shithub: libvpx

Info  •  Files  •  Log  •  Branches

vp9: Adjustments to nonrd-pickmode for vbr

For VBR: (1) allow newmv mode for golden ref to
select interpolation filter (as in last ref case), and
(2) don't use the more aggressive tx-skip testing logic for large blocks.

Only affects 1 pass real-time vbr mode (speed >= 5).

PSNR/SSIM metrics on ytlive set are all positive, ~0.5-2% gain.

Change-Id: I0ffbb0a9755563a5acd6230c58236e4f19a47266

--- a/vp9/encoder/vp9_pickmode.c

+++ b/vp9/encoder/vp9_pickmode.c

@@ -1667,8 +1667,9 @@

     if ((this_mode == NEWMV || filter_ref == SWITCHABLE) && pred_filter_search

         && (ref_frame == LAST_FRAME ||

-            (ref_frame == GOLDEN_FRAME && cpi->use_svc))

-        && (((mi->mv[0].as_mv.row | mi->mv[0].as_mv.col) & 0x07) != 0)) {

+           (ref_frame == GOLDEN_FRAME &&

+           (cpi->use_svc || cpi->oxcf.rc_mode == VPX_VBR))) &&

+           (((mi->mv[0].as_mv.row | mi->mv[0].as_mv.col) & 0x07) != 0)) {

       int pf_rate[3];

       int64_t pf_dist[3];

       unsigned int pf_var[3];

@@ -1723,7 +1724,7 @@

       vp9_build_inter_predictors_sby(xd, mi_row, mi_col, bsize);

       // For large partition blocks, extra testing is done.

-      if (bsize > BLOCK_32X32 &&

+      if (cpi->oxcf.rc_mode == VPX_CBR && bsize > BLOCK_32X32 &&

         !cyclic_refresh_segment_id_boosted(xd->mi[0]->segment_id) &&

         cm->base_qindex) {

         model_rd_for_sb_y_large(cpi, bsize, x, xd, &this_rdc.rate,