1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
|
// Copyright (c) 2010 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
// yuv_row internal functions to handle YUV conversion and scaling to RGB.
// These functions are used from both yuv_convert.cc and yuv_scale.cc.
// TODO(fbarchard): Write function that can handle rotation and scaling.
#ifndef MEDIA_BASE_YUV_ROW_H_
#define MEDIA_BASE_YUV_ROW_H_
#include "chromium_types.h"
extern "C" {
// Can only do 1x.
// This is the second fastest of the scalers.
void FastConvertYUVToRGB32Row(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width);
void FastConvertYUVToRGB32Row_C(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
unsigned int x_shift);
void FastConvertYUVToRGB32Row(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width);
// Can do 1x, half size or any scale down by an integer amount.
// Step can be negative (mirroring, rotate 180).
// This is the third fastest of the scalers.
// Only defined on Windows x86-32.
void ConvertYUVToRGB32Row_SSE(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int step);
// Rotate is like Convert, but applies different step to Y versus U and V.
// This allows rotation by 90 or 270, by stepping by stride.
// This is the forth fastest of the scalers.
// Only defined on Windows x86-32.
void RotateConvertYUVToRGB32Row_SSE(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int ystep,
int uvstep);
// Doubler does 4 pixels at a time. Each pixel is replicated.
// This is the fastest of the scalers.
// Only defined on Windows x86-32.
void DoubleYUVToRGB32Row_SSE(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width);
// Handles arbitrary scaling up or down.
// Mirroring is supported, but not 90 or 270 degree rotation.
// Chroma is under sampled every 2 pixels for performance.
void ScaleYUVToRGB32Row(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int source_dx);
void ScaleYUVToRGB32Row(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int source_dx);
void ScaleYUVToRGB32Row_C(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int source_dx);
// Handles arbitrary scaling up or down with bilinear filtering.
// Mirroring is supported, but not 90 or 270 degree rotation.
// Chroma is under sampled every 2 pixels for performance.
// This is the slowest of the scalers.
void LinearScaleYUVToRGB32Row(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int source_dx);
void LinearScaleYUVToRGB32Row(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int source_dx);
void LinearScaleYUVToRGB32Row_C(const uint8_t* y_buf,
const uint8_t* u_buf,
const uint8_t* v_buf,
uint8_t* rgb_buf,
int width,
int source_dx);
#if defined(_MSC_VER) && !defined(__CLR_VER) && !defined(__clang__)
#if defined(VISUALC_HAS_AVX2)
#define SIMD_ALIGNED(var) __declspec(align(32)) var
#else
#define SIMD_ALIGNED(var) __declspec(align(16)) var
#endif
#elif defined(__GNUC__) || defined(__clang__)
// Caveat GCC 4.2 to 4.7 have a known issue using vectors with const.
#if defined(CLANG_HAS_AVX2) || defined(GCC_HAS_AVX2)
#define SIMD_ALIGNED(var) var __attribute__((aligned(32)))
#else
#define SIMD_ALIGNED(var) var __attribute__((aligned(16)))
#endif
#else
#define SIMD_ALIGNED(var) var
#endif
extern SIMD_ALIGNED(const int16_t kCoefficientsRgbY[768][4]);
// x64 uses MMX2 (SSE) so emms is not required.
// Warning C4799: function has no EMMS instruction.
// EMMS() is slow and should be called by the calling function once per image.
#if defined(ARCH_CPU_X86) && !defined(ARCH_CPU_X86_64)
#if defined(_MSC_VER)
#define EMMS() __asm emms
#pragma warning(disable: 4799)
#else
#define EMMS() asm("emms")
#endif
#else
#define EMMS() ((void)0)
#endif
} // extern "C"
#endif // MEDIA_BASE_YUV_ROW_H_
|
