Improve rdopt decisions for ext-inter
Relative to previous ext-inter:
lowres: -0.177%
or -0.029% (with USE_RECT_INTERINTRA = 0)
* When predicting interintra modes, the previous code did not provide
the intra predictor with the correct context during rdopt. Add an
explicit 'ctx' parameter to the relevant functions, to provide this
context.
This fixes a nondeterminism bug, which was causing test failures in
*EncoderThreadTest*
* For rectangular blocks, build_intra_predictors_for_interintra needs
to overwrite part of the context buffer in order to set up the
correct context for intra prediction. We now restore the original
contents afterwards.
* Add a flag to enable/disable rectangular interintra prediction;
disabling improves encoding speed but reduces BDRATE improvement.
Change-Id: I7458c036c7f94df9ab1ba0c7efa79aeaa7e17118
diff --git a/av1/encoder/encodeframe.c b/av1/encoder/encodeframe.c
index d26f65f..7e39cc5 100644
--- a/av1/encoder/encodeframe.c
+++ b/av1/encoder/encodeframe.c
@@ -919,7 +919,7 @@
x->pred_mv[LAST_FRAME] = mbmi->mv[0].as_mv;
}
- av1_build_inter_predictors_sb(xd, mi_row, mi_col, cm->sb_size);
+ av1_build_inter_predictors_sb(xd, mi_row, mi_col, NULL, cm->sb_size);
ref = xd->plane[0].dst.buf;
ref_stride = xd->plane[0].dst.stride;
@@ -5550,10 +5550,10 @@
} else {
#endif // CONFIG_WARPED_MOTION
if (!(cpi->sf.reuse_inter_pred_sby && ctx->pred_pixel_ready) || seg_skip)
- av1_build_inter_predictors_sby(xd, mi_row, mi_col,
+ av1_build_inter_predictors_sby(xd, mi_row, mi_col, NULL,
AOMMAX(bsize, BLOCK_8X8));
- av1_build_inter_predictors_sbuv(xd, mi_row, mi_col,
+ av1_build_inter_predictors_sbuv(xd, mi_row, mi_col, NULL,
AOMMAX(bsize, BLOCK_8X8));
#if CONFIG_WARPED_MOTION
}
diff --git a/av1/encoder/firstpass.c b/av1/encoder/firstpass.c
index cb577c5..f066b46 100644
--- a/av1/encoder/firstpass.c
+++ b/av1/encoder/firstpass.c
@@ -853,7 +853,8 @@
xd->mi[0]->mbmi.tx_size = TX_4X4;
xd->mi[0]->mbmi.ref_frame[0] = LAST_FRAME;
xd->mi[0]->mbmi.ref_frame[1] = NONE;
- av1_build_inter_predictors_sby(xd, mb_row << 1, mb_col << 1, bsize);
+ av1_build_inter_predictors_sby(xd, mb_row << 1, mb_col << 1, NULL,
+ bsize);
av1_encode_sby_pass1(cm, x, bsize);
sum_mvr += mv.row;
sum_mvr_abs += abs(mv.row);
diff --git a/av1/encoder/mbgraph.c b/av1/encoder/mbgraph.c
index 1fd1682..1f63222 100644
--- a/av1/encoder/mbgraph.c
+++ b/av1/encoder/mbgraph.c
@@ -74,7 +74,7 @@
xd->mi[0]->mbmi.ref_frame[1] = NONE;
#endif // CONFIG_EXT_INTER
- av1_build_inter_predictors_sby(xd, mb_row, mb_col, BLOCK_16X16);
+ av1_build_inter_predictors_sby(xd, mb_row, mb_col, NULL, BLOCK_16X16);
/* restore UMV window */
x->mv_col_min = tmp_col_min;
diff --git a/av1/encoder/rdopt.c b/av1/encoder/rdopt.c
index 941f465..ecdd04d 100644
--- a/av1/encoder/rdopt.c
+++ b/av1/encoder/rdopt.c
@@ -6371,13 +6371,11 @@
}
}
-static INLINE void restore_dst_buf(MACROBLOCKD *xd,
- uint8_t *orig_dst[MAX_MB_PLANE],
- int orig_dst_stride[MAX_MB_PLANE]) {
+static INLINE void restore_dst_buf(MACROBLOCKD *xd, BUFFER_SET dst) {
int i;
for (i = 0; i < MAX_MB_PLANE; i++) {
- xd->plane[i].dst.buf = orig_dst[i];
- xd->plane[i].dst.stride = orig_dst_stride[i];
+ xd->plane[i].dst.buf = dst.plane[i];
+ xd->plane[i].dst.stride = dst.stride[i];
}
}
@@ -6934,8 +6932,8 @@
static int64_t build_and_cost_compound_wedge(
const AV1_COMP *const cpi, MACROBLOCK *x, const int_mv *const cur_mv,
const BLOCK_SIZE bsize, const int this_mode, int rs2, int rate_mv,
- int *out_rate_mv, uint8_t **preds0, uint8_t **preds1, int *strides,
- int mi_row, int mi_col) {
+ BUFFER_SET *ctx, int *out_rate_mv, uint8_t **preds0, uint8_t **preds1,
+ int *strides, int mi_row, int mi_col) {
MACROBLOCKD *xd = &x->e_mbd;
MB_MODE_INFO *const mbmi = &xd->mi[0]->mbmi;
int rate_sum;
@@ -6951,7 +6949,7 @@
if (have_newmv_in_inter_mode(this_mode)) {
*out_rate_mv = interinter_compound_motion_search(cpi, x, bsize, this_mode,
mi_row, mi_col);
- av1_build_inter_predictors_sby(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sby(xd, mi_row, mi_col, ctx, bsize);
model_rd_for_sb(cpi, bsize, x, xd, 0, 0, &rate_sum, &dist_sum,
&tmp_skip_txfm_sb, &tmp_skip_sse_sb);
rd = RDCOST(x->rdmult, x->rddiv, rs2 + *out_rate_mv + rate_sum, dist_sum);
@@ -7056,10 +7054,7 @@
double pts[144], pts_inref[144];
#endif // CONFIG_WARPED_MOTION
int64_t rd = INT64_MAX;
- uint8_t *orig_dst[MAX_MB_PLANE];
- int orig_dst_stride[MAX_MB_PLANE];
- uint8_t *tmp_dst[MAX_MB_PLANE];
- int tmp_dst_stride[MAX_MB_PLANE];
+ BUFFER_SET orig_dst, tmp_dst;
int rs = 0;
InterpFilter assign_filter = SWITCHABLE;
@@ -7312,12 +7307,12 @@
// one for future predictions. In the end, copy from tmp_buf to
// dst if necessary.
for (i = 0; i < MAX_MB_PLANE; i++) {
- tmp_dst[i] = tmp_buf + i * MAX_SB_SQUARE;
- tmp_dst_stride[i] = MAX_SB_SIZE;
+ tmp_dst.plane[i] = tmp_buf + i * MAX_SB_SQUARE;
+ tmp_dst.stride[i] = MAX_SB_SIZE;
}
for (i = 0; i < MAX_MB_PLANE; i++) {
- orig_dst[i] = xd->plane[i].dst.buf;
- orig_dst_stride[i] = xd->plane[i].dst.stride;
+ orig_dst.plane[i] = xd->plane[i].dst.buf;
+ orig_dst.stride[i] = xd->plane[i].dst.stride;
}
// We don't include the cost of the second reference here, because there
@@ -7383,7 +7378,7 @@
assign_filter == SWITCHABLE ? EIGHTTAP_REGULAR : assign_filter;
#endif
rs = av1_get_switchable_rate(cpi, xd);
- av1_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sb(xd, mi_row, mi_col, &orig_dst, bsize);
model_rd_for_sb(cpi, bsize, x, xd, 0, MAX_MB_PLANE - 1, &tmp_rate,
&tmp_dist, &skip_txfm_sb, &skip_sse_sb);
rd = RDCOST(x->rdmult, x->rddiv, rs + tmp_rate, tmp_dist);
@@ -7403,7 +7398,7 @@
#else
InterpFilter best_filter = mbmi->interp_filter;
#endif
- restore_dst_buf(xd, tmp_dst, tmp_dst_stride);
+ restore_dst_buf(xd, tmp_dst);
// EIGHTTAP_REGULAR mode is calculated beforehand
for (i = 1; i < filter_set_size; ++i) {
int tmp_skip_sb = 0;
@@ -7419,7 +7414,7 @@
mbmi->interp_filter = i;
#endif
tmp_rs = av1_get_switchable_rate(cpi, xd);
- av1_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sb(xd, mi_row, mi_col, &orig_dst, bsize);
model_rd_for_sb(cpi, bsize, x, xd, 0, MAX_MB_PLANE - 1, &tmp_rate,
&tmp_dist, &tmp_skip_sb, &tmp_skip_sse);
tmp_rd = RDCOST(x->rdmult, x->rddiv, tmp_rs + tmp_rate, tmp_dist);
@@ -7436,16 +7431,16 @@
skip_sse_sb = tmp_skip_sse;
best_in_temp = !best_in_temp;
if (best_in_temp) {
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
} else {
- restore_dst_buf(xd, tmp_dst, tmp_dst_stride);
+ restore_dst_buf(xd, tmp_dst);
}
}
}
if (best_in_temp) {
- restore_dst_buf(xd, tmp_dst, tmp_dst_stride);
+ restore_dst_buf(xd, tmp_dst);
} else {
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
}
#if CONFIG_DUAL_FILTER
av1_copy(mbmi->interp_filter, best_filter);
@@ -7523,7 +7518,7 @@
switch (cur_type) {
case COMPOUND_AVERAGE:
- av1_build_inter_predictors_sby(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sby(xd, mi_row, mi_col, &orig_dst, bsize);
av1_subtract_plane(x, bsize, 0);
rd = estimate_yrd_for_sb(cpi, bsize, x, &rate_sum, &dist_sum,
&tmp_skip_txfm_sb, &tmp_skip_sse_sb,
@@ -7539,8 +7534,8 @@
best_rd_compound / 3 < ref_best_rd) {
int tmp_rate_mv = 0;
best_rd_cur = build_and_cost_compound_wedge(
- cpi, x, cur_mv, bsize, this_mode, rs2, rate_mv, &tmp_rate_mv,
- preds0, preds1, strides, mi_row, mi_col);
+ cpi, x, cur_mv, bsize, this_mode, rs2, rate_mv, &orig_dst,
+ &tmp_rate_mv, preds0, preds1, strides, mi_row, mi_col);
if (best_rd_cur < best_rd_compound) {
best_rd_compound = best_rd_cur;
@@ -7574,7 +7569,7 @@
}
if (ref_best_rd < INT64_MAX && best_rd_compound / 3 > ref_best_rd) {
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
return INT64_MAX;
}
@@ -7614,15 +7609,16 @@
xd->plane[j].dst.buf = tmp_buf + j * MAX_SB_SQUARE;
xd->plane[j].dst.stride = bw;
}
- av1_build_inter_predictors_sby(xd, mi_row, mi_col, bsize);
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ av1_build_inter_predictors_sby(xd, mi_row, mi_col, &orig_dst, bsize);
+ restore_dst_buf(xd, orig_dst);
mbmi->ref_frame[1] = INTRA_FRAME;
mbmi->use_wedge_interintra = 0;
for (j = 0; j < INTERINTRA_MODES; ++j) {
mbmi->interintra_mode = (INTERINTRA_MODE)j;
rmode = interintra_mode_cost[mbmi->interintra_mode];
- av1_build_intra_predictors_for_interintra(xd, bsize, 0, intrapred, bw);
+ av1_build_intra_predictors_for_interintra(xd, bsize, 0, &orig_dst,
+ intrapred, bw);
av1_combine_interintra(xd, bsize, 0, tmp_buf, bw, intrapred, bw);
model_rd_for_sb(cpi, bsize, x, xd, 0, 0, &rate_sum, &dist_sum,
&tmp_skip_txfm_sb, &tmp_skip_sse_sb);
@@ -7634,7 +7630,8 @@
}
mbmi->interintra_mode = best_interintra_mode;
rmode = interintra_mode_cost[mbmi->interintra_mode];
- av1_build_intra_predictors_for_interintra(xd, bsize, 0, intrapred, bw);
+ av1_build_intra_predictors_for_interintra(xd, bsize, 0, &orig_dst,
+ intrapred, bw);
av1_combine_interintra(xd, bsize, 0, tmp_buf, bw, intrapred, bw);
av1_subtract_plane(x, bsize, 0);
rd = estimate_yrd_for_sb(cpi, bsize, x, &rate_sum, &dist_sum,
@@ -7674,7 +7671,7 @@
do_masked_motion_search(cpi, x, mask, bw, bsize, mi_row, mi_col,
&tmp_mv, &tmp_rate_mv, 0, mv_idx);
mbmi->mv[0].as_int = tmp_mv.as_int;
- av1_build_inter_predictors_sby(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sby(xd, mi_row, mi_col, &orig_dst, bsize);
model_rd_for_sb(cpi, bsize, x, xd, 0, 0, &rate_sum, &dist_sum,
&tmp_skip_txfm_sb, &tmp_skip_sse_sb);
rd = RDCOST(x->rdmult, x->rddiv,
@@ -7746,7 +7743,7 @@
if (pred_exists == 0) {
int tmp_rate;
int64_t tmp_dist;
- av1_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sb(xd, mi_row, mi_col, &orig_dst, bsize);
model_rd_for_sb(cpi, bsize, x, xd, 0, MAX_MB_PLANE - 1, &tmp_rate,
&tmp_dist, &skip_txfm_sb, &skip_sse_sb);
rd = RDCOST(x->rdmult, x->rddiv, rs + tmp_rate, tmp_dist);
@@ -7767,7 +7764,7 @@
int64_t mrd =
AOMMIN(modelled_rd[mode0][refs[0]], modelled_rd[mode1][refs[1]]);
if (rd / 4 * 3 > mrd && ref_best_rd < INT64_MAX) {
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
return INT64_MAX;
}
} else if (!is_comp_interintra_pred) {
@@ -7780,7 +7777,7 @@
// if current pred_error modeled rd is substantially more than the best
// so far, do not bother doing full rd
if (rd / 2 > ref_best_rd) {
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
return INT64_MAX;
}
}
@@ -7849,10 +7846,10 @@
// is needed in only one direction
if (!av1_is_interp_needed(xd)) tmp_rate2 -= rs;
#endif // CONFIG_EXT_INTERP
- av1_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sb(xd, mi_row, mi_col, &orig_dst, bsize);
#if CONFIG_EXT_INTER
} else {
- av1_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sb(xd, mi_row, mi_col, &orig_dst, bsize);
#endif // CONFIG_EXT_INTER
}
av1_build_obmc_inter_prediction(cm, xd, mi_row, mi_col, above_pred_buf,
@@ -7965,7 +7962,7 @@
continue;
} else {
#endif // CONFIG_MOTION_VAR || CONFIG_WARPED_MOTION
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
return INT64_MAX;
#if CONFIG_MOTION_VAR || CONFIG_WARPED_MOTION
}
@@ -7988,7 +7985,7 @@
#if CONFIG_MOTION_VAR || CONFIG_WARPED_MOTION
continue;
#else
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
return INT64_MAX;
#endif // CONFIG_MOTION_VAR || CONFIG_WARPED_MOTION
}
@@ -8091,7 +8088,7 @@
if (best_rd == INT64_MAX) {
av1_invalid_rd_stats(rd_stats);
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
return INT64_MAX;
}
*mbmi = best_mbmi;
@@ -8124,7 +8121,7 @@
#endif // CONFIG_AOM_HIGHBITDEPTH
#endif // !(CONFIG_MOTION_VAR || CONFIG_WARPED_MOTION)
- restore_dst_buf(xd, orig_dst, orig_dst_stride);
+ restore_dst_buf(xd, orig_dst);
return 0; // The rate-distortion cost will be re-calculated by caller.
}
@@ -8912,6 +8909,15 @@
// Mode must by compatible
assert(is_interintra_allowed_mode(this_mode));
+#if !USE_RECT_INTERINTRA
+ // Note: If the subsampling is unequal, any block size we pick
+ // either a rectangular luma block or a rectangular chroma block.
+ // So in this case, we can't use any interintra modes
+ if (xd->plane[1].subsampling_x != xd->plane[1].subsampling_y ||
+ xd->plane[2].subsampling_x != xd->plane[2].subsampling_y)
+ continue;
+#endif
+
if (!is_interintra_allowed_bsize(bsize)) continue;
}
@@ -9715,7 +9721,7 @@
}
} else {
#endif // CONFIG_WARPED_MOTION
- av1_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
+ av1_build_inter_predictors_sb(xd, mi_row, mi_col, NULL, bsize);
#if CONFIG_WARPED_MOTION
}
#endif // CONFIG_WARPED_MOTION
@@ -10901,7 +10907,8 @@
// then dont bother looking at UV
int is_cost_valid_uv;
RD_STATS rd_stats_uv;
- av1_build_inter_predictors_sbuv(&x->e_mbd, mi_row, mi_col, BLOCK_8X8);
+ av1_build_inter_predictors_sbuv(&x->e_mbd, mi_row, mi_col, NULL,
+ BLOCK_8X8);
#if CONFIG_VAR_TX
is_cost_valid_uv =
inter_block_uvrd(cpi, x, &rd_stats_uv, BLOCK_8X8, tmp_best_rdu);
