Merge "Harmonize and fix coeff context computation" into nextgenv2
diff --git a/LICENSE b/LICENSE
index 1ce4434..fc340c3 100644
--- a/LICENSE
+++ b/LICENSE
@@ -1,31 +1,27 @@
-Copyright (c) 2010, The WebM Project authors. All rights reserved.
+Copyright (c) 2016, Alliance for Open Media. All rights reserved.
Redistribution and use in source and binary forms, with or without
-modification, are permitted provided that the following conditions are
-met:
+modification, are permitted provided that the following conditions
+are met:
- * Redistributions of source code must retain the above copyright
- notice, this list of conditions and the following disclaimer.
+1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
- * Redistributions in binary form must reproduce the above copyright
- notice, this list of conditions and the following disclaimer in
- the documentation and/or other materials provided with the
- distribution.
-
- * Neither the name of Google, nor the WebM Project, nor the names
- of its contributors may be used to endorse or promote products
- derived from this software without specific prior written
- permission.
+2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in
+ the documentation and/or other materials provided with the
+ distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
+COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
+INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
+BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
+ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+POSSIBILITY OF SUCH DAMAGE.
diff --git a/PATENTS b/PATENTS
index caedf60..be491f5 100644
--- a/PATENTS
+++ b/PATENTS
@@ -1,23 +1,108 @@
-Additional IP Rights Grant (Patents)
-------------------------------------
+Alliance for Open Media Patent License 1.0
-"These implementations" means the copyrightable works that implement the WebM
-codecs distributed by Google as part of the WebM Project.
+1. License Terms.
-Google hereby grants to you a perpetual, worldwide, non-exclusive, no-charge,
-royalty-free, irrevocable (except as stated in this section) patent license to
-make, have made, use, offer to sell, sell, import, transfer, and otherwise
-run, modify and propagate the contents of these implementations of WebM, where
-such license applies only to those patent claims, both currently owned by
-Google and acquired in the future, licensable by Google that are necessarily
-infringed by these implementations of WebM. This grant does not include claims
-that would be infringed only as a consequence of further modification of these
-implementations. If you or your agent or exclusive licensee institute or order
-or agree to the institution of patent litigation or any other patent
-enforcement activity against any entity (including a cross-claim or
-counterclaim in a lawsuit) alleging that any of these implementations of WebM
-or any code incorporated within any of these implementations of WebM
-constitute direct or contributory patent infringement, or inducement of
-patent infringement, then any patent rights granted to you under this License
-for these implementations of WebM shall terminate as of the date such
-litigation is filed.
+1.1. Patent License. Subject to the terms and conditions of this License, each
+ Licensor, on behalf of itself and successors in interest and assigns,
+ grants Licensee a non-sublicensable, perpetual, worldwide, non-exclusive,
+ no-charge, royalty-free, irrevocable (except as expressly stated in this
+ License) patent license to its Necessary Claims to make, use, sell, offer
+ for sale, import or distribute any Implementation.
+
+1.2. Conditions.
+
+1.2.1. Availability. As a condition to the grant of rights to Licensee to make,
+ sell, offer for sale, import or distribute an Implementation under
+ Section 1.1, Licensee must make its Necessary Claims available under
+ this License, and must reproduce this License with any Implementation
+ as follows:
+
+ a. For distribution in source code, by including this License in the
+ root directory of the source code with its Implementation.
+
+ b. For distribution in any other form (including binary, object form,
+ and/or hardware description code (e.g., HDL, RTL, Gate Level Netlist,
+ GDSII, etc.)), by including this License in the documentation, legal
+ notices, and/or other written materials provided with the
+ Implementation.
+
+1.2.2. Additional Conditions. This license is directly from Licensor to
+ Licensee. Licensee acknowledges as a condition of benefiting from it
+ that no rights from Licensor are received from suppliers, distributors,
+ or otherwise in connection with this License.
+
+1.3. Defensive Termination. If any Licensee, its Affiliates, or its agents
+ initiates patent litigation or files, maintains, or voluntarily
+ participates in a lawsuit against another entity or any person asserting
+ that any Implementation infringes Necessary Claims, any patent licenses
+ granted under this License directly to the Licensee are immediately
+ terminated as of the date of the initiation of action unless 1) that suit
+ was in response to a corresponding suit regarding an Implementation first
+ brought against an initiating entity, or 2) that suit was brought to
+ enforce the terms of this License (including intervention in a third-party
+ action by a Licensee).
+
+1.4. Disclaimers. The Reference Implementation and Specification are provided
+ "AS IS" and without warranty. The entire risk as to implementing or
+ otherwise using the Reference Implementation or Specification is assumed
+ by the implementer and user. Licensor expressly disclaims any warranties
+ (express, implied, or otherwise), including implied warranties of
+ merchantability, non-infringement, fitness for a particular purpose, or
+ title, related to the material. IN NO EVENT WILL LICENSOR BE LIABLE TO
+ ANY OTHER PARTY FOR LOST PROFITS OR ANY FORM OF INDIRECT, SPECIAL,
+ INCIDENTAL, OR CONSEQUENTIAL DAMAGES OF ANY CHARACTER FROM ANY CAUSES OF
+ ACTION OF ANY KIND WITH RESPECT TO THIS LICENSE, WHETHER BASED ON BREACH
+ OF CONTRACT, TORT (INCLUDING NEGLIGENCE), OR OTHERWISE, AND WHETHER OR
+ NOT THE OTHER PARTRY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+2. Definitions.
+
+2.1. Affiliate. Affiliate means an entity that directly or indirectly
+ Controls, is Controlled by, or is under common Control of that party.
+
+2.2. Control. Control means direct or indirect control of more than 50% of
+ the voting power to elect directors of that corporation, or for any other
+ entity, the power to direct management of such entity.
+
+2.3. Decoder. "Decoder" means any decoder that conforms fully with all
+ non-optional portions of the Specification.
+
+2.4. Encoder. "Encoder" means any encoder that produces a bitstream that can
+ be decoded by a Decoder only to the extent it produces such a bitstream.
+
+2.5. Final Deliverable. Final Deliverable means the final version of a
+ deliverable approved by the Alliance for Open Media as a Final
+ Deliverable.
+
+2.6. Implementation. "Implementation" means any implementation, including the
+ Reference Implementation, that is an Encoder and/or a Decoder. An
+ Implementation also includes components of an Implementation only to the
+ extent they are used as part of an Implementation.
+
+2.7. License. License means this license.
+
+2.8. Licensee. Licensee means any person or entity who exercises patent
+ rights granted under this License.
+
+2.9. Licensor. "Licensor" means (i) any Licensee that makes, sells, offers
+ for sale, imports or distributes any Implementation, or (ii) a person
+ or entity that has a licensing obligation to the Implementation as a
+ result of its membership and/or participation in the Alliance for Open
+ Media working group that developed the Specification.
+
+2.10. Necessary Claims. "Necessary Claims" means all claims of patents or
+ patent applications, (a) that currently or at any time in the future,
+ are owned or controlled by the Licensor, and (b) (i) would be an
+ Essential Claim as defined by the W3C Policy as of February 5, 2004
+ (https://www.w3.org/Consortium/Patent-Policy-20040205/#def-essential)
+ as if the Specification was a W3C Recommendation; or (ii) are infringed
+ by the Reference Implementation.
+
+2.11. Reference Implementation. Reference Implementation means an Encoder
+ and/or Decoder released by the Alliance for Open Media as a Final
+ Deliverable.
+
+2.12. Specification. Specification means the specification designated by
+ the Alliance for Open Media as a Final Deliverable for which this
+ License was issued.
+
diff --git a/aom/internal/aom_codec_internal.h b/aom/internal/aom_codec_internal.h
index fc758b8..4399279 100644
--- a/aom/internal/aom_codec_internal.h
+++ b/aom/internal/aom_codec_internal.h
@@ -43,6 +43,7 @@
*/
#ifndef AOM_INTERNAL_AOM_CODEC_INTERNAL_H_
#define AOM_INTERNAL_AOM_CODEC_INTERNAL_H_
+#include "./aom_config.h"
#include "../aom_decoder.h"
#include "../aom_encoder.h"
#include <stdarg.h>
@@ -439,6 +440,24 @@
aom_codec_err_t error, const char *fmt,
...) CLANG_ANALYZER_NORETURN;
+#if CONFIG_DEBUG
+#define AOM_CHECK_MEM_ERROR(error_info, lval, expr) \
+ do { \
+ lval = (expr); \
+ if (!lval) \
+ aom_internal_error(error_info, AOM_CODEC_MEM_ERROR, \
+ "Failed to allocate " #lval " at %s:%d", __FILE__, \
+ __LINE__); \
+ } while (0)
+#else
+#define AOM_CHECK_MEM_ERROR(error_info, lval, expr) \
+ do { \
+ lval = (expr); \
+ if (!lval) \
+ aom_internal_error(error_info, AOM_CODEC_MEM_ERROR, \
+ "Failed to allocate " #lval); \
+ } while (0)
+#endif
#ifdef __cplusplus
} // extern "C"
#endif
diff --git a/aom_dsp/x86/sad_avx2.c b/aom_dsp/x86/sad_avx2.c
index bfa734d..6ce61fd 100644
--- a/aom_dsp/x86/sad_avx2.c
+++ b/aom_dsp/x86/sad_avx2.c
@@ -81,8 +81,10 @@
FSAD32_H(32); \
FSAD32_H(16);
-FSAD64;
-FSAD32;
+/* clang-format off */
+FSAD64
+FSAD32
+/* clang-format on */
#undef FSAD64
#undef FSAD32
@@ -170,8 +172,10 @@
FSADAVG32_H(32); \
FSADAVG32_H(16);
-FSADAVG64;
-FSADAVG32;
+/* clang-format off */
+FSADAVG64
+FSADAVG32
+/* clang-format on */
#undef FSADAVG64
#undef FSADAVG32
diff --git a/aom_mem/aom_mem.c b/aom_mem/aom_mem.c
index 43f8237..144085f 100644
--- a/aom_mem/aom_mem.c
+++ b/aom_mem/aom_mem.c
@@ -18,35 +18,47 @@
#include "include/aom_mem_intrnl.h"
#include "aom/aom_integer.h"
+static size_t GetAlignedMallocSize(size_t size, size_t align) {
+ return size + align - 1 + ADDRESS_STORAGE_SIZE;
+}
+
+static size_t *GetMallocAddressLocation(void *const mem) {
+ return ((size_t *)mem) - 1;
+}
+
+static void SetActualMallocAddress(void *const mem,
+ const void *const malloc_addr) {
+ size_t *const malloc_addr_location = GetMallocAddressLocation(mem);
+ *malloc_addr_location = (size_t)malloc_addr;
+}
+
+static void *GetActualMallocAddress(void *const mem) {
+ const size_t *const malloc_addr_location = GetMallocAddressLocation(mem);
+ return (void *)(*malloc_addr_location);
+}
+
void *aom_memalign(size_t align, size_t size) {
- void *addr, *x = NULL;
-
- addr = malloc(size + align - 1 + ADDRESS_STORAGE_SIZE);
-
+ void *x = NULL;
+ const size_t aligned_size = GetAlignedMallocSize(size, align);
+ void *const addr = malloc(aligned_size);
if (addr) {
x = align_addr((unsigned char *)addr + ADDRESS_STORAGE_SIZE, (int)align);
- /* save the actual malloc address */
- ((size_t *)x)[-1] = (size_t)addr;
+ SetActualMallocAddress(x, addr);
}
-
return x;
}
void *aom_malloc(size_t size) { return aom_memalign(DEFAULT_ALIGNMENT, size); }
void *aom_calloc(size_t num, size_t size) {
- void *x;
-
- x = aom_memalign(DEFAULT_ALIGNMENT, num * size);
-
- if (x) memset(x, 0, num * size);
-
+ const size_t total_size = num * size;
+ void *const x = aom_malloc(total_size);
+ if (x) memset(x, 0, total_size);
return x;
}
void *aom_realloc(void *memblk, size_t size) {
- void *addr, *new_addr = NULL;
- int align = DEFAULT_ALIGNMENT;
+ void *new_addr = NULL;
/*
The realloc() function changes the size of the object pointed to by
@@ -61,19 +73,14 @@
else if (!size)
aom_free(memblk);
else {
- addr = (void *)(((size_t *)memblk)[-1]);
+ void *addr = GetActualMallocAddress(memblk);
+ const size_t aligned_size = GetAlignedMallocSize(size, DEFAULT_ALIGNMENT);
memblk = NULL;
-
- new_addr = realloc(addr, size + align + ADDRESS_STORAGE_SIZE);
-
- if (new_addr) {
- addr = new_addr;
- new_addr =
- (void *)(((size_t)((unsigned char *)new_addr + ADDRESS_STORAGE_SIZE) +
- (align - 1)) &
- (size_t)-align);
- /* save the actual malloc address */
- ((size_t *)new_addr)[-1] = (size_t)addr;
+ addr = realloc(addr, aligned_size);
+ if (addr) {
+ new_addr = align_addr((unsigned char *)addr + ADDRESS_STORAGE_SIZE,
+ DEFAULT_ALIGNMENT);
+ SetActualMallocAddress(new_addr, addr);
}
}
@@ -82,7 +89,7 @@
void aom_free(void *memblk) {
if (memblk) {
- void *addr = (void *)(((size_t *)memblk)[-1]);
+ void *addr = GetActualMallocAddress(memblk);
free(addr);
}
}
diff --git a/av1/common/common.h b/av1/common/common.h
index 5b8100b..551055a 100644
--- a/av1/common/common.h
+++ b/av1/common/common.h
@@ -16,7 +16,6 @@
#include <assert.h>
-#include "./aom_config.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_mem/aom_mem.h"
#include "aom/aom_integer.h"
@@ -49,24 +48,8 @@
return num_values > 0 ? get_msb(num_values) + 1 : 0;
}
-#if CONFIG_DEBUG
-#define CHECK_MEM_ERROR(cm, lval, expr) \
- do { \
- lval = (expr); \
- if (!lval) \
- aom_internal_error(&cm->error, AOM_CODEC_MEM_ERROR, \
- "Failed to allocate " #lval " at %s:%d", __FILE__, \
- __LINE__); \
- } while (0)
-#else
-#define CHECK_MEM_ERROR(cm, lval, expr) \
- do { \
- lval = (expr); \
- if (!lval) \
- aom_internal_error(&cm->error, AOM_CODEC_MEM_ERROR, \
- "Failed to allocate " #lval); \
- } while (0)
-#endif
+#define CHECK_MEM_ERROR(cm, lval, expr) \
+ AOM_CHECK_MEM_ERROR(&cm->error, lval, expr)
// TODO(yaowu: validate the usage of these codes or develop new ones.)
#define AV1_SYNC_CODE_0 0x49
#define AV1_SYNC_CODE_1 0x83
diff --git a/av1/common/entropymode.c b/av1/common/entropymode.c
index b2cb6ff..e1593e3 100644
--- a/av1/common/entropymode.c
+++ b/av1/common/entropymode.c
@@ -857,6 +857,12 @@
},
};
+#if CONFIG_EXT_TX && CONFIG_RECT_TX && CONFIG_VAR_TX
+// the probability of (0) using recursive square tx partition vs.
+// (1) biggest rect tx for 4X8-8X4/8X16-16X8/16X32-32X16 blocks
+static const aom_prob default_rect_tx_prob[TX_SIZES - 1] = { 192, 192, 192 };
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX && CONFIG_VAR_TX
+
int av1_get_palette_color_context(const uint8_t *color_map, int cols, int r,
int c, int n, int *color_order) {
int i, j, max, max_idx, temp;
@@ -1298,6 +1304,9 @@
av1_copy(fc->tx_size_probs, default_tx_size_prob);
#if CONFIG_VAR_TX
av1_copy(fc->txfm_partition_prob, default_txfm_partition_probs);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ av1_copy(fc->rect_tx_prob, default_rect_tx_prob);
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
#endif
av1_copy(fc->skip_probs, default_skip_probs);
#if CONFIG_REF_MV
@@ -1451,6 +1460,15 @@
}
#endif // CONFIG_EXT_INTER
+#if CONFIG_VAR_TX && CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (cm->tx_mode == TX_MODE_SELECT) {
+ for (i = 0; i < TX_SIZES - 1; ++i) {
+ fc->rect_tx_prob[i] =
+ av1_mode_mv_merge_probs(pre_fc->rect_tx_prob[i], counts->rect_tx[i]);
+ }
+ }
+#endif // CONFIG_VAR_TX && CONFIG_EXT_TX && CONFIG_RECT_TX
+
for (i = 0; i < BLOCK_SIZE_GROUPS; i++)
aom_tree_merge_probs(av1_intra_mode_tree, pre_fc->y_mode_prob[i],
counts->y_mode[i], fc->y_mode_prob[i]);
@@ -1478,10 +1496,11 @@
}
#if CONFIG_VAR_TX
- if (cm->tx_mode == TX_MODE_SELECT)
+ if (cm->tx_mode == TX_MODE_SELECT) {
for (i = 0; i < TXFM_PARTITION_CONTEXTS; ++i)
fc->txfm_partition_prob[i] = av1_mode_mv_merge_probs(
pre_fc->txfm_partition_prob[i], counts->txfm_partition[i]);
+ }
#endif
for (i = 0; i < SKIP_CONTEXTS; ++i)
diff --git a/av1/common/entropymode.h b/av1/common/entropymode.h
index 956bc99..7968484 100644
--- a/av1/common/entropymode.h
+++ b/av1/common/entropymode.h
@@ -96,6 +96,10 @@
aom_prob tx_size_probs[TX_SIZES - 1][TX_SIZE_CONTEXTS][TX_SIZES - 1];
#if CONFIG_VAR_TX
aom_prob txfm_partition_prob[TXFM_PARTITION_CONTEXTS];
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ // TODO(yuec) make this flag harmonize with the original syntax
+ aom_prob rect_tx_prob[TX_SIZES - 1];
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
#endif
aom_prob skip_probs[SKIP_CONTEXTS];
#if CONFIG_REF_MV
@@ -179,6 +183,9 @@
unsigned int tx_size[TX_SIZES - 1][TX_SIZE_CONTEXTS][TX_SIZES];
#if CONFIG_VAR_TX
unsigned int txfm_partition[TXFM_PARTITION_CONTEXTS][2];
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ unsigned int rect_tx[TX_SIZES - 1][2];
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
#endif
unsigned int skip[SKIP_CONTEXTS][2];
#if CONFIG_REF_MV
diff --git a/av1/common/loopfilter.c b/av1/common/loopfilter.c
index 7086be8..2147bb8 100644
--- a/av1/common/loopfilter.c
+++ b/av1/common/loopfilter.c
@@ -1246,12 +1246,18 @@
tx_size_mask = 0;
#if CONFIG_VAR_TX
- if (is_inter_block(mbmi) && !mbmi->skip)
- tx_size =
- (plane->plane_type == PLANE_TYPE_UV)
- ? uv_txsize_lookup[sb_type][mbmi->inter_tx_size
- [blk_row][blk_col]][ss_x][ss_y]
- : mbmi->inter_tx_size[blk_row][blk_col];
+ if (is_inter_block(mbmi) && !mbmi->skip) {
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ TX_SIZE mb_tx_size = is_rect_tx(mbmi->tx_size)
+ ? mbmi->tx_size
+ : mbmi->inter_tx_size[blk_row][blk_col];
+#else
+ TX_SIZE mb_tx_size = mbmi->inter_tx_size[blk_row][blk_col];
+#endif
+ tx_size = (plane->plane_type == PLANE_TYPE_UV)
+ ? uv_txsize_lookup[sb_type][mb_tx_size][ss_x][ss_y]
+ : mb_tx_size;
+ }
#if CONFIG_EXT_TX && CONFIG_RECT_TX
tx_size_r =
diff --git a/av1/common/odintrin.h b/av1/common/odintrin.h
index 4c5b9ae..5324cff 100644
--- a/av1/common/odintrin.h
+++ b/av1/common/odintrin.h
@@ -16,6 +16,10 @@
#include "aom_dsp/aom_dsp_common.h"
#include "aom_ports/bitops.h"
+#ifdef __cplusplus
+extern "C" {
+#endif
+
/*Smallest blocks are 4x4*/
#define OD_LOG_BSIZE0 (2)
/*There are 5 block sizes total (4x4, 8x8, 16x16, 32x32 and 64x64).*/
@@ -55,4 +59,8 @@
We define a special version of the macro to use when x can be zero.*/
#define OD_ILOG(x) ((x) ? OD_ILOG_NZ(x) : 0)
+#ifdef __cplusplus
+} // extern "C"
#endif
+
+#endif // AV1_COMMON_ODINTRIN_H_
diff --git a/av1/common/warped_motion.c b/av1/common/warped_motion.c
index 2ff1fce..146a14d 100644
--- a/av1/common/warped_motion.c
+++ b/av1/common/warped_motion.c
@@ -477,7 +477,7 @@
uint16_t *dst = CONVERT_TO_SHORTPTR(dst8);
uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
int gm_err = 0, no_gm_err = 0;
- int gm_sumerr = 0, no_gm_sumerr = 0;
+ int64_t gm_sumerr = 0, no_gm_sumerr = 0;
for (i = p_row; i < p_row + p_height; ++i) {
for (j = p_col; j < p_col + p_width; ++j) {
int in[2], out[2];
@@ -492,8 +492,8 @@
stride, bd);
no_gm_err = dst[(j - p_col) + (i - p_row) * p_stride] -
ref[(j - p_col) + (i - p_row) * stride];
- gm_sumerr += gm_err * gm_err;
- no_gm_sumerr += no_gm_err * no_gm_err;
+ gm_sumerr += (int64_t)gm_err * gm_err;
+ no_gm_sumerr += (int64_t)no_gm_err * no_gm_err;
}
}
return (double)gm_sumerr / no_gm_sumerr;
diff --git a/av1/decoder/decodeframe.c b/av1/decoder/decodeframe.c
index 2daf4bf..8d3202c 100644
--- a/av1/decoder/decodeframe.c
+++ b/av1/decoder/decodeframe.c
@@ -1308,10 +1308,9 @@
const int step = num_4x4_blocks_txsize_lookup[max_tx_size];
int block = 0;
#if CONFIG_EXT_TX && CONFIG_RECT_TX
- const TX_SIZE tx_size =
- plane ? get_uv_tx_size(mbmi, pd) : mbmi->tx_size;
-
- if (tx_size >= TX_SIZES) { // rect txsize is used
+ if (is_rect_tx(mbmi->tx_size)) {
+ const TX_SIZE tx_size =
+ plane ? get_uv_tx_size(mbmi, pd) : mbmi->tx_size;
const int stepr = num_4x4_blocks_high_txsize_lookup[tx_size];
const int stepc = num_4x4_blocks_wide_txsize_lookup[tx_size];
const int max_blocks_wide =
@@ -3491,6 +3490,12 @@
#if CONFIG_VAR_TX
for (k = 0; k < TXFM_PARTITION_CONTEXTS; ++k)
av1_diff_update_prob(&r, &fc->txfm_partition_prob[k]);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (cm->tx_mode == TX_MODE_SELECT) {
+ for (i = 1; i < TX_SIZES - 1; ++i)
+ av1_diff_update_prob(&r, &fc->rect_tx_prob[i]);
+ }
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
#endif
for (k = 0; k < SKIP_CONTEXTS; ++k)
diff --git a/av1/decoder/decodemv.c b/av1/decoder/decodemv.c
index 316d317..535c5b5 100644
--- a/av1/decoder/decodemv.c
+++ b/av1/decoder/decodemv.c
@@ -1632,13 +1632,34 @@
const int width = num_4x4_blocks_wide_lookup[bsize];
const int height = num_4x4_blocks_high_lookup[bsize];
int idx, idy;
- for (idy = 0; idy < height; idy += bs)
- for (idx = 0; idx < width; idx += bs)
- read_tx_size_vartx(cm, xd, mbmi, xd->counts, max_tx_size, idy, idx,
- r);
+ int tx_size_cat = inter_tx_size_cat_lookup[bsize];
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ int is_rect_tx_allowed = inter_block && is_rect_tx_allowed_bsize(bsize);
+ int use_rect_tx = 0;
+
+ if (is_rect_tx_allowed) {
+ use_rect_tx = aom_read(r, cm->fc->rect_tx_prob[tx_size_cat]);
+ if (xd->counts) {
+ ++xd->counts->rect_tx[tx_size_cat][use_rect_tx];
+ }
+ }
+
+ if (use_rect_tx) {
+ mbmi->tx_size = max_txsize_rect_lookup[bsize];
+ set_txfm_ctxs(mbmi->tx_size, xd->n8_w, xd->n8_h, xd);
+ } else {
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
+ for (idy = 0; idy < height; idy += bs)
+ for (idx = 0; idx < width; idx += bs)
+ read_tx_size_vartx(cm, xd, mbmi, xd->counts, max_tx_size, idy, idx,
+ r);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ }
+#endif
if (xd->counts) {
const int ctx = get_tx_size_context(xd);
- ++xd->counts->tx_size[max_tx_size - TX_8X8][ctx][mbmi->tx_size];
+ ++xd->counts
+ ->tx_size[tx_size_cat][ctx][txsize_sqr_up_map[mbmi->tx_size]];
}
} else {
if (inter_block)
diff --git a/av1/encoder/bitstream.c b/av1/encoder/bitstream.c
index 3bd5adc..7c33aea 100644
--- a/av1/encoder/bitstream.c
+++ b/av1/encoder/bitstream.c
@@ -1121,9 +1121,25 @@
const int width = num_4x4_blocks_wide_lookup[bsize];
const int height = num_4x4_blocks_high_lookup[bsize];
int idx, idy;
- for (idy = 0; idy < height; idy += bs)
- for (idx = 0; idx < width; idx += bs)
- write_tx_size_vartx(cm, xd, mbmi, max_tx_size, idy, idx, w);
+
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (is_rect_tx_allowed(mbmi)) {
+ int tx_size_cat = inter_tx_size_cat_lookup[bsize];
+
+ aom_write(w, is_rect_tx(mbmi->tx_size),
+ cm->fc->rect_tx_prob[tx_size_cat]);
+ }
+
+ if (is_rect_tx(mbmi->tx_size)) {
+ set_txfm_ctxs(mbmi->tx_size, xd->n8_w, xd->n8_h, xd);
+ } else {
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
+ for (idy = 0; idy < height; idy += bs)
+ for (idx = 0; idx < width; idx += bs)
+ write_tx_size_vartx(cm, xd, mbmi, max_tx_size, idy, idx, w);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ }
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
} else {
set_txfm_ctxs(mbmi->tx_size, xd->n8_w, xd->n8_h, xd);
write_selected_tx_size(cm, xd, w);
@@ -1650,7 +1666,7 @@
TX_SIZE tx_size =
plane ? get_uv_tx_size(mbmi, &xd->plane[plane]) : mbmi->tx_size;
- if (is_inter_block(mbmi) && tx_size < TX_SIZES) {
+ if (is_inter_block(mbmi) && !is_rect_tx(tx_size)) {
#else
if (is_inter_block(mbmi)) {
#endif
@@ -3271,6 +3287,13 @@
#if CONFIG_VAR_TX
update_txfm_partition_probs(cm, header_bc, counts);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (cm->tx_mode == TX_MODE_SELECT) {
+ for (i = 1; i < TX_SIZES - 1; ++i)
+ av1_cond_prob_diff_update(header_bc, &fc->rect_tx_prob[i],
+ counts->rect_tx[i]);
+ }
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
#endif
update_skip_probs(cm, header_bc, counts);
diff --git a/av1/encoder/encodeframe.c b/av1/encoder/encodeframe.c
index f172fbb..200b02c 100644
--- a/av1/encoder/encodeframe.c
+++ b/av1/encoder/encodeframe.c
@@ -5031,7 +5031,7 @@
av1_encode_sb(x, AOMMAX(bsize, BLOCK_8X8));
#if CONFIG_VAR_TX
#if CONFIG_EXT_TX && CONFIG_RECT_TX
- if (mbmi->tx_size >= TX_SIZES)
+ if (is_rect_tx(mbmi->tx_size))
av1_tokenize_sb(cpi, td, t, !output_enabled, AOMMAX(bsize, BLOCK_8X8));
else
#endif
@@ -5054,8 +5054,17 @@
assert(IMPLIES(is_rect_tx(mbmi->tx_size), is_rect_tx_allowed(mbmi)));
#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
#if CONFIG_VAR_TX
- if (is_inter)
- tx_partition_count_update(cm, xd, bsize, mi_row, mi_col, td->counts);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (is_rect_tx_allowed(mbmi)) {
+ td->counts->rect_tx[tx_size_cat][is_rect_tx(mbmi->tx_size)]++;
+ }
+ if (!is_rect_tx_allowed(mbmi) || !is_rect_tx(mbmi->tx_size)) {
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (is_inter)
+ tx_partition_count_update(cm, xd, bsize, mi_row, mi_col, td->counts);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ }
+#endif
#endif
++td->counts->tx_size[tx_size_cat][tx_size_ctx][coded_tx_size];
} else {
@@ -5118,6 +5127,11 @@
is_inter_block(mbmi) && !(mbmi->skip || seg_skip)) {
if (!output_enabled)
tx_partition_set_contexts(cm, xd, bsize, mi_row, mi_col);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (is_rect_tx(mbmi->tx_size)) {
+ set_txfm_ctxs(mbmi->tx_size, xd->n8_w, xd->n8_h, xd);
+ }
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
} else {
TX_SIZE tx_size;
// The new intra coding scheme requires no change of transform size
diff --git a/av1/encoder/encodemb.c b/av1/encoder/encodemb.c
index 5a21ebd..718bddb 100644
--- a/av1/encoder/encodemb.c
+++ b/av1/encoder/encodemb.c
@@ -791,7 +791,7 @@
#endif
#if CONFIG_VAR_TX
- // Assert not magic number (uninitialised).
+ // Assert not magic number (uninitialized).
assert(x->blk_skip[plane][(blk_row << bwl) + blk_col] != 234);
if (x->blk_skip[plane][(blk_row << bwl) + blk_col] == 0) {
@@ -976,9 +976,6 @@
int idx, idy;
int block = 0;
int step = num_4x4_blocks_txsize_lookup[max_tx_size];
-#if CONFIG_EXT_TX && CONFIG_RECT_TX
- const TX_SIZE tx_size = plane ? get_uv_tx_size(mbmi, pd) : mbmi->tx_size;
-#endif
av1_get_entropy_contexts(bsize, TX_4X4, pd, ctx.ta[plane], ctx.tl[plane]);
#else
const struct macroblockd_plane *const pd = &xd->plane[plane];
@@ -991,7 +988,7 @@
#if CONFIG_VAR_TX
#if CONFIG_EXT_TX && CONFIG_RECT_TX
- if (tx_size >= TX_SIZES) {
+ if (is_rect_tx(mbmi->tx_size)) {
av1_foreach_transformed_block_in_plane(xd, bsize, plane, encode_block,
&arg);
} else {
diff --git a/av1/encoder/encoder.c b/av1/encoder/encoder.c
index c2e8275..65e1f3c 100644
--- a/av1/encoder/encoder.c
+++ b/av1/encoder/encoder.c
@@ -2526,9 +2526,8 @@
SNPRINT2(results, "\t%7.3f", consistency);
SNPRINT2(results, "\t%7.3f", cpi->worst_consistency);
}
-
- fprintf(f, "%s\t Time Rc-Err Abs Err\n", headings);
- fprintf(f, "%s\t%8.0f %7.2f %7.2f\n", results, total_encode_time,
+ fprintf(f, "%s\t Time\tRcErr\tAbsErr\n", headings);
+ fprintf(f, "%s\t%8.0f\t%7.2f\t%7.2f\n", results, total_encode_time,
rate_err, fabs(rate_err));
}
diff --git a/av1/encoder/rdopt.c b/av1/encoder/rdopt.c
index 4d28bf0..f27a100 100644
--- a/av1/encoder/rdopt.c
+++ b/av1/encoder/rdopt.c
@@ -2766,6 +2766,61 @@
return best_rd;
}
+// Return value 0: early termination triggered, no valid rd cost available;
+// 1: rd cost values are valid.
+static int super_block_uvrd(const AV1_COMP *cpi, MACROBLOCK *x, int *rate,
+ int64_t *distortion, int *skippable, int64_t *sse,
+ BLOCK_SIZE bsize, int64_t ref_best_rd) {
+ MACROBLOCKD *const xd = &x->e_mbd;
+ MB_MODE_INFO *const mbmi = &xd->mi[0]->mbmi;
+ const TX_SIZE uv_tx_size = get_uv_tx_size(mbmi, &xd->plane[1]);
+ int plane;
+ int pnrate = 0, pnskip = 1;
+ int64_t pndist = 0, pnsse = 0;
+ int is_cost_valid = 1;
+
+ if (ref_best_rd < 0) is_cost_valid = 0;
+
+ if (is_inter_block(mbmi) && is_cost_valid) {
+ int plane;
+ for (plane = 1; plane < MAX_MB_PLANE; ++plane)
+ av1_subtract_plane(x, bsize, plane);
+ }
+
+ *rate = 0;
+ *distortion = 0;
+ *sse = 0;
+ *skippable = 1;
+
+ for (plane = 1; plane < MAX_MB_PLANE; ++plane) {
+ txfm_rd_in_plane(x, cpi, &pnrate, &pndist, &pnskip, &pnsse, ref_best_rd,
+ plane, bsize, uv_tx_size, cpi->sf.use_fast_coef_costing);
+ if (pnrate == INT_MAX) {
+ is_cost_valid = 0;
+ break;
+ }
+ *rate += pnrate;
+ *distortion += pndist;
+ *sse += pnsse;
+ *skippable &= pnskip;
+ if (RDCOST(x->rdmult, x->rddiv, *rate, *distortion) > ref_best_rd &&
+ RDCOST(x->rdmult, x->rddiv, 0, *sse) > ref_best_rd) {
+ is_cost_valid = 0;
+ break;
+ }
+ }
+
+ if (!is_cost_valid) {
+ // reset cost value
+ *rate = INT_MAX;
+ *distortion = INT64_MAX;
+ *sse = INT64_MAX;
+ *skippable = 0;
+ }
+
+ return is_cost_valid;
+}
+
#if CONFIG_VAR_TX
void av1_tx_block_rd_b(const AV1_COMP *cpi, MACROBLOCK *x, TX_SIZE tx_size,
int blk_row, int blk_col, int plane, int block,
@@ -3148,6 +3203,61 @@
mbmi->tx_type = tx_type;
inter_block_yrd(cpi, x, rate, dist, skippable, sse, bsize, ref_best_rd);
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (is_rect_tx_allowed(mbmi)) {
+ int rate_rect_tx, skippable_rect_tx = 0;
+ int64_t dist_rect_tx, sse_rect_tx, rd, rd_rect_tx;
+ int tx_size_cat = inter_tx_size_cat_lookup[bsize];
+ TX_SIZE tx_size = max_txsize_rect_lookup[bsize];
+ TX_SIZE var_tx_size = mbmi->tx_size;
+
+ txfm_rd_in_plane(x, cpi, &rate_rect_tx, &dist_rect_tx, &skippable_rect_tx,
+ &sse_rect_tx, ref_best_rd, 0, bsize, tx_size,
+ cpi->sf.use_fast_coef_costing);
+
+ if (*rate != INT_MAX) {
+ *rate += av1_cost_bit(cm->fc->rect_tx_prob[tx_size_cat], 0);
+ if (*skippable) {
+ rd = RDCOST(x->rdmult, x->rddiv, s1, *sse);
+ } else {
+ rd = RDCOST(x->rdmult, x->rddiv, *rate + s0, *dist);
+ if (is_inter && !xd->lossless[xd->mi[0]->mbmi.segment_id] &&
+ !(*skippable))
+ rd = AOMMIN(rd, RDCOST(x->rdmult, x->rddiv, s1, *sse));
+ }
+ } else {
+ rd = INT64_MAX;
+ }
+
+ if (rate_rect_tx != INT_MAX) {
+ rate_rect_tx += av1_cost_bit(cm->fc->rect_tx_prob[tx_size_cat], 1);
+ if (skippable_rect_tx) {
+ rd_rect_tx = RDCOST(x->rdmult, x->rddiv, s1, sse_rect_tx);
+ } else {
+ rd_rect_tx =
+ RDCOST(x->rdmult, x->rddiv, rate_rect_tx + s0, dist_rect_tx);
+ if (is_inter && !xd->lossless[xd->mi[0]->mbmi.segment_id] &&
+ !(skippable_rect_tx))
+ rd_rect_tx =
+ AOMMIN(rd_rect_tx, RDCOST(x->rdmult, x->rddiv, s1, sse_rect_tx));
+ }
+ } else {
+ rd_rect_tx = INT64_MAX;
+ }
+
+ if (rd_rect_tx < rd) {
+ *rate = rate_rect_tx;
+ *dist = dist_rect_tx;
+ *sse = sse_rect_tx;
+ *skippable = skippable_rect_tx;
+ if (!xd->lossless[mbmi->segment_id]) x->blk_skip[0][0] = *skippable;
+ mbmi->tx_size = tx_size;
+ mbmi->inter_tx_size[0][0] = mbmi->tx_size;
+ } else {
+ mbmi->tx_size = var_tx_size;
+ }
+ }
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
if (*rate == INT_MAX) return INT64_MAX;
@@ -3361,17 +3471,24 @@
if (ref_best_rd < 0) is_cost_valid = 0;
+ *rate = 0;
+ *distortion = 0;
+ *sse = 0;
+ *skippable = 1;
+
+#if CONFIG_EXT_TX && CONFIG_RECT_TX
+ if (is_rect_tx(mbmi->tx_size)) {
+ return super_block_uvrd(cpi, x, rate, distortion, skippable, sse, bsize,
+ ref_best_rd);
+ }
+#endif // CONFIG_EXT_TX && CONFIG_RECT_TX
+
if (is_inter_block(mbmi) && is_cost_valid) {
int plane;
for (plane = 1; plane < MAX_MB_PLANE; ++plane)
av1_subtract_plane(x, bsize, plane);
}
- *rate = 0;
- *distortion = 0;
- *sse = 0;
- *skippable = 1;
-
for (plane = 1; plane < MAX_MB_PLANE; ++plane) {
const struct macroblockd_plane *const pd = &xd->plane[plane];
const BLOCK_SIZE plane_bsize = get_plane_block_size(bsize, pd);
@@ -3429,61 +3546,6 @@
}
#endif // CONFIG_VAR_TX
-// Return value 0: early termination triggered, no valid rd cost available;
-// 1: rd cost values are valid.
-static int super_block_uvrd(const AV1_COMP *cpi, MACROBLOCK *x, int *rate,
- int64_t *distortion, int *skippable, int64_t *sse,
- BLOCK_SIZE bsize, int64_t ref_best_rd) {
- MACROBLOCKD *const xd = &x->e_mbd;
- MB_MODE_INFO *const mbmi = &xd->mi[0]->mbmi;
- const TX_SIZE uv_tx_size = get_uv_tx_size(mbmi, &xd->plane[1]);
- int plane;
- int pnrate = 0, pnskip = 1;
- int64_t pndist = 0, pnsse = 0;
- int is_cost_valid = 1;
-
- if (ref_best_rd < 0) is_cost_valid = 0;
-
- if (is_inter_block(mbmi) && is_cost_valid) {
- int plane;
- for (plane = 1; plane < MAX_MB_PLANE; ++plane)
- av1_subtract_plane(x, bsize, plane);
- }
-
- *rate = 0;
- *distortion = 0;
- *sse = 0;
- *skippable = 1;
-
- for (plane = 1; plane < MAX_MB_PLANE; ++plane) {
- txfm_rd_in_plane(x, cpi, &pnrate, &pndist, &pnskip, &pnsse, ref_best_rd,
- plane, bsize, uv_tx_size, cpi->sf.use_fast_coef_costing);
- if (pnrate == INT_MAX) {
- is_cost_valid = 0;
- break;
- }
- *rate += pnrate;
- *distortion += pndist;
- *sse += pnsse;
- *skippable &= pnskip;
- if (RDCOST(x->rdmult, x->rddiv, *rate, *distortion) > ref_best_rd &&
- RDCOST(x->rdmult, x->rddiv, 0, *sse) > ref_best_rd) {
- is_cost_valid = 0;
- break;
- }
- }
-
- if (!is_cost_valid) {
- // reset cost value
- *rate = INT_MAX;
- *distortion = INT64_MAX;
- *sse = INT64_MAX;
- *skippable = 0;
- }
-
- return is_cost_valid;
-}
-
static void rd_pick_palette_intra_sbuv(
AV1_COMP *cpi, MACROBLOCK *x, int dc_mode_cost,
PALETTE_MODE_INFO *palette_mode_info, uint8_t *best_palette_color_map,
diff --git a/third_party/googletest/README.libvpx b/third_party/googletest/README.libaom
similarity index 100%
rename from third_party/googletest/README.libvpx
rename to third_party/googletest/README.libaom
diff --git a/third_party/libwebm/README.libvpx b/third_party/libwebm/README.libaom
similarity index 100%
rename from third_party/libwebm/README.libvpx
rename to third_party/libwebm/README.libaom
diff --git a/third_party/libyuv/README.libvpx b/third_party/libyuv/README.libaom
similarity index 100%
rename from third_party/libyuv/README.libvpx
rename to third_party/libyuv/README.libaom
diff --git a/third_party/x86inc/README.libvpx b/third_party/x86inc/README.libaom
similarity index 100%
rename from third_party/x86inc/README.libvpx
rename to third_party/x86inc/README.libaom
