2010-05-18 17:58:33 +02:00
|
|
|
/*
|
2010-09-09 14:16:39 +02:00
|
|
|
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
|
2010-05-18 17:58:33 +02:00
|
|
|
*
|
2010-06-18 18:39:21 +02:00
|
|
|
* Use of this source code is governed by a BSD-style license
|
2010-06-04 22:19:40 +02:00
|
|
|
* that can be found in the LICENSE file in the root of the source
|
|
|
|
|
* tree. An additional intellectual property rights grant can be found
|
2010-06-18 18:39:21 +02:00
|
|
|
* in the file PATENTS. All contributing project authors may
|
2010-06-04 22:19:40 +02:00
|
|
|
* be found in the AUTHORS file in the root of the source tree.
|
2010-05-18 17:58:33 +02:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#ifndef SUBPIXEL_H
|
|
|
|
|
#define SUBPIXEL_H
|
|
|
|
|
|
|
|
|
|
#define prototype_subpixel_predict(sym) \
|
|
|
|
|
void sym(unsigned char *src, int src_pitch, int xofst, int yofst, \
|
|
|
|
|
unsigned char *dst, int dst_pitch)
|
|
|
|
|
|
|
|
|
|
#if ARCH_X86 || ARCH_X86_64
|
|
|
|
|
#include "x86/subpixel_x86.h"
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#if ARCH_ARM
|
|
|
|
|
#include "arm/subpixel_arm.h"
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_sixtap16x16
|
|
|
|
|
#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_sixtap16x16);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_sixtap8x8
|
|
|
|
|
#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_sixtap8x8);
|
|
|
|
|
|
Dual 16x16 inter prediction.
This patch introduces the concept of dual inter16x16 prediction. A
16x16 inter-predicted macroblock can use 2 references instead of 1,
where both references use the same mvmode (new, near/est, zero). In the
case of newmv, this means that two MVs are coded instead of one. The
frame can be encoded in 3 ways: all MBs single-prediction, all MBs dual
prediction, or per-MB single/dual prediction selection ("hybrid"), in
which case a single bit is coded per-MB to indicate whether the MB uses
single or dual inter prediction.
In the future, we can (maybe?) get further gains by mixing this with
Adrian's 32x32 work, per-segment dual prediction settings, or adding
support for dual splitmv/8x8mv inter prediction.
Gain (on derf-set, CQ mode) is ~2.8% (SSIM) or ~3.6% (glb PSNR). Most
gain is at medium/high bitrates, but there's minor gains at low bitrates
also. Output was confirmed to match between encoder and decoder.
Note for optimization people: this patch introduces a 2nd version of
16x16/8x8 sixtap/bilin functions, which does an avg instead of a
store. They may want to look and make sure this is implemented to
their satisfaction so we can optimize it best in the future.
Change-ID: I59dc84b07cbb3ccf073ac0f756d03d294cb19281
2011-12-06 20:53:02 +01:00
|
|
|
#ifndef vp8_subpix_sixtap_avg16x16
|
|
|
|
|
#define vp8_subpix_sixtap_avg16x16 vp8_sixtap_predict_avg16x16_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_sixtap_avg16x16);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_sixtap_avg8x8
|
|
|
|
|
#define vp8_subpix_sixtap_avg8x8 vp8_sixtap_predict_avg8x8_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_sixtap_avg8x8);
|
2010-05-18 17:58:33 +02:00
|
|
|
#ifndef vp8_subpix_sixtap8x4
|
|
|
|
|
#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_sixtap8x4);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_sixtap4x4
|
|
|
|
|
#define vp8_subpix_sixtap4x4 vp8_sixtap_predict_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_sixtap4x4);
|
|
|
|
|
|
2012-04-18 22:51:58 +02:00
|
|
|
#ifndef vp8_subpix_sixtap_avg4x4
|
|
|
|
|
#define vp8_subpix_sixtap_avg4x4 vp8_sixtap_predict_avg_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_sixtap_avg4x4);
|
|
|
|
|
|
2012-03-19 15:53:05 +01:00
|
|
|
#if CONFIG_ENHANCED_INTERP
|
|
|
|
|
#ifndef vp8_subpix_eighttap16x16
|
|
|
|
|
#define vp8_subpix_eighttap16x16 vp8_eighttap_predict16x16_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap16x16);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap8x8
|
|
|
|
|
#define vp8_subpix_eighttap8x8 vp8_eighttap_predict8x8_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap8x8);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap_avg16x16
|
|
|
|
|
#define vp8_subpix_eighttap_avg16x16 vp8_eighttap_predict_avg16x16_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap_avg16x16);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap_avg8x8
|
|
|
|
|
#define vp8_subpix_eighttap_avg8x8 vp8_eighttap_predict_avg8x8_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap_avg8x8);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap8x4
|
|
|
|
|
#define vp8_subpix_eighttap8x4 vp8_eighttap_predict8x4_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap8x4);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap4x4
|
|
|
|
|
#define vp8_subpix_eighttap4x4 vp8_eighttap_predict_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap4x4);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap16x16_sharp
|
|
|
|
|
#define vp8_subpix_eighttap16x16_sharp vp8_eighttap_predict16x16_sharp_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap16x16_sharp);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap8x8_sharp
|
|
|
|
|
#define vp8_subpix_eighttap8x8_sharp vp8_eighttap_predict8x8_sharp_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap8x8_sharp);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap_avg16x16_sharp
|
|
|
|
|
#define vp8_subpix_eighttap_avg16x16_sharp vp8_eighttap_predict_avg16x16_sharp_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap_avg16x16_sharp);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap_avg8x8_sharp
|
|
|
|
|
#define vp8_subpix_eighttap_avg8x8_sharp vp8_eighttap_predict_avg8x8_sharp_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap_avg8x8_sharp);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap8x4_sharp
|
|
|
|
|
#define vp8_subpix_eighttap8x4_sharp vp8_eighttap_predict8x4_sharp_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap8x4_sharp);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_eighttap4x4_sharp
|
|
|
|
|
#define vp8_subpix_eighttap4x4_sharp vp8_eighttap_predict_sharp_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_eighttap4x4_sharp);
|
|
|
|
|
#endif /* CONFIG_ENAHNCED_INTERP */
|
|
|
|
|
|
2010-05-18 17:58:33 +02:00
|
|
|
#ifndef vp8_subpix_bilinear16x16
|
|
|
|
|
#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_bilinear16x16);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_bilinear8x8
|
|
|
|
|
#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_bilinear8x8);
|
|
|
|
|
|
Dual 16x16 inter prediction.
This patch introduces the concept of dual inter16x16 prediction. A
16x16 inter-predicted macroblock can use 2 references instead of 1,
where both references use the same mvmode (new, near/est, zero). In the
case of newmv, this means that two MVs are coded instead of one. The
frame can be encoded in 3 ways: all MBs single-prediction, all MBs dual
prediction, or per-MB single/dual prediction selection ("hybrid"), in
which case a single bit is coded per-MB to indicate whether the MB uses
single or dual inter prediction.
In the future, we can (maybe?) get further gains by mixing this with
Adrian's 32x32 work, per-segment dual prediction settings, or adding
support for dual splitmv/8x8mv inter prediction.
Gain (on derf-set, CQ mode) is ~2.8% (SSIM) or ~3.6% (glb PSNR). Most
gain is at medium/high bitrates, but there's minor gains at low bitrates
also. Output was confirmed to match between encoder and decoder.
Note for optimization people: this patch introduces a 2nd version of
16x16/8x8 sixtap/bilin functions, which does an avg instead of a
store. They may want to look and make sure this is implemented to
their satisfaction so we can optimize it best in the future.
Change-ID: I59dc84b07cbb3ccf073ac0f756d03d294cb19281
2011-12-06 20:53:02 +01:00
|
|
|
#ifndef vp8_subpix_bilinear_avg16x16
|
|
|
|
|
#define vp8_subpix_bilinear_avg16x16 vp8_bilinear_predict_avg16x16_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_bilinear_avg16x16);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_bilinear_avg8x8
|
|
|
|
|
#define vp8_subpix_bilinear_avg8x8 vp8_bilinear_predict_avg8x8_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_bilinear_avg8x8);
|
|
|
|
|
|
2010-05-18 17:58:33 +02:00
|
|
|
#ifndef vp8_subpix_bilinear8x4
|
|
|
|
|
#define vp8_subpix_bilinear8x4 vp8_bilinear_predict8x4_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_bilinear8x4);
|
|
|
|
|
|
|
|
|
|
#ifndef vp8_subpix_bilinear4x4
|
|
|
|
|
#define vp8_subpix_bilinear4x4 vp8_bilinear_predict4x4_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_bilinear4x4);
|
|
|
|
|
|
2012-04-18 22:51:58 +02:00
|
|
|
#ifndef vp8_subpix_bilinear_avg4x4
|
|
|
|
|
#define vp8_subpix_bilinear_avg4x4 vp8_bilinear_predict_avg4x4_c
|
|
|
|
|
#endif
|
|
|
|
|
extern prototype_subpixel_predict(vp8_subpix_bilinear_avg4x4);
|
|
|
|
|
|
2010-05-18 17:58:33 +02:00
|
|
|
typedef prototype_subpixel_predict((*vp8_subpix_fn_t));
|
|
|
|
|
typedef struct
|
|
|
|
|
{
|
2012-03-19 15:53:05 +01:00
|
|
|
#if CONFIG_ENHANCED_INTERP
|
|
|
|
|
vp8_subpix_fn_t eighttap16x16;
|
|
|
|
|
vp8_subpix_fn_t eighttap8x8;
|
|
|
|
|
vp8_subpix_fn_t eighttap_avg16x16;
|
|
|
|
|
vp8_subpix_fn_t eighttap_avg8x8;
|
|
|
|
|
vp8_subpix_fn_t eighttap8x4;
|
|
|
|
|
vp8_subpix_fn_t eighttap4x4;
|
|
|
|
|
vp8_subpix_fn_t eighttap16x16_sharp;
|
|
|
|
|
vp8_subpix_fn_t eighttap8x8_sharp;
|
|
|
|
|
vp8_subpix_fn_t eighttap_avg16x16_sharp;
|
|
|
|
|
vp8_subpix_fn_t eighttap_avg8x8_sharp;
|
|
|
|
|
vp8_subpix_fn_t eighttap8x4_sharp;
|
|
|
|
|
vp8_subpix_fn_t eighttap4x4_sharp;
|
|
|
|
|
#endif
|
2010-05-18 17:58:33 +02:00
|
|
|
vp8_subpix_fn_t sixtap16x16;
|
|
|
|
|
vp8_subpix_fn_t sixtap8x8;
|
Dual 16x16 inter prediction.
This patch introduces the concept of dual inter16x16 prediction. A
16x16 inter-predicted macroblock can use 2 references instead of 1,
where both references use the same mvmode (new, near/est, zero). In the
case of newmv, this means that two MVs are coded instead of one. The
frame can be encoded in 3 ways: all MBs single-prediction, all MBs dual
prediction, or per-MB single/dual prediction selection ("hybrid"), in
which case a single bit is coded per-MB to indicate whether the MB uses
single or dual inter prediction.
In the future, we can (maybe?) get further gains by mixing this with
Adrian's 32x32 work, per-segment dual prediction settings, or adding
support for dual splitmv/8x8mv inter prediction.
Gain (on derf-set, CQ mode) is ~2.8% (SSIM) or ~3.6% (glb PSNR). Most
gain is at medium/high bitrates, but there's minor gains at low bitrates
also. Output was confirmed to match between encoder and decoder.
Note for optimization people: this patch introduces a 2nd version of
16x16/8x8 sixtap/bilin functions, which does an avg instead of a
store. They may want to look and make sure this is implemented to
their satisfaction so we can optimize it best in the future.
Change-ID: I59dc84b07cbb3ccf073ac0f756d03d294cb19281
2011-12-06 20:53:02 +01:00
|
|
|
vp8_subpix_fn_t sixtap_avg16x16;
|
|
|
|
|
vp8_subpix_fn_t sixtap_avg8x8;
|
2010-05-18 17:58:33 +02:00
|
|
|
vp8_subpix_fn_t sixtap8x4;
|
|
|
|
|
vp8_subpix_fn_t sixtap4x4;
|
2012-04-18 22:51:58 +02:00
|
|
|
vp8_subpix_fn_t sixtap_avg4x4;
|
2010-05-18 17:58:33 +02:00
|
|
|
vp8_subpix_fn_t bilinear16x16;
|
|
|
|
|
vp8_subpix_fn_t bilinear8x8;
|
Dual 16x16 inter prediction.
This patch introduces the concept of dual inter16x16 prediction. A
16x16 inter-predicted macroblock can use 2 references instead of 1,
where both references use the same mvmode (new, near/est, zero). In the
case of newmv, this means that two MVs are coded instead of one. The
frame can be encoded in 3 ways: all MBs single-prediction, all MBs dual
prediction, or per-MB single/dual prediction selection ("hybrid"), in
which case a single bit is coded per-MB to indicate whether the MB uses
single or dual inter prediction.
In the future, we can (maybe?) get further gains by mixing this with
Adrian's 32x32 work, per-segment dual prediction settings, or adding
support for dual splitmv/8x8mv inter prediction.
Gain (on derf-set, CQ mode) is ~2.8% (SSIM) or ~3.6% (glb PSNR). Most
gain is at medium/high bitrates, but there's minor gains at low bitrates
also. Output was confirmed to match between encoder and decoder.
Note for optimization people: this patch introduces a 2nd version of
16x16/8x8 sixtap/bilin functions, which does an avg instead of a
store. They may want to look and make sure this is implemented to
their satisfaction so we can optimize it best in the future.
Change-ID: I59dc84b07cbb3ccf073ac0f756d03d294cb19281
2011-12-06 20:53:02 +01:00
|
|
|
vp8_subpix_fn_t bilinear_avg16x16;
|
|
|
|
|
vp8_subpix_fn_t bilinear_avg8x8;
|
2010-05-18 17:58:33 +02:00
|
|
|
vp8_subpix_fn_t bilinear8x4;
|
|
|
|
|
vp8_subpix_fn_t bilinear4x4;
|
2012-04-18 22:51:58 +02:00
|
|
|
vp8_subpix_fn_t bilinear_avg4x4;
|
2010-05-18 17:58:33 +02:00
|
|
|
} vp8_subpix_rtcd_vtable_t;
|
|
|
|
|
|
|
|
|
|
#if CONFIG_RUNTIME_CPU_DETECT
|
|
|
|
|
#define SUBPIX_INVOKE(ctx,fn) (ctx)->fn
|
|
|
|
|
#else
|
|
|
|
|
#define SUBPIX_INVOKE(ctx,fn) vp8_subpix_##fn
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#endif
|