TPL based qp decisions
Enable the TPL based qp decisions in Q mode. Calculate the quantizer
step size scaling based on the temporal dependency. The coding
performance is improved
overall PSNR SSIM
lowres -0.70% -0.09%
midres2 -1.26% -0.75%
ugc480p -2.48% -1.29%
hdres2 -1.47% -1.17%
STATS_CHANGED
Change-Id: Id925ec661c0a00047ad45aad34458620c8a1490c
diff --git a/av1/encoder/encoder_utils.c b/av1/encoder/encoder_utils.c
index 557268f..70edf58 100644
--- a/av1/encoder/encoder_utils.c
+++ b/av1/encoder/encoder_utils.c
@@ -542,6 +542,19 @@
*q = av1_rc_pick_q_and_bounds(cpi, cm->width, cm->height, cpi->gf_frame_index,
bottom_index, top_index);
+#if !CONFIG_REALTIME_ONLY
+ if (cpi->oxcf.rc_cfg.mode == AOM_Q &&
+ cpi->ppi->tpl_data.tpl_frame[cpi->gf_frame_index].is_valid &&
+ is_frame_tpl_eligible(gf_group, cpi->gf_frame_index) &&
+ !frame_is_intra_only(cm)) {
+ *q = av1_get_arf_q_index_q_mode(
+ cpi, &cpi->ppi->tpl_data.tpl_frame[cpi->gf_frame_index]);
+ *top_index = *bottom_index = *q;
+ if (gf_group->update_type[cpi->gf_frame_index] == ARF_UPDATE)
+ cpi->ppi->p_rc.arf_q = *q;
+ }
+#endif
+
// Configure experimental use of segmentation for enhanced coding of
// static regions if indicated.
// Only allowed in the second pass of a two pass encode, as it requires
