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
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
|
// Copyright 2010 Google Inc. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the COPYING file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
// -----------------------------------------------------------------------------
//
// YUV->RGB conversion functions
//
// Author: Skal (pascal.massimino@gmail.com)
#include "src/dsp/yuv.h"
#include <assert.h>
#include <stdlib.h>
//-----------------------------------------------------------------------------
// Plain-C version
#define ROW_FUNC(FUNC_NAME, FUNC, XSTEP) \
static void FUNC_NAME(const uint8_t* y, \
const uint8_t* u, const uint8_t* v, \
uint8_t* dst, int len) { \
const uint8_t* const end = dst + (len & ~1) * (XSTEP); \
while (dst != end) { \
FUNC(y[0], u[0], v[0], dst); \
FUNC(y[1], u[0], v[0], dst + (XSTEP)); \
y += 2; \
++u; \
++v; \
dst += 2 * (XSTEP); \
} \
if (len & 1) { \
FUNC(y[0], u[0], v[0], dst); \
} \
} \
// All variants implemented.
ROW_FUNC(YuvToRgbRow, VP8YuvToRgb, 3)
ROW_FUNC(YuvToBgrRow, VP8YuvToBgr, 3)
ROW_FUNC(YuvToRgbaRow, VP8YuvToRgba, 4)
ROW_FUNC(YuvToBgraRow, VP8YuvToBgra, 4)
ROW_FUNC(YuvToArgbRow, VP8YuvToArgb, 4)
ROW_FUNC(YuvToRgba4444Row, VP8YuvToRgba4444, 2)
ROW_FUNC(YuvToRgb565Row, VP8YuvToRgb565, 2)
#undef ROW_FUNC
// Main call for processing a plane with a WebPSamplerRowFunc function:
void WebPSamplerProcessPlane(const uint8_t* y, int y_stride,
const uint8_t* u, const uint8_t* v, int uv_stride,
uint8_t* dst, int dst_stride,
int width, int height, WebPSamplerRowFunc func) {
int j;
for (j = 0; j < height; ++j) {
func(y, u, v, dst, width);
y += y_stride;
if (j & 1) {
u += uv_stride;
v += uv_stride;
}
dst += dst_stride;
}
}
//-----------------------------------------------------------------------------
// Main call
WebPSamplerRowFunc WebPSamplers[MODE_LAST];
extern void WebPInitSamplersSSE2(void);
extern void WebPInitSamplersMIPS32(void);
extern void WebPInitSamplersMIPSdspR2(void);
static volatile VP8CPUInfo yuv_last_cpuinfo_used =
(VP8CPUInfo)&yuv_last_cpuinfo_used;
WEBP_TSAN_IGNORE_FUNCTION void WebPInitSamplers(void) {
if (yuv_last_cpuinfo_used == VP8GetCPUInfo) return;
WebPSamplers[MODE_RGB] = YuvToRgbRow;
WebPSamplers[MODE_RGBA] = YuvToRgbaRow;
WebPSamplers[MODE_BGR] = YuvToBgrRow;
WebPSamplers[MODE_BGRA] = YuvToBgraRow;
WebPSamplers[MODE_ARGB] = YuvToArgbRow;
WebPSamplers[MODE_RGBA_4444] = YuvToRgba4444Row;
WebPSamplers[MODE_RGB_565] = YuvToRgb565Row;
WebPSamplers[MODE_rgbA] = YuvToRgbaRow;
WebPSamplers[MODE_bgrA] = YuvToBgraRow;
WebPSamplers[MODE_Argb] = YuvToArgbRow;
WebPSamplers[MODE_rgbA_4444] = YuvToRgba4444Row;
// If defined, use CPUInfo() to overwrite some pointers with faster versions.
if (VP8GetCPUInfo != NULL) {
#if defined(WEBP_USE_SSE2)
if (VP8GetCPUInfo(kSSE2)) {
WebPInitSamplersSSE2();
}
#endif // WEBP_USE_SSE2
#if defined(WEBP_USE_MIPS32)
if (VP8GetCPUInfo(kMIPS32)) {
WebPInitSamplersMIPS32();
}
#endif // WEBP_USE_MIPS32
#if defined(WEBP_USE_MIPS_DSP_R2)
if (VP8GetCPUInfo(kMIPSdspR2)) {
WebPInitSamplersMIPSdspR2();
}
#endif // WEBP_USE_MIPS_DSP_R2
}
yuv_last_cpuinfo_used = VP8GetCPUInfo;
}
//-----------------------------------------------------------------------------
// ARGB -> YUV converters
static void ConvertARGBToY_C(const uint32_t* argb, uint8_t* y, int width) {
int i;
for (i = 0; i < width; ++i) {
const uint32_t p = argb[i];
y[i] = VP8RGBToY((p >> 16) & 0xff, (p >> 8) & 0xff, (p >> 0) & 0xff,
YUV_HALF);
}
}
void WebPConvertARGBToUV_C(const uint32_t* argb, uint8_t* u, uint8_t* v,
int src_width, int do_store) {
// No rounding. Last pixel is dealt with separately.
const int uv_width = src_width >> 1;
int i;
for (i = 0; i < uv_width; ++i) {
const uint32_t v0 = argb[2 * i + 0];
const uint32_t v1 = argb[2 * i + 1];
// VP8RGBToU/V expects four accumulated pixels. Hence we need to
// scale r/g/b value by a factor 2. We just shift v0/v1 one bit less.
const int r = ((v0 >> 15) & 0x1fe) + ((v1 >> 15) & 0x1fe);
const int g = ((v0 >> 7) & 0x1fe) + ((v1 >> 7) & 0x1fe);
const int b = ((v0 << 1) & 0x1fe) + ((v1 << 1) & 0x1fe);
const int tmp_u = VP8RGBToU(r, g, b, YUV_HALF << 2);
const int tmp_v = VP8RGBToV(r, g, b, YUV_HALF << 2);
if (do_store) {
u[i] = tmp_u;
v[i] = tmp_v;
} else {
// Approximated average-of-four. But it's an acceptable diff.
u[i] = (u[i] + tmp_u + 1) >> 1;
v[i] = (v[i] + tmp_v + 1) >> 1;
}
}
if (src_width & 1) { // last pixel
const uint32_t v0 = argb[2 * i + 0];
const int r = (v0 >> 14) & 0x3fc;
const int g = (v0 >> 6) & 0x3fc;
const int b = (v0 << 2) & 0x3fc;
const int tmp_u = VP8RGBToU(r, g, b, YUV_HALF << 2);
const int tmp_v = VP8RGBToV(r, g, b, YUV_HALF << 2);
if (do_store) {
u[i] = tmp_u;
v[i] = tmp_v;
} else {
u[i] = (u[i] + tmp_u + 1) >> 1;
v[i] = (v[i] + tmp_v + 1) >> 1;
}
}
}
//-----------------------------------------------------------------------------
static void ConvertRGB24ToY_C(const uint8_t* rgb, uint8_t* y, int width) {
int i;
for (i = 0; i < width; ++i, rgb += 3) {
y[i] = VP8RGBToY(rgb[0], rgb[1], rgb[2], YUV_HALF);
}
}
static void ConvertBGR24ToY_C(const uint8_t* bgr, uint8_t* y, int width) {
int i;
for (i = 0; i < width; ++i, bgr += 3) {
y[i] = VP8RGBToY(bgr[2], bgr[1], bgr[0], YUV_HALF);
}
}
void WebPConvertRGBA32ToUV_C(const uint16_t* rgb,
uint8_t* u, uint8_t* v, int width) {
int i;
for (i = 0; i < width; i += 1, rgb += 4) {
const int r = rgb[0], g = rgb[1], b = rgb[2];
u[i] = VP8RGBToU(r, g, b, YUV_HALF << 2);
v[i] = VP8RGBToV(r, g, b, YUV_HALF << 2);
}
}
//-----------------------------------------------------------------------------
#if !WEBP_NEON_OMIT_C_CODE
#define MAX_Y ((1 << 10) - 1) // 10b precision over 16b-arithmetic
static uint16_t clip_y(int v) {
return (v < 0) ? 0 : (v > MAX_Y) ? MAX_Y : (uint16_t)v;
}
static uint64_t SharpYUVUpdateY_C(const uint16_t* ref, const uint16_t* src,
uint16_t* dst, int len) {
uint64_t diff = 0;
int i;
for (i = 0; i < len; ++i) {
const int diff_y = ref[i] - src[i];
const int new_y = (int)dst[i] + diff_y;
dst[i] = clip_y(new_y);
diff += (uint64_t)abs(diff_y);
}
return diff;
}
static void SharpYUVUpdateRGB_C(const int16_t* ref, const int16_t* src,
int16_t* dst, int len) {
int i;
for (i = 0; i < len; ++i) {
const int diff_uv = ref[i] - src[i];
dst[i] += diff_uv;
}
}
static void SharpYUVFilterRow_C(const int16_t* A, const int16_t* B, int len,
const uint16_t* best_y, uint16_t* out) {
int i;
for (i = 0; i < len; ++i, ++A, ++B) {
const int v0 = (A[0] * 9 + A[1] * 3 + B[0] * 3 + B[1] + 8) >> 4;
const int v1 = (A[1] * 9 + A[0] * 3 + B[1] * 3 + B[0] + 8) >> 4;
out[2 * i + 0] = clip_y(best_y[2 * i + 0] + v0);
out[2 * i + 1] = clip_y(best_y[2 * i + 1] + v1);
}
}
#endif // !WEBP_NEON_OMIT_C_CODE
#undef MAX_Y
//-----------------------------------------------------------------------------
void (*WebPConvertRGB24ToY)(const uint8_t* rgb, uint8_t* y, int width);
void (*WebPConvertBGR24ToY)(const uint8_t* bgr, uint8_t* y, int width);
void (*WebPConvertRGBA32ToUV)(const uint16_t* rgb,
uint8_t* u, uint8_t* v, int width);
void (*WebPConvertARGBToY)(const uint32_t* argb, uint8_t* y, int width);
void (*WebPConvertARGBToUV)(const uint32_t* argb, uint8_t* u, uint8_t* v,
int src_width, int do_store);
uint64_t (*WebPSharpYUVUpdateY)(const uint16_t* ref, const uint16_t* src,
uint16_t* dst, int len);
void (*WebPSharpYUVUpdateRGB)(const int16_t* ref, const int16_t* src,
int16_t* dst, int len);
void (*WebPSharpYUVFilterRow)(const int16_t* A, const int16_t* B, int len,
const uint16_t* best_y, uint16_t* out);
static volatile VP8CPUInfo rgba_to_yuv_last_cpuinfo_used =
(VP8CPUInfo)&rgba_to_yuv_last_cpuinfo_used;
extern void WebPInitConvertARGBToYUVSSE2(void);
extern void WebPInitConvertARGBToYUVNEON(void);
extern void WebPInitSharpYUVSSE2(void);
extern void WebPInitSharpYUVNEON(void);
WEBP_TSAN_IGNORE_FUNCTION void WebPInitConvertARGBToYUV(void) {
if (rgba_to_yuv_last_cpuinfo_used == VP8GetCPUInfo) return;
WebPConvertARGBToY = ConvertARGBToY_C;
WebPConvertARGBToUV = WebPConvertARGBToUV_C;
WebPConvertRGB24ToY = ConvertRGB24ToY_C;
WebPConvertBGR24ToY = ConvertBGR24ToY_C;
WebPConvertRGBA32ToUV = WebPConvertRGBA32ToUV_C;
#if !WEBP_NEON_OMIT_C_CODE
WebPSharpYUVUpdateY = SharpYUVUpdateY_C;
WebPSharpYUVUpdateRGB = SharpYUVUpdateRGB_C;
WebPSharpYUVFilterRow = SharpYUVFilterRow_C;
#endif
if (VP8GetCPUInfo != NULL) {
#if defined(WEBP_USE_SSE2)
if (VP8GetCPUInfo(kSSE2)) {
WebPInitConvertARGBToYUVSSE2();
WebPInitSharpYUVSSE2();
}
#endif // WEBP_USE_SSE2
}
#if defined(WEBP_USE_NEON)
if (WEBP_NEON_OMIT_C_CODE ||
(VP8GetCPUInfo != NULL && VP8GetCPUInfo(kNEON))) {
WebPInitConvertARGBToYUVNEON();
WebPInitSharpYUVNEON();
}
#endif // WEBP_USE_NEON
assert(WebPConvertARGBToY != NULL);
assert(WebPConvertARGBToUV != NULL);
assert(WebPConvertRGB24ToY != NULL);
assert(WebPConvertBGR24ToY != NULL);
assert(WebPConvertRGBA32ToUV != NULL);
assert(WebPSharpYUVUpdateY != NULL);
assert(WebPSharpYUVUpdateRGB != NULL);
assert(WebPSharpYUVFilterRow != NULL);
rgba_to_yuv_last_cpuinfo_used = VP8GetCPUInfo;
}
|