Commit ccef8842 authored by Jingning Han's avatar Jingning Han Committed by Gerrit Code Review

Allow full coeff probability model and cost update

This commit moves the simplified coefficient probability model
and costing update to speed 4, and turns on chessboard pattern
mode search for sub 720p sequences. The overall coding performance
of speed 3 is improved:
derf  0.889%
stdhd 1.744%

The speed 3 runtime for test sequences are improved:
bus cif at 1000 kbps 9823 ms -> 9642 ms
pedestrian 1080p 2000 kbps 189559 ms -> 183284 ms

Change-Id: Iecbc7496a68f31fd49fb09f8dfd97c028d675a5d
parent 1e305479
......@@ -110,14 +110,13 @@ static void set_good_speed_feature(VP9_COMP *cpi, VP9_COMMON *cm,
if (speed >= 3) {
sf->tx_size_search_method = frame_is_intra_only(cm) ? USE_FULL_RD
: USE_LARGESTALL;
if (MIN(cm->width, cm->height) >= 720) {
if (MIN(cm->width, cm->height) >= 720)
sf->disable_split_mask = DISABLE_ALL_SPLIT;
sf->cb_partition_search = frame_is_boosted(cpi) ? 0 : 1;
} else {
else
sf->disable_split_mask = DISABLE_ALL_INTER_SPLIT;
}
sf->adaptive_pred_interp_filter = 0;
sf->cb_partition_search = frame_is_boosted(cpi) ? 0 : 1;
sf->cb_pred_filter_search = 1;
sf->motion_field_mode_search = frame_is_boosted(cpi) ? 0 : 1;
......@@ -126,8 +125,6 @@ static void set_good_speed_feature(VP9_COMP *cpi, VP9_COMMON *cm,
sf->recode_loop = ALLOW_RECODE_KFMAXBW;
sf->adaptive_rd_thresh = 3;
sf->mode_skip_start = 6;
sf->use_fast_coef_updates = ONE_LOOP_REDUCED;
sf->use_fast_coef_costing = 1;
}
if (speed >= 4) {
......@@ -140,6 +137,8 @@ static void set_good_speed_feature(VP9_COMP *cpi, VP9_COMMON *cm,
sf->disable_filter_search_var_thresh = 200;
sf->use_lastframe_partitioning = LAST_FRAME_PARTITION_ALL;
sf->use_lp32x32fdct = 1;
sf->use_fast_coef_updates = ONE_LOOP_REDUCED;
sf->use_fast_coef_costing = 1;
}
if (speed >= 5) {
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment