Merge "Test vector mismatch fix"
diff --git a/build/make/gen_msvs_proj.sh b/build/make/gen_msvs_proj.sh
index c2ef44a..fb948fd 100755
--- a/build/make/gen_msvs_proj.sh
+++ b/build/make/gen_msvs_proj.sh
@@ -447,6 +447,8 @@
obj_int_extract)
tag Tool \
Name="VCCLCompilerTool" \
+ Optimization="2" \
+ FavorSizeorSpeed="1" \
AdditionalIncludeDirectories="$incs" \
PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE" \
RuntimeLibrary="$release_runtime" \
@@ -462,6 +464,8 @@
tag Tool \
Name="VCCLCompilerTool" \
+ Optimization="2" \
+ FavorSizeorSpeed="1" \
AdditionalIncludeDirectories="$incs" \
PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
RuntimeLibrary="$release_runtime" \
@@ -476,6 +480,8 @@
tag Tool \
Name="VCCLCompilerTool" \
AdditionalIncludeDirectories="$incs" \
+ Optimization="2" \
+ FavorSizeorSpeed="1" \
PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \
RuntimeLibrary="$release_runtime" \
UsePrecompiledHeader="0" \
diff --git a/libmkv/EbmlIDs.h b/libmkv/EbmlIDs.h
index 4297470..3418e36 100644
--- a/libmkv/EbmlIDs.h
+++ b/libmkv/EbmlIDs.h
@@ -120,7 +120,7 @@
//video
Video = 0xE0,
FlagInterlaced = 0x9A,
-// StereoMode = 0x53B8,
+ StereoMode = 0x53B8,
PixelWidth = 0xB0,
PixelHeight = 0xBA,
PixelCropBottom = 0x54AA,
diff --git a/vp8/common/alloccommon.c b/vp8/common/alloccommon.c
index edef360..b0c7363 100644
--- a/vp8/common/alloccommon.c
+++ b/vp8/common/alloccommon.c
@@ -129,32 +129,32 @@
{
case 0:
cm->no_lpf = 0;
- cm->simpler_lpf = 0;
+ cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 0;
cm->full_pixel = 0;
break;
case 1:
cm->no_lpf = 0;
- cm->simpler_lpf = 1;
+ cm->filter_type = SIMPLE_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 0;
break;
case 2:
cm->no_lpf = 1;
- cm->simpler_lpf = 0;
+ cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 0;
break;
case 3:
cm->no_lpf = 1;
- cm->simpler_lpf = 1;
+ cm->filter_type = SIMPLE_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 1;
break;
default:
/*4,5,6,7 are reserved for future use*/
cm->no_lpf = 0;
- cm->simpler_lpf = 0;
+ cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 0;
cm->full_pixel = 0;
break;
@@ -169,7 +169,7 @@
oci->mb_no_coeff_skip = 1;
oci->no_lpf = 0;
- oci->simpler_lpf = 0;
+ oci->filter_type = NORMAL_LOOPFILTER;
oci->use_bilinear_mc_filter = 0;
oci->full_pixel = 0;
oci->multi_token_partition = ONE_PARTITION;
diff --git a/vp8/common/arm/loopfilter_arm.c b/vp8/common/arm/loopfilter_arm.c
index 3532a03..6d1caa4 100644
--- a/vp8/common/arm/loopfilter_arm.c
+++ b/vp8/common/arm/loopfilter_arm.c
@@ -38,9 +38,8 @@
/*ARMV6 loopfilter functions*/
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -51,20 +50,18 @@
}
void vp8_loop_filter_mbhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -75,20 +72,18 @@
}
void vp8_loop_filter_mbvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -101,12 +96,11 @@
}
void vp8_loop_filter_bhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -114,9 +108,8 @@
/* Vertical B Filtering */
void vp8_loop_filter_bv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_vertical_edge_armv6(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_armv6(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -129,12 +122,11 @@
}
void vp8_loop_filter_bvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -145,9 +137,8 @@
/* NEON loopfilter functions */
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_horizontal_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -155,20 +146,18 @@
}
void vp8_loop_filter_mbhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_vertical_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -176,20 +165,18 @@
}
void vp8_loop_filter_mbvs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -199,12 +186,11 @@
}
void vp8_loop_filter_bhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -212,9 +198,8 @@
/* Vertical B Filtering */
void vp8_loop_filter_bv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -224,12 +209,11 @@
}
void vp8_loop_filter_bvs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_neon(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_neon(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
diff --git a/vp8/common/loopfilter.c b/vp8/common/loopfilter.c
index 221998b..a324271 100644
--- a/vp8/common/loopfilter.c
+++ b/vp8/common/loopfilter.c
@@ -25,9 +25,8 @@
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -38,20 +37,18 @@
}
void vp8_loop_filter_mbhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -62,20 +59,18 @@
}
void vp8_loop_filter_mbvs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -88,12 +83,11 @@
}
void vp8_loop_filter_bhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -101,9 +95,8 @@
/* Vertical B Filtering */
void vp8_loop_filter_bv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_vertical_edge_c(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_c(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_c(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -116,12 +109,11 @@
}
void vp8_loop_filter_bvs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -368,17 +360,17 @@
if (filter_level)
{
if (mb_col > 0)
- cm->lf_mbv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
+ cm->lf_mbv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level]);
if (!skip_lf)
- cm->lf_bv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
+ cm->lf_bv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level]);
/* don't apply across umv border */
if (mb_row > 0)
- cm->lf_mbh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
+ cm->lf_mbh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level]);
if (!skip_lf)
- cm->lf_bh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
+ cm->lf_bh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level]);
}
y_ptr += 16;
@@ -472,17 +464,17 @@
if (filter_level)
{
if (mb_col > 0)
- cm->lf_mbv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_mbv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
if (!skip_lf)
- cm->lf_bv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_bv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
/* don't apply across umv border */
if (mb_row > 0)
- cm->lf_mbh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_mbh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
if (!skip_lf)
- cm->lf_bh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_bh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
}
y_ptr += 16;
@@ -581,15 +573,15 @@
if (filter_level)
{
if (mb_col > 0)
- cm->lf_mbv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_mbv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
if (!skip_lf)
- cm->lf_bv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_bv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
- cm->lf_mbh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_mbh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
if (!skip_lf)
- cm->lf_bh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
+ cm->lf_bh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level]);
}
y_ptr += 16;
diff --git a/vp8/common/loopfilter.h b/vp8/common/loopfilter.h
index 2e5997c..ca136b3 100644
--- a/vp8/common/loopfilter.h
+++ b/vp8/common/loopfilter.h
@@ -41,7 +41,7 @@
#define prototype_loopfilter_block(sym) \
void sym(unsigned char *y, unsigned char *u, unsigned char *v,\
- int ystride, int uv_stride, loop_filter_info *lfi, int simpler)
+ int ystride, int uv_stride, loop_filter_info *lfi)
#if ARCH_X86 || ARCH_X86_64
#include "x86/loopfilter_x86.h"
diff --git a/vp8/common/onyx.h b/vp8/common/onyx.h
index 426b8fc..a059519 100644
--- a/vp8/common/onyx.h
+++ b/vp8/common/onyx.h
@@ -109,6 +109,7 @@
int noise_sensitivity; // parameter used for applying pre processing blur: recommendation 0
int Sharpness; // parameter used for sharpening output: recommendation 0:
int cpu_used;
+ unsigned int rc_max_intra_bitrate_pct;
// mode ->
//(0)=Realtime/Live Encoding. This mode is optimized for realtim encoding (for example, capturing
@@ -139,8 +140,9 @@
int end_usage; // vbr or cbr
- // shoot to keep buffer full at all times by undershooting a bit 95 recommended
+ // buffer targeting aggressiveness
int under_shoot_pct;
+ int over_shoot_pct;
// buffering parameters
int starting_buffer_level; // in seconds
diff --git a/vp8/common/onyxc_int.h b/vp8/common/onyxc_int.h
index c8c2277..03aed81 100644
--- a/vp8/common/onyxc_int.h
+++ b/vp8/common/onyxc_int.h
@@ -119,7 +119,6 @@
/* profile settings */
int mb_no_coeff_skip;
int no_lpf;
- int simpler_lpf;
int use_bilinear_mc_filter;
int full_pixel;
diff --git a/vp8/common/ppc/loopfilter_altivec.c b/vp8/common/ppc/loopfilter_altivec.c
index e602fee..71bf6e2 100644
--- a/vp8/common/ppc/loopfilter_altivec.c
+++ b/vp8/common/ppc/loopfilter_altivec.c
@@ -53,9 +53,8 @@
// Horizontal MB filtering
void loop_filter_mbh_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
mbloop_filter_horizontal_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
if (u_ptr)
@@ -63,9 +62,8 @@
}
void loop_filter_mbhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
(void)u_ptr;
(void)v_ptr;
(void)uv_stride;
@@ -74,9 +72,8 @@
// Vertical MB Filtering
void loop_filter_mbv_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
mbloop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
if (u_ptr)
@@ -84,9 +81,8 @@
}
void loop_filter_mbvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
(void)u_ptr;
(void)v_ptr;
(void)uv_stride;
@@ -95,9 +91,8 @@
// Horizontal B Filtering
void loop_filter_bh_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
// These should all be done at once with one call, instead of 3
loop_filter_horizontal_edge_y_ppc(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
loop_filter_horizontal_edge_y_ppc(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
@@ -108,9 +103,8 @@
}
void loop_filter_bhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
(void)u_ptr;
(void)v_ptr;
(void)uv_stride;
@@ -121,9 +115,8 @@
// Vertical B Filtering
void loop_filter_bv_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
loop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->flim, lfi->lim, lfi->thr);
if (u_ptr)
@@ -131,9 +124,8 @@
}
void loop_filter_bvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void)simpler_lpf;
(void)u_ptr;
(void)v_ptr;
(void)uv_stride;
diff --git a/vp8/common/x86/loopfilter_x86.c b/vp8/common/x86/loopfilter_x86.c
index 5837bc0..a52420c 100644
--- a/vp8/common/x86/loopfilter_x86.c
+++ b/vp8/common/x86/loopfilter_x86.c
@@ -42,9 +42,8 @@
#if HAVE_MMX
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -56,21 +55,19 @@
void vp8_loop_filter_mbhs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -82,21 +79,19 @@
void vp8_loop_filter_mbvs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -110,12 +105,11 @@
void vp8_loop_filter_bhs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -124,9 +118,8 @@
/* Vertical B Filtering */
void vp8_loop_filter_bv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_vertical_edge_mmx(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_mmx(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -140,12 +133,11 @@
void vp8_loop_filter_bvs_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -156,9 +148,8 @@
/* Horizontal MB filtering */
#if HAVE_SSE2
void vp8_loop_filter_mbh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -167,21 +158,19 @@
void vp8_loop_filter_mbhs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
@@ -190,21 +179,19 @@
void vp8_loop_filter_mbvs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -215,12 +202,11 @@
void vp8_loop_filter_bhs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -229,9 +215,8 @@
/* Vertical B Filtering */
void vp8_loop_filter_bv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
- (void) simpler_lpf;
vp8_loop_filter_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
@@ -242,12 +227,11 @@
void vp8_loop_filter_bvs_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
- int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
+ int y_stride, int uv_stride, loop_filter_info *lfi)
{
(void) u_ptr;
(void) v_ptr;
(void) uv_stride;
- (void) simpler_lpf;
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
diff --git a/vp8/decoder/threading.c b/vp8/decoder/threading.c
index fe6dd09..16afd59 100644
--- a/vp8/decoder/threading.c
+++ b/vp8/decoder/threading.c
@@ -368,17 +368,17 @@
if (filter_level)
{
if (mb_col > 0)
- pc->lf_mbv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_mbv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
if (!skip_lf)
- pc->lf_bv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_bv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
/* don't apply across umv border */
if (mb_row > 0)
- pc->lf_mbh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_mbh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
if (!skip_lf)
- pc->lf_bh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_bh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
}
}
@@ -860,17 +860,17 @@
if (filter_level)
{
if (mb_col > 0)
- pc->lf_mbv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_mbv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
if (!skip_lf)
- pc->lf_bv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_bv(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
/* don't apply across umv border */
if (mb_row > 0)
- pc->lf_mbh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_mbh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
if (!skip_lf)
- pc->lf_bh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level], pc->simpler_lpf);
+ pc->lf_bh(xd->dst.y_buffer, xd->dst.u_buffer, xd->dst.v_buffer, recon_y_stride, recon_uv_stride, &lfi[filter_level]);
}
}
diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c
index 92e0cbb..c7579e8 100644
--- a/vp8/encoder/onyx_if.c
+++ b/vp8/encoder/onyx_if.c
@@ -1459,10 +1459,6 @@
cpi->avg_frame_qindex = cpi->oxcf.worst_allowed_q;
// Initialise the starting buffer levels
- cpi->oxcf.starting_buffer_level =
- rescale(cpi->oxcf.starting_buffer_level,
- cpi->oxcf.target_bandwidth, 1000);
-
cpi->buffer_level = cpi->oxcf.starting_buffer_level;
cpi->bits_off_target = cpi->oxcf.starting_buffer_level;
@@ -1635,6 +1631,10 @@
// Convert target bandwidth from Kbit/s to Bit/s
cpi->oxcf.target_bandwidth *= 1000;
+ cpi->oxcf.starting_buffer_level =
+ rescale(cpi->oxcf.starting_buffer_level,
+ cpi->oxcf.target_bandwidth, 1000);
+
// Set or reset optimal and maximum buffer levels.
if (cpi->oxcf.optimal_buffer_level == 0)
cpi->oxcf.optimal_buffer_level = cpi->oxcf.target_bandwidth / 8;
@@ -1684,8 +1684,6 @@
// Only allow dropped frames in buffered mode
cpi->drop_frames_allowed = cpi->oxcf.allow_df && cpi->buffered_mode;
- cm->filter_type = (LOOPFILTERTYPE) cpi->filter_type;
-
if (!cm->use_bilinear_mc_filter)
cm->mcomp_filter_type = SIXTAP;
else
@@ -1699,9 +1697,6 @@
cm->horiz_scale = cpi->horiz_scale;
cm->vert_scale = cpi->vert_scale ;
- // As per VP8
- cpi->intra_frame_target = (4 * (cm->Width + cm->Height) / 15) * 1000;
-
// VP8 sharpness level mapping 0-7 (vs 0-10 in general VPx dialogs)
if (cpi->oxcf.Sharpness > 7)
cpi->oxcf.Sharpness = 7;
@@ -1731,10 +1726,6 @@
vp8_alloc_compressor_data(cpi);
}
- // Clamp KF frame size to quarter of data rate
- if (cpi->intra_frame_target > cpi->target_bandwidth >> 2)
- cpi->intra_frame_target = cpi->target_bandwidth >> 2;
-
if (cpi->oxcf.fixed_q >= 0)
{
cpi->last_q[0] = cpi->oxcf.fixed_q;
@@ -1960,7 +1951,6 @@
cpi->frames_till_gf_update_due = 0;
cpi->key_frame_count = 1;
- cpi->tot_key_frame_bits = 0;
cpi->ni_av_qi = cpi->oxcf.worst_allowed_q;
cpi->ni_tot_qi = 0;
@@ -1986,7 +1976,6 @@
for (i = 0; i < KEY_FRAME_CONTEXT; i++)
{
- cpi->prior_key_frame_size[i] = cpi->intra_frame_target;
cpi->prior_key_frame_distance[i] = (int)cpi->output_frame_rate;
}
@@ -2670,16 +2659,17 @@
if (cpi->pass == 2)
vp8_calc_auto_iframe_target_size(cpi);
- // 1 Pass there is no information on which to base size so use bandwidth per second * fixed fraction
else
#endif
- cpi->this_frame_target = cpi->oxcf.target_bandwidth / 2;
-
- // in error resilient mode the first frame is bigger since it likely contains
- // all the static background
- if (cpi->oxcf.error_resilient_mode == 1 || (cpi->compressor_speed == 2))
{
- cpi->this_frame_target *= 3; // 5;
+ /* 1 Pass there is no information on which to base size so use
+ * bandwidth per second * fraction of the initial buffer
+ * level
+ */
+ cpi->this_frame_target = cpi->oxcf.starting_buffer_level / 2;
+
+ if(cpi->this_frame_target > cpi->oxcf.target_bandwidth * 3 / 2)
+ cpi->this_frame_target = cpi->oxcf.target_bandwidth * 3 / 2;
}
// Key frame from VFW/auto-keyframe/first frame
@@ -2713,6 +2703,19 @@
}
}
+ /* Apply limits on keyframe target.
+ *
+ * TODO: move this after consolidating
+ * vp8_calc_iframe_target_size() and vp8_calc_auto_iframe_target_size()
+ */
+ if (cm->frame_type == KEY_FRAME && cpi->oxcf.rc_max_intra_bitrate_pct)
+ {
+ unsigned int max_rate = cpi->av_per_frame_bandwidth
+ * cpi->oxcf.rc_max_intra_bitrate_pct / 100;
+
+ if (cpi->this_frame_target > max_rate)
+ cpi->this_frame_target = max_rate;
+ }
return 1;
}
@@ -4370,9 +4373,9 @@
vp8_clear_system_state(); //__asm emms;
if (cpi->total_coded_error_left != 0.0)
- fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6ld %6ld"
- "%6ld %6ld %6ld %5ld %5ld %5ld %8ld %8.2f %10d %10.3f"
- "%10.3f %8ld\n",
+ fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6d %6d"
+ "%6d %6d %6d %5d %5d %5d %8d %8.2f %10d %10.3f"
+ "%10.3f %8d\n",
cpi->common.current_video_frame, cpi->this_frame_target,
cpi->projected_frame_size,
(cpi->projected_frame_size - cpi->this_frame_target),
@@ -4389,9 +4392,9 @@
(double)cpi->bits_left / cpi->total_coded_error_left,
cpi->tot_recode_hits);
else
- fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6ld %6ld"
- "%6ld %6ld %6ld %5ld %5ld %5ld %8ld %8.2f %10d %10.3f"
- "%8ld\n",
+ fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6d %6d"
+ "%6d %6d %6d %5d %5d %5d %8d %8.2f %10d %10.3f"
+ "%8d\n",
cpi->common.current_video_frame,
cpi->this_frame_target, cpi->projected_frame_size,
(cpi->projected_frame_size - cpi->this_frame_target),
@@ -5184,35 +5187,6 @@
}
-static int calc_low_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd)
-{
- int i, j;
- int Total = 0;
-
- unsigned char *src = source->y_buffer;
- unsigned char *dst = dest->y_buffer;
- (void)rtcd;
-
- // Loop through the Y plane raw and reconstruction data summing (square differences)
- for (i = 0; i < source->y_height; i += 16)
- {
- for (j = 0; j < source->y_width; j += 16)
- {
- unsigned int sse;
- VARIANCE_INVOKE(rtcd, mse16x16)(src + j, source->y_stride, dst + j, dest->y_stride, &sse);
-
- if (sse < 8096)
- Total += sse;
- }
-
- src += 16 * source->y_stride;
- dst += 16 * dest->y_stride;
- }
-
- return Total;
-}
-
-
int vp8_get_quantizer(VP8_PTR c)
{
VP8_COMP *cpi = (VP8_COMP *) c;
diff --git a/vp8/encoder/onyx_int.h b/vp8/encoder/onyx_int.h
index 4c18542..e7ad8f3 100644
--- a/vp8/encoder/onyx_int.h
+++ b/vp8/encoder/onyx_int.h
@@ -385,14 +385,11 @@
int active_arnr_frames; // <= cpi->oxcf.arnr_max_frames
INT64 key_frame_count;
- INT64 tot_key_frame_bits;
- int prior_key_frame_size[KEY_FRAME_CONTEXT];
int prior_key_frame_distance[KEY_FRAME_CONTEXT];
int per_frame_bandwidth; // Current section per frame bandwidth target
int av_per_frame_bandwidth; // Average frame size target for clip
int min_frame_bandwidth; // Minimum allocation that should be used for any frame
int last_key_frame_size;
- int intra_frame_target;
int inter_frame_target;
double output_frame_rate;
long long last_time_stamp_seen;
@@ -509,7 +506,6 @@
int auto_adjust_key_quantizer;
int keyquantizer;
int auto_worst_q;
- int filter_type;
int cpu_used;
int chroma_boost;
int horiz_scale;
diff --git a/vp8/encoder/ratectrl.c b/vp8/encoder/ratectrl.c
index 9821d29..09a69bc 100644
--- a/vp8/encoder/ratectrl.c
+++ b/vp8/encoder/ratectrl.c
@@ -662,8 +662,6 @@
int min_frame_target;
int Adjustment;
- // Set the min frame bandwidth.
- //min_frame_target = estimate_min_frame_size( cpi );
min_frame_target = 0;
if (cpi->pass == 2)
@@ -817,11 +815,6 @@
}
}
- // Set a reduced data rate target for our initial Q calculation.
- // This should help to save bits during earier sections.
- if ((cpi->oxcf.under_shoot_pct > 0) && (cpi->oxcf.under_shoot_pct <= 100))
- cpi->this_frame_target = (cpi->this_frame_target * cpi->oxcf.under_shoot_pct) / 100;
-
// Sanity check that the total sum of adjustments is not above the maximum allowed
// That is that having allowed for KF and GF penalties we have not pushed the
// current interframe target to low. If the adjustment we apply here is not capable of recovering
@@ -858,11 +851,6 @@
percent_low =
(cpi->oxcf.optimal_buffer_level - cpi->buffer_level) /
one_percent_bits;
-
- if (percent_low > 100)
- percent_low = 100;
- else if (percent_low < 0)
- percent_low = 0;
}
// Are we overshooting the long term clip data rate...
else if (cpi->bits_off_target < 0)
@@ -870,16 +858,16 @@
// Adjust per frame data target downwards to compensate.
percent_low = (int)(100 * -cpi->bits_off_target /
(cpi->total_byte_count * 8));
-
- if (percent_low > 100)
- percent_low = 100;
- else if (percent_low < 0)
- percent_low = 0;
}
+ if (percent_low > cpi->oxcf.under_shoot_pct)
+ percent_low = cpi->oxcf.under_shoot_pct;
+ else if (percent_low < 0)
+ percent_low = 0;
+
// lower the target bandwidth for this frame.
- cpi->this_frame_target =
- (cpi->this_frame_target * (100 - (percent_low / 2))) / 100;
+ cpi->this_frame_target -= (cpi->this_frame_target * percent_low)
+ / 200;
// Are we using allowing control of active_worst_allowed_q
// according to buffer level.
@@ -950,20 +938,29 @@
}
else
{
- int percent_high;
+ int percent_high = 0;
- if (cpi->bits_off_target > cpi->oxcf.optimal_buffer_level)
+ if ((cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
+ && (cpi->buffer_level > cpi->oxcf.optimal_buffer_level))
{
- percent_high = (int)(100 * (cpi->bits_off_target - cpi->oxcf.optimal_buffer_level) / (cpi->total_byte_count * 8));
-
- if (percent_high > 100)
- percent_high = 100;
- else if (percent_high < 0)
- percent_high = 0;
-
- cpi->this_frame_target = (cpi->this_frame_target * (100 + (percent_high / 2))) / 100;
-
+ percent_high = (cpi->buffer_level
+ - cpi->oxcf.optimal_buffer_level)
+ / one_percent_bits;
}
+ else if (cpi->bits_off_target > cpi->oxcf.optimal_buffer_level)
+ {
+ percent_high = (int)((100 * cpi->bits_off_target)
+ / (cpi->total_byte_count * 8));
+ }
+
+ if (percent_high > cpi->oxcf.over_shoot_pct)
+ percent_high = cpi->oxcf.over_shoot_pct;
+ else if (percent_high < 0)
+ percent_high = 0;
+
+ cpi->this_frame_target += (cpi->this_frame_target *
+ percent_high) / 200;
+
// Are we allowing control of active_worst_allowed_q according to bufferl level.
if (cpi->auto_worst_q)
@@ -1419,119 +1416,85 @@
return Q;
}
-static int estimate_min_frame_size(VP8_COMP *cpi)
-{
- double correction_factor;
- int bits_per_mb_at_max_q;
- // This funtion returns a default value for the first few frames untill the correction factor has had time to adapt.
- if (cpi->common.current_video_frame < 10)
- {
- if (cpi->pass == 2)
- return (cpi->min_frame_bandwidth);
- else
- return cpi->per_frame_bandwidth / 3;
- }
-
- /* // Select the appropriate correction factor based upon type of frame.
- if ( cpi->common.frame_type == KEY_FRAME )
- correction_factor = cpi->key_frame_rate_correction_factor;
- else
- {
- if ( cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame )
- correction_factor = cpi->gf_rate_correction_factor;
- else
- correction_factor = cpi->rate_correction_factor;
- }*/
-
- // We estimate at half the value we get from vp8_bits_per_mb
- correction_factor = cpi->rate_correction_factor / 2.0;
-
- bits_per_mb_at_max_q = (int)(.5 + correction_factor * vp8_bits_per_mb[cpi->common.frame_type][MAXQ]);
-
- return (bits_per_mb_at_max_q * cpi->common.MBs) >> BPER_MB_NORMBITS;
-}
-
-void vp8_adjust_key_frame_context(VP8_COMP *cpi)
+static int estimate_keyframe_frequency(VP8_COMP *cpi)
{
int i;
- int av_key_frames_per_second;
- // Average key frame frequency and size
- unsigned int total_weight = 0;
- unsigned int av_key_frame_frequency = 0;
- unsigned int av_key_frame_bits = 0;
+ // Average key frame frequency
+ int av_key_frame_frequency = 0;
- unsigned int output_frame_rate = (unsigned int)(100 * cpi->output_frame_rate);
- unsigned int target_bandwidth = (unsigned int)(100 * cpi->target_bandwidth);
-
- // Clear down mmx registers to allow floating point in what follows
- vp8_clear_system_state(); //__asm emms;
-
- // Update the count of total key frame bits
- cpi->tot_key_frame_bits += cpi->projected_frame_size;
-
- // First key frame at start of sequence is a special case. We have no frequency data.
+ /* First key frame at start of sequence is a special case. We have no
+ * frequency data.
+ */
if (cpi->key_frame_count == 1)
{
- av_key_frame_frequency = (int)cpi->output_frame_rate * 2; // Assume a default of 1 kf every 2 seconds
- av_key_frame_bits = cpi->projected_frame_size;
- av_key_frames_per_second = output_frame_rate / av_key_frame_frequency; // Note output_frame_rate not cpi->output_frame_rate
+ /* Assume a default of 1 kf every 2 seconds, or the max kf interval,
+ * whichever is smaller.
+ */
+ int key_freq = cpi->oxcf.key_freq>0 ? cpi->oxcf.key_freq : 1;
+ av_key_frame_frequency = (int)cpi->output_frame_rate * 2;
+
+ if (cpi->oxcf.auto_key && av_key_frame_frequency > key_freq)
+ av_key_frame_frequency = cpi->oxcf.key_freq;
+
+ cpi->prior_key_frame_distance[KEY_FRAME_CONTEXT - 1]
+ = av_key_frame_frequency;
}
else
{
+ unsigned int total_weight = 0;
int last_kf_interval =
(cpi->frames_since_key > 0) ? cpi->frames_since_key : 1;
- // reset keyframe context and calculate weighted average of last KEY_FRAME_CONTEXT keyframes
+ /* reset keyframe context and calculate weighted average of last
+ * KEY_FRAME_CONTEXT keyframes
+ */
for (i = 0; i < KEY_FRAME_CONTEXT; i++)
{
if (i < KEY_FRAME_CONTEXT - 1)
- {
- cpi->prior_key_frame_size[i] = cpi->prior_key_frame_size[i+1];
- cpi->prior_key_frame_distance[i] = cpi->prior_key_frame_distance[i+1];
- }
+ cpi->prior_key_frame_distance[i]
+ = cpi->prior_key_frame_distance[i+1];
else
- {
- cpi->prior_key_frame_size[i] = cpi->projected_frame_size;
cpi->prior_key_frame_distance[i] = last_kf_interval;
- }
- av_key_frame_bits += prior_key_frame_weight[i] * cpi->prior_key_frame_size[i];
- av_key_frame_frequency += prior_key_frame_weight[i] * cpi->prior_key_frame_distance[i];
- total_weight += prior_key_frame_weight[i];
+ av_key_frame_frequency += prior_key_frame_weight[i]
+ * cpi->prior_key_frame_distance[i];
+ total_weight += prior_key_frame_weight[i];
}
- av_key_frame_bits /= total_weight;
av_key_frame_frequency /= total_weight;
- av_key_frames_per_second = output_frame_rate / av_key_frame_frequency;
}
+ return av_key_frame_frequency;
+}
+
+
+void vp8_adjust_key_frame_context(VP8_COMP *cpi)
+{
+ // Clear down mmx registers to allow floating point in what follows
+ vp8_clear_system_state();
// Do we have any key frame overspend to recover?
- if ((cpi->pass != 2) && (cpi->projected_frame_size > cpi->per_frame_bandwidth))
+ // Two-pass overspend handled elsewhere.
+ if ((cpi->pass != 2)
+ && (cpi->projected_frame_size > cpi->per_frame_bandwidth))
{
- // Update the count of key frame overspend to be recovered in subsequent frames
- // A portion of the KF overspend is treated as gf overspend (and hence recovered more quickly)
- // as the kf is also a gf. Otherwise the few frames following each kf tend to get more bits
- // allocated than those following other gfs.
- cpi->kf_overspend_bits += (cpi->projected_frame_size - cpi->per_frame_bandwidth) * 7 / 8;
- cpi->gf_overspend_bits += (cpi->projected_frame_size - cpi->per_frame_bandwidth) * 1 / 8;
- if(!av_key_frame_frequency)
- av_key_frame_frequency = 60;
+ int overspend;
- // Work out how much to try and recover per frame.
- // For one pass we estimate the number of frames to spread it over based upon past history.
- // For two pass we know how many frames there will be till the next kf.
- if (cpi->pass == 2)
- {
- if (cpi->frames_to_key > 16)
- cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits / (int)cpi->frames_to_key;
- else
- cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits / 16;
- }
- else
- cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits / (int)av_key_frame_frequency;
+ /* Update the count of key frame overspend to be recovered in
+ * subsequent frames. A portion of the KF overspend is treated as gf
+ * overspend (and hence recovered more quickly) as the kf is also a
+ * gf. Otherwise the few frames following each kf tend to get more
+ * bits allocated than those following other gfs.
+ */
+ overspend = (cpi->projected_frame_size - cpi->per_frame_bandwidth);
+ cpi->kf_overspend_bits += overspend * 7 / 8;
+ cpi->gf_overspend_bits += overspend * 1 / 8;
+
+ /* Work out how much to try and recover per frame. */
+ cpi->kf_bitrate_adjustment = cpi->kf_overspend_bits
+ / estimate_keyframe_frequency(cpi);
}
cpi->frames_since_key = 0;
@@ -1539,6 +1502,7 @@
cpi->key_frame_count++;
}
+
void vp8_compute_frame_size_bounds(VP8_COMP *cpi, int *frame_under_shoot_limit, int *frame_over_shoot_limit)
{
// Set-up bounds on acceptable frame size:
diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c
index 1f329d7..e99d6f0 100644
--- a/vp8/encoder/rdopt.c
+++ b/vp8/encoder/rdopt.c
@@ -243,9 +243,6 @@
vp8_set_speed_features(cpi);
- if (cpi->common.simpler_lpf)
- cpi->common.filter_type = SIMPLE_LOOPFILTER;
-
q = (int)pow(Qvalue, 1.25);
if (q < 8)
diff --git a/vp8/encoder/x86/quantize_sse4.asm b/vp8/encoder/x86/quantize_sse4.asm
new file mode 100644
index 0000000..258899e
--- /dev/null
+++ b/vp8/encoder/x86/quantize_sse4.asm
@@ -0,0 +1,254 @@
+;
+; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+; Use of this source code is governed by a BSD-style license and patent
+; grant that can be found in the LICENSE file in the root of the source
+; tree. All contributing project authors may be found in the AUTHORS
+; file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+%include "asm_enc_offsets.asm"
+
+
+; void vp8_regular_quantize_b_sse4 | arg
+; (BLOCK *b, | 0
+; BLOCKD *d) | 1
+
+global sym(vp8_regular_quantize_b_sse4)
+sym(vp8_regular_quantize_b_sse4):
+
+%if ABI_IS_32BIT
+ push rbp
+ mov rbp, rsp
+ GET_GOT rbx
+ push rdi
+ push rsi
+
+ ALIGN_STACK 16, rax
+ %define qcoeff 0 ; 32
+ %define stack_size 32
+ sub rsp, stack_size
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ SAVE_XMM 8, u
+ push rdi
+ push rsi
+ %endif
+%endif
+ ; end prolog
+
+%if ABI_IS_32BIT
+ mov rdi, arg(0) ; BLOCK *b
+ mov rsi, arg(1) ; BLOCKD *d
+%else
+ %ifidn __OUTPUT_FORMAT__,x64
+ mov rdi, rcx ; BLOCK *b
+ mov rsi, rdx ; BLOCKD *d
+ %else
+ ;mov rdi, rdi ; BLOCK *b
+ ;mov rsi, rsi ; BLOCKD *d
+ %endif
+%endif
+
+ mov rax, [rdi + vp8_block_coeff]
+ mov rcx, [rdi + vp8_block_zbin]
+ mov rdx, [rdi + vp8_block_round]
+ movd xmm7, [rdi + vp8_block_zbin_extra]
+
+ ; z
+ movdqa xmm0, [rax]
+ movdqa xmm1, [rax + 16]
+
+ ; duplicate zbin_oq_value
+ pshuflw xmm7, xmm7, 0
+ punpcklwd xmm7, xmm7
+
+ movdqa xmm2, xmm0
+ movdqa xmm3, xmm1
+
+ ; sz
+ psraw xmm0, 15
+ psraw xmm1, 15
+
+ ; (z ^ sz)
+ pxor xmm2, xmm0
+ pxor xmm3, xmm1
+
+ ; x = abs(z)
+ psubw xmm2, xmm0
+ psubw xmm3, xmm1
+
+ ; zbin
+ movdqa xmm4, [rcx]
+ movdqa xmm5, [rcx + 16]
+
+ ; *zbin_ptr + zbin_oq_value
+ paddw xmm4, xmm7
+ paddw xmm5, xmm7
+
+ movdqa xmm6, xmm2
+ movdqa xmm7, xmm3
+
+ ; x - (*zbin_ptr + zbin_oq_value)
+ psubw xmm6, xmm4
+ psubw xmm7, xmm5
+
+ ; round
+ movdqa xmm4, [rdx]
+ movdqa xmm5, [rdx + 16]
+
+ mov rax, [rdi + vp8_block_quant_shift]
+ mov rcx, [rdi + vp8_block_quant]
+ mov rdx, [rdi + vp8_block_zrun_zbin_boost]
+
+ ; x + round
+ paddw xmm2, xmm4
+ paddw xmm3, xmm5
+
+ ; quant
+ movdqa xmm4, [rcx]
+ movdqa xmm5, [rcx + 16]
+
+ ; y = x * quant_ptr >> 16
+ pmulhw xmm4, xmm2
+ pmulhw xmm5, xmm3
+
+ ; y += x
+ paddw xmm2, xmm4
+ paddw xmm3, xmm5
+
+ pxor xmm4, xmm4
+%if ABI_IS_32BIT
+ movdqa [rsp + qcoeff], xmm4
+ movdqa [rsp + qcoeff + 16], xmm4
+%else
+ pxor xmm8, xmm8
+%endif
+
+ ; quant_shift
+ movdqa xmm5, [rax]
+
+ ; zrun_zbin_boost
+ mov rax, rdx
+
+%macro ZIGZAG_LOOP 5
+ ; x
+ pextrw ecx, %4, %2
+
+ ; if (x >= zbin)
+ sub cx, WORD PTR[rdx] ; x - zbin
+ lea rdx, [rdx + 2] ; zbin_boost_ptr++
+ jl rq_zigzag_loop_%1 ; x < zbin
+
+ pextrw edi, %3, %2 ; y
+
+ ; downshift by quant_shift[rc]
+ pextrb ecx, xmm5, %1 ; quant_shift[rc]
+ sar edi, cl ; also sets Z bit
+ je rq_zigzag_loop_%1 ; !y
+%if ABI_IS_32BIT
+ mov WORD PTR[rsp + qcoeff + %1 *2], di
+%else
+ pinsrw %5, edi, %2 ; qcoeff[rc]
+%endif
+ mov rdx, rax ; reset to b->zrun_zbin_boost
+rq_zigzag_loop_%1:
+%endmacro
+; in vp8_default_zig_zag1d order: see vp8/common/entropy.c
+ZIGZAG_LOOP 0, 0, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 1, 1, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 4, 4, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 8, 0, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 5, 5, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 2, 2, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 3, 3, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 6, 6, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 9, 1, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 12, 4, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 13, 5, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 10, 2, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 7, 7, xmm2, xmm6, xmm4
+ZIGZAG_LOOP 11, 3, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 14, 6, xmm3, xmm7, xmm8
+ZIGZAG_LOOP 15, 7, xmm3, xmm7, xmm8
+
+ mov rcx, [rsi + vp8_blockd_dequant]
+ mov rdi, [rsi + vp8_blockd_dqcoeff]
+
+%if ABI_IS_32BIT
+ movdqa xmm4, [rsp + qcoeff]
+ movdqa xmm5, [rsp + qcoeff + 16]
+%else
+ %define xmm5 xmm8
+%endif
+
+ ; y ^ sz
+ pxor xmm4, xmm0
+ pxor xmm5, xmm1
+ ; x = (y ^ sz) - sz
+ psubw xmm4, xmm0
+ psubw xmm5, xmm1
+
+ ; dequant
+ movdqa xmm0, [rcx]
+ movdqa xmm1, [rcx + 16]
+
+ mov rcx, [rsi + vp8_blockd_qcoeff]
+
+ pmullw xmm0, xmm4
+ pmullw xmm1, xmm5
+
+ ; store qcoeff
+ movdqa [rcx], xmm4
+ movdqa [rcx + 16], xmm5
+
+ ; store dqcoeff
+ movdqa [rdi], xmm0
+ movdqa [rdi + 16], xmm1
+
+ ; select the last value (in zig_zag order) for EOB
+ pxor xmm6, xmm6
+ pcmpeqw xmm4, xmm6
+ pcmpeqw xmm5, xmm6
+
+ packsswb xmm4, xmm5
+ pshufb xmm4, [GLOBAL(zig_zag1d)]
+ pmovmskb edx, xmm4
+ xor rdi, rdi
+ mov eax, -1
+ xor dx, ax
+ bsr eax, edx
+ sub edi, edx
+ sar edi, 31
+ add eax, 1
+ and eax, edi
+
+ mov [rsi + vp8_blockd_eob], eax
+
+ ; begin epilog
+%if ABI_IS_32BIT
+ add rsp, stack_size
+ pop rsp
+
+ pop rsi
+ pop rdi
+ RESTORE_GOT
+ pop rbp
+%else
+ %undef xmm5
+ %ifidn __OUTPUT_FORMAT__,x64
+ pop rsi
+ pop rdi
+ RESTORE_XMM
+ %endif
+%endif
+
+ ret
+
+SECTION_RODATA
+align 16
+; vp8/common/entropy.c: vp8_default_zig_zag1d
+zig_zag1d:
+ db 0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15
diff --git a/vp8/encoder/x86/quantize_x86.h b/vp8/encoder/x86/quantize_x86.h
index f093580..bbe475f 100644
--- a/vp8/encoder/x86/quantize_x86.h
+++ b/vp8/encoder/x86/quantize_x86.h
@@ -51,4 +51,17 @@
#endif /* HAVE_SSSE3 */
+
+#if HAVE_SSE4_1
+extern prototype_quantize_block(vp8_regular_quantize_b_sse4);
+
+#if !CONFIG_RUNTIME_CPU_DETECT
+
+#undef vp8_quantize_quantb
+#define vp8_quantize_quantb vp8_regular_quantize_b_sse4
+
+#endif /* !CONFIG_RUNTIME_CPU_DETECT */
+
+#endif /* HAVE_SSE4_1 */
+
#endif /* QUANTIZE_X86_H */
diff --git a/vp8/encoder/x86/temporal_filter_apply_sse2.asm b/vp8/encoder/x86/temporal_filter_apply_sse2.asm
index 30674c8..b777ef5 100644
--- a/vp8/encoder/x86/temporal_filter_apply_sse2.asm
+++ b/vp8/encoder/x86/temporal_filter_apply_sse2.asm
@@ -164,10 +164,10 @@
movdqa xmm6, [rdi+32]
movdqa xmm7, [rdi+48]
; += modifier
- paddw xmm4, xmm0
- paddw xmm5, xmm2
- paddw xmm6, xmm1
- paddw xmm7, xmm3
+ paddd xmm4, xmm0
+ paddd xmm5, xmm2
+ paddd xmm6, xmm1
+ paddd xmm7, xmm3
; write back
movdqa [rdi], xmm4
movdqa [rdi+16], xmm5
diff --git a/vp8/encoder/x86/x86_csystemdependent.c b/vp8/encoder/x86/x86_csystemdependent.c
index 47a1788..b01319f 100644
--- a/vp8/encoder/x86/x86_csystemdependent.c
+++ b/vp8/encoder/x86/x86_csystemdependent.c
@@ -313,6 +313,8 @@
cpi->rtcd.variance.sad8x8x8 = vp8_sad8x8x8_sse4;
cpi->rtcd.variance.sad4x4x8 = vp8_sad4x4x8_sse4;
cpi->rtcd.search.full_search = vp8_full_search_sadx8;
+
+ cpi->rtcd.quantize.quantb = vp8_regular_quantize_b_sse4;
}
#endif
diff --git a/vp8/vp8_cx_iface.c b/vp8/vp8_cx_iface.c
index 973265c..db60bfe 100644
--- a/vp8/vp8_cx_iface.c
+++ b/vp8/vp8_cx_iface.c
@@ -151,7 +151,8 @@
RANGE_CHECK_HI(cfg, g_lag_in_frames, 0);
#endif
RANGE_CHECK(cfg, rc_end_usage, VPX_VBR, VPX_CQ);
- RANGE_CHECK_HI(cfg, rc_undershoot_pct, 100);
+ RANGE_CHECK_HI(cfg, rc_undershoot_pct, 1000);
+ RANGE_CHECK_HI(cfg, rc_overshoot_pct, 1000);
RANGE_CHECK_HI(cfg, rc_2pass_vbr_bias_pct, 100);
RANGE_CHECK(cfg, kf_mode, VPX_KF_DISABLED, VPX_KF_AUTO);
//RANGE_CHECK_BOOL(cfg, g_delete_firstpassfile);
@@ -304,6 +305,7 @@
}
oxcf->target_bandwidth = cfg.rc_target_bitrate;
+ oxcf->rc_max_intra_bitrate_pct = cfg.rc_max_intra_bitrate_pct;
oxcf->best_allowed_q = cfg.rc_min_quantizer;
oxcf->worst_allowed_q = cfg.rc_max_quantizer;
@@ -311,7 +313,7 @@
oxcf->fixed_q = -1;
oxcf->under_shoot_pct = cfg.rc_undershoot_pct;
- //oxcf->over_shoot_pct = cfg.rc_overshoot_pct;
+ oxcf->over_shoot_pct = cfg.rc_overshoot_pct;
oxcf->maximum_buffer_size = cfg.rc_buf_sz;
oxcf->starting_buffer_level = cfg.rc_buf_initial_sz;
@@ -357,6 +359,7 @@
printf("key_freq: %d\n", oxcf->key_freq);
printf("end_usage: %d\n", oxcf->end_usage);
printf("under_shoot_pct: %d\n", oxcf->under_shoot_pct);
+ printf("over_shoot_pct: %d\n", oxcf->over_shoot_pct);
printf("starting_buffer_level: %d\n", oxcf->starting_buffer_level);
printf("optimal_buffer_level: %d\n", oxcf->optimal_buffer_level);
printf("maximum_buffer_size: %d\n", oxcf->maximum_buffer_size);
@@ -1083,11 +1086,11 @@
{0}, /* rc_twopass_stats_in */
#endif
256, /* rc_target_bandwidth */
-
+ 0, /* rc_max_intra_bitrate_pct */
4, /* rc_min_quantizer */
63, /* rc_max_quantizer */
- 95, /* rc_undershoot_pct */
- 200, /* rc_overshoot_pct */
+ 100, /* rc_undershoot_pct */
+ 100, /* rc_overshoot_pct */
6000, /* rc_max_buffer_size */
4000, /* rc_buffer_initial_size; */
diff --git a/vp8/vp8cx.mk b/vp8/vp8cx.mk
index a7b68db..c178371 100644
--- a/vp8/vp8cx.mk
+++ b/vp8/vp8cx.mk
@@ -117,6 +117,7 @@
VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/variance_impl_ssse3.asm
VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/quantize_ssse3.asm
VP8_CX_SRCS-$(HAVE_SSE4_1) += encoder/x86/sad_sse4.asm
+VP8_CX_SRCS-$(HAVE_SSE4_1) += encoder/x86/quantize_sse4.asm
VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_mmx.asm
VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/encodeopt.asm
VP8_CX_SRCS-$(ARCH_X86_64) += encoder/x86/ssim_opt.asm
diff --git a/vpx/vpx_encoder.h b/vpx/vpx_encoder.h
index 9c44414..6dbce0d 100644
--- a/vpx/vpx_encoder.h
+++ b/vpx/vpx_encoder.h
@@ -398,6 +398,21 @@
unsigned int rc_target_bitrate;
+ /*!\brief Max data rate for Intra frames
+ *
+ * This value controls additional clamping on the maximum size of a
+ * keyframe. It is expressed as a percentage of the average
+ * per-frame bitrate, with the special (and default) value 0 meaning
+ * unlimited, or no additional clamping beyond the codec's built-in
+ * algorithm.
+ *
+ * For example, to allocate no more than 4.5 frames worth of bitrate
+ * to a keyframe, set this to 450.
+ *
+ */
+ unsigned int rc_max_intra_bitrate_pct;
+
+
/*
* quantizer settings
*/
@@ -430,20 +445,28 @@
*/
- /*!\brief Rate control undershoot tolerance
+ /*!\brief Rate control adaptation undershoot control
*
- * This value, expressed as a percentage of the target bitrate, describes
- * the target bitrate for easier frames, allowing bits to be saved for
- * harder frames. Set to zero to use the codec default.
+ * This value, expressed as a percentage of the target bitrate,
+ * controls the maximum allowed adaptation speed of the codec.
+ * This factor controls the maximum amount of bits that can
+ * be subtracted from the target bitrate in order to compensate
+ * for prior overshoot.
+ *
+ * Valid values in the range 0-1000.
*/
unsigned int rc_undershoot_pct;
- /*!\brief Rate control overshoot tolerance
+ /*!\brief Rate control adaptation overshoot control
*
- * This value, expressed as a percentage of the target bitrate, describes
- * the maximum allowed bitrate for a given frame. Set to zero to use the
- * codec default.
+ * This value, expressed as a percentage of the target bitrate,
+ * controls the maximum allowed adaptation speed of the codec.
+ * This factor controls the maximum amount of bits that can
+ * be added to the target bitrate in order to compensate for
+ * prior undershoot.
+ *
+ * Valid values in the range 0-1000.
*/
unsigned int rc_overshoot_pct;
diff --git a/vpxenc.c b/vpxenc.c
index 39256b6..6ea83fb 100644
--- a/vpxenc.c
+++ b/vpxenc.c
@@ -256,6 +256,16 @@
return stats->buf;
}
+/* Stereo 3D packed frame format */
+typedef enum stereo_format
+{
+ STEREO_FORMAT_MONO = 0,
+ STEREO_FORMAT_LEFT_RIGHT = 1,
+ STEREO_FORMAT_BOTTOM_TOP = 2,
+ STEREO_FORMAT_TOP_BOTTOM = 3,
+ STEREO_FORMAT_RIGHT_LEFT = 11
+} stereo_format_t;
+
enum video_file_type
{
FILE_TYPE_RAW,
@@ -606,7 +616,8 @@
static void
write_webm_file_header(EbmlGlobal *glob,
const vpx_codec_enc_cfg_t *cfg,
- const struct vpx_rational *fps)
+ const struct vpx_rational *fps,
+ stereo_format_t stereo_fmt)
{
{
EbmlLoc start;
@@ -650,6 +661,7 @@
Ebml_StartSubElement(glob, &videoStart, Video);
Ebml_SerializeUnsigned(glob, PixelWidth, pixelWidth);
Ebml_SerializeUnsigned(glob, PixelHeight, pixelHeight);
+ Ebml_SerializeUnsigned(glob, StereoMode, stereo_fmt);
Ebml_SerializeFloat(glob, FrameRate, frameRate);
Ebml_EndSubElement(glob, &videoStart); //Video
}
@@ -916,6 +928,16 @@
"Frame width");
static const arg_def_t height = ARG_DEF("h", "height", 1,
"Frame height");
+static const struct arg_enum_list stereo_mode_enum[] = {
+ {"mono" , STEREO_FORMAT_MONO},
+ {"left-right", STEREO_FORMAT_LEFT_RIGHT},
+ {"bottom-top", STEREO_FORMAT_BOTTOM_TOP},
+ {"top-bottom", STEREO_FORMAT_TOP_BOTTOM},
+ {"right-left", STEREO_FORMAT_RIGHT_LEFT},
+ {NULL, 0}
+};
+static const arg_def_t stereo_mode = ARG_DEF_ENUM(NULL, "stereo-mode", 1,
+ "Stereo 3D video format", stereo_mode_enum);
static const arg_def_t timebase = ARG_DEF(NULL, "timebase", 1,
"Stream timebase (frame duration)");
static const arg_def_t error_resilient = ARG_DEF(NULL, "error-resilient", 1,
@@ -926,7 +948,7 @@
static const arg_def_t *global_args[] =
{
&use_yv12, &use_i420, &usage, &threads, &profile,
- &width, &height, &timebase, &framerate, &error_resilient,
+ &width, &height, &stereo_mode, &timebase, &framerate, &error_resilient,
&lag_in_frames, NULL
};
@@ -962,11 +984,14 @@
"Client initial buffer size (ms)");
static const arg_def_t buf_optimal_sz = ARG_DEF(NULL, "buf-optimal-sz", 1,
"Client optimal buffer size (ms)");
+static const arg_def_t max_intra_rate_pct = ARG_DEF(NULL, "max-intra-rate", 1,
+ "Max I-frame bitrate (pct)");
static const arg_def_t *rc_args[] =
{
&dropframe_thresh, &resize_allowed, &resize_up_thresh, &resize_down_thresh,
&end_usage, &target_bitrate, &min_quantizer, &max_quantizer,
&undershoot_pct, &overshoot_pct, &buf_sz, &buf_initial_sz, &buf_optimal_sz,
+ &max_intra_rate_pct,
NULL
};
@@ -1084,7 +1109,6 @@
#define ARG_CTRL_CNT_MAX 10
-
int main(int argc, const char **argv_)
{
vpx_codec_ctx_t encoder;
@@ -1120,6 +1144,7 @@
uint64_t psnr_samples_total = 0;
double psnr_totals[4] = {0, 0, 0, 0};
int psnr_count = 0;
+ stereo_format_t stereo_fmt = STEREO_FORMAT_MONO;
exec_name = argv_[0];
ebml.last_pts_ms = -1;
@@ -1259,6 +1284,8 @@
cfg.g_w = arg_parse_uint(&arg);
else if (arg_match(&arg, &height, argi))
cfg.g_h = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &stereo_mode, argi))
+ stereo_fmt = arg_parse_enum_or_int(&arg);
else if (arg_match(&arg, &timebase, argi))
cfg.g_timebase = arg_parse_rational(&arg);
else if (arg_match(&arg, &error_resilient, argi))
@@ -1279,6 +1306,8 @@
cfg.rc_end_usage = arg_parse_enum_or_int(&arg);
else if (arg_match(&arg, &target_bitrate, argi))
cfg.rc_target_bitrate = arg_parse_uint(&arg);
+ else if (arg_match(&arg, &max_intra_rate_pct, argi))
+ cfg.rc_max_intra_bitrate_pct = arg_parse_uint(&arg);
else if (arg_match(&arg, &min_quantizer, argi))
cfg.rc_min_quantizer = arg_parse_uint(&arg);
else if (arg_match(&arg, &max_quantizer, argi))
@@ -1557,7 +1586,7 @@
if(write_webm)
{
ebml.stream = outfile;
- write_webm_file_header(&ebml, &cfg, &arg_framerate);
+ write_webm_file_header(&ebml, &cfg, &arg_framerate, stereo_fmt);
}
else
write_ivf_file_header(outfile, &cfg, codec->fourcc, 0);
