Support sub8x8 chroma component prediction

This commit allows the sub8x8 blocks to compose and filter their
chroma components for supertx in cb4x4 mode. The coding gains of
supertx and cb4x4 are largely additive:

          supertx      cb4x4       cb4x4 + supertx
lowres     -1.0%       -2.7%        -3.64%
midres     -0.8%       -1.3%        -2.10%

Change-Id: Ie7d09f6fceb36ce375e56773728f05dd628786fe

av1/common/reconinter.c

@@ -1007,8 +1007,8 @@ void av1_setup_pre_planes(MACROBLOCKD *xd, int idx,
 
 #if CONFIG_SUPERTX
 #if CONFIG_CB4X4
-static const uint8_t mask_4[4] = { 64, 62, 12, 0 };
-static const uint8_t mask_4_uv[4] = { 64, 62, 12, 0 };
+static const uint8_t mask_4[4] = { 64, 52, 12, 0 };
+static const uint8_t mask_4_uv[4] = { 64, 52, 12, 0 };
 #endif  // CONFIG_CB4X4
 static const uint8_t mask_8[8] = { 64, 64, 62, 52, 12, 2, 0, 0 };
 

av1/decoder/decodeframe.c

@@ -1044,8 +1044,6 @@ static void dec_predict_sb_complex(AV1Decoder *const pbi, MACROBLOCKD *const xd,
 
           // weighted average to smooth the boundary
           for (i = 0; i < MAX_MB_PLANE; i++) {
-            if (bsize == BLOCK_8X8 && i != 0)
-              continue;  // Skip <4x4 chroma smoothing
             xd->plane[i].dst.buf = dst_buf[i];
             xd->plane[i].dst.stride = dst_stride[i];
             av1_build_masked_inter_predictor_complex(
@@ -1109,8 +1107,6 @@ static void dec_predict_sb_complex(AV1Decoder *const pbi, MACROBLOCKD *const xd,
 
           // Smooth
           for (i = 0; i < MAX_MB_PLANE; i++) {
-            if (bsize == BLOCK_8X8 && i != 0)
-              continue;  // Skip <4x4 chroma smoothing
             xd->plane[i].dst.buf = dst_buf[i];
             xd->plane[i].dst.stride = dst_stride[i];
             av1_build_masked_inter_predictor_complex(
@@ -1163,8 +1159,10 @@ static void dec_predict_sb_complex(AV1Decoder *const pbi, MACROBLOCKD *const xd,
                                  dst_buf3, dst_stride3);
       }
       for (i = 0; i < MAX_MB_PLANE; i++) {
+#if !CONFIG_CB4X4
         if (bsize == BLOCK_8X8 && i != 0)
           continue;  // Skip <4x4 chroma smoothing
+#endif
         if (mi_row < cm->mi_rows && mi_col + hbs < cm->mi_cols) {
           av1_build_masked_inter_predictor_complex(
               xd, dst_buf[i], dst_stride[i], dst_buf1[i], dst_stride1[i],

av1/encoder/encodeframe.c

@@ -6074,9 +6074,6 @@ static void predict_sb_complex(const AV1_COMP *const cpi, ThreadData *td,
 
           // Smooth
           for (i = 0; i < MAX_MB_PLANE; i++) {
-            if (bsize == BLOCK_8X8 && i != 0)
-              continue;  // Skip <4x4 chroma smoothing
-
             xd->plane[i].dst.buf = dst_buf[i];
             xd->plane[i].dst.stride = dst_stride[i];
             av1_build_masked_inter_predictor_complex(
@@ -6138,9 +6135,6 @@ static void predict_sb_complex(const AV1_COMP *const cpi, ThreadData *td,
                        dst_stride1, 2);
 
           for (i = 0; i < MAX_MB_PLANE; i++) {
-            if (bsize == BLOCK_8X8 && i != 0)
-              continue;  // Skip <4x4 chroma smoothing
-
             xd->plane[i].dst.buf = dst_buf[i];
             xd->plane[i].dst.stride = dst_stride[i];
             av1_build_masked_inter_predictor_complex(
@@ -6195,8 +6189,10 @@ static void predict_sb_complex(const AV1_COMP *const cpi, ThreadData *td,
                              pc_tree->split[3]);
       }
       for (i = 0; i < MAX_MB_PLANE; i++) {
+#if !CONFIG_CB4X4
         if (bsize == BLOCK_8X8 && i != 0)
           continue;  // Skip <4x4 chroma smoothing
+#endif
         if (mi_row < cm->mi_rows && mi_col + hbs < cm->mi_cols) {
           av1_build_masked_inter_predictor_complex(
               xd, dst_buf[i], dst_stride[i], dst_buf1[i], dst_stride1[i],