q_for_deltaq_objective(AV1_COMP *const cpi, BLOCK_SIZE bsize,
- int analysis_type, int mi_row,
- int mi_col) {
+ int mi_row, int mi_col) {
AV1_COMMON *const cm = &cpi->common;
assert(IMPLIES(cpi->gf_group.size > 0,
cpi->gf_group.index < cpi->gf_group.size));
@@ -3789,28 +3788,13 @@
int offset = 0;
double beta = 1.0;
- if (analysis_type == 0) {
- if (mc_dep_cost > 0 && intra_cost > 0) {
- const double r0 = cpi->rd.r0;
- const double rk = (double)intra_cost / mc_dep_cost;
- beta = (r0 / rk);
- assert(beta > 0.0);
- }
-#if !USE_TPL_CLASSIC_MODEL
- } else if (analysis_type == 1) {
- const double mc_count_base = (mi_count * cpi->rd.mc_count_base);
- beta = (mc_count + 1.0) / (mc_count_base + 1.0);
- beta = pow(beta, 0.5);
- } else if (analysis_type == 2) {
- const double mc_saved_base = (mi_count * cpi->rd.mc_saved_base);
- beta = (mc_saved + 1.0) / (mc_saved_base + 1.0);
- beta = pow(beta, 0.5);
-#endif // !USE_TPL_CLASSIC_MODEL
+ if (mc_dep_cost > 0 && intra_cost > 0) {
+ const double r0 = cpi->rd.r0;
+ const double rk = (double)intra_cost / mc_dep_cost;
+ beta = (r0 / rk);
+ assert(beta > 0.0);
}
- offset = (7 * av1_get_deltaq_offset(cpi, cm->base_qindex, beta)) / 8;
- // printf("[%d/%d]: beta %g offset %d\n", pyr_lev_from_top,
- // cpi->gf_group.pyramid_height, beta, offset);
-
+ offset = av1_get_deltaq_offset(cpi, cm->base_qindex, beta);
aom_clear_system_state();
const DeltaQInfo *const delta_q_info = &cm->delta_q_info;
@@ -3852,8 +3836,7 @@
} else if (cpi->oxcf.deltaq_mode == DELTA_Q_OBJECTIVE) {
assert(cpi->oxcf.enable_tpl_model);
// Setup deltaq based on tpl stats
- current_qindex =
- get_q_for_deltaq_objective(cpi, sb_size, 0, mi_row, mi_col);
+ current_qindex = get_q_for_deltaq_objective(cpi, sb_size, mi_row, mi_col);
}
const int delta_q_res = delta_q_info->delta_q_res;
