/src/libavif/ext/libyuv/source/convert.cc
Line | Count | Source |
1 | | /* |
2 | | * Copyright 2011 The LibYuv Project Authors. All rights reserved. |
3 | | * |
4 | | * Use of this source code is governed by a BSD-style license |
5 | | * that can be found in the LICENSE file in the root of the source |
6 | | * tree. An additional intellectual property rights grant can be found |
7 | | * in the file PATENTS. All contributing project authors may |
8 | | * be found in the AUTHORS file in the root of the source tree. |
9 | | */ |
10 | | |
11 | | #include "libyuv/convert.h" |
12 | | |
13 | | #include <limits.h> |
14 | | |
15 | | #include "libyuv/basic_types.h" |
16 | | #include "libyuv/convert_from_argb.h" |
17 | | #include "libyuv/cpu_id.h" |
18 | | #include "libyuv/planar_functions.h" |
19 | | #include "libyuv/rotate.h" |
20 | | #include "libyuv/row.h" |
21 | | #include "libyuv/scale.h" // For ScalePlane() |
22 | | #include "libyuv/scale_row.h" // For FixedDiv |
23 | | #include "libyuv/scale_uv.h" // For UVScale() |
24 | | |
25 | | #ifdef __cplusplus |
26 | | namespace libyuv { |
27 | | |
28 | | extern const struct ArgbConstants kArgbI601Constants; |
29 | | extern const struct ArgbConstants kArgbJPEGConstants; |
30 | | extern "C" { |
31 | | #endif |
32 | | |
33 | | // Subsample amount uses a shift. |
34 | | // v is value |
35 | | // a is amount to add to round up |
36 | | // s is shift to subsample down |
37 | 0 | #define SUBSAMPLE(v, a, s) (v < 0) ? (-((-v + a) >> s)) : ((v + a) >> s) |
38 | 0 | static __inline int Abs(int v) { |
39 | 0 | return v >= 0 ? v : -v; |
40 | 0 | } |
41 | | |
42 | | // Any I4xx To I420 format |
43 | | static int I4xxToI420(const uint8_t* src_y, |
44 | | int src_stride_y, |
45 | | const uint8_t* src_u, |
46 | | int src_stride_u, |
47 | | const uint8_t* src_v, |
48 | | int src_stride_v, |
49 | | uint8_t* dst_y, |
50 | | int dst_stride_y, |
51 | | uint8_t* dst_u, |
52 | | int dst_stride_u, |
53 | | uint8_t* dst_v, |
54 | | int dst_stride_v, |
55 | | int src_y_width, |
56 | | int src_y_height, |
57 | | int src_uv_width, |
58 | 0 | int src_uv_height) { |
59 | 0 | int r; |
60 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || |
61 | 0 | src_y_width <= 0 || src_y_height == 0 || src_y_height == INT_MIN || |
62 | 0 | src_uv_width <= 0 || src_uv_height == 0) { |
63 | 0 | return -1; |
64 | 0 | } |
65 | 0 | const int dst_y_width = src_y_width; |
66 | 0 | const int dst_y_height = Abs(src_y_height); |
67 | 0 | const int dst_uv_width = SUBSAMPLE(dst_y_width, 1, 1); |
68 | 0 | const int dst_uv_height = SUBSAMPLE(dst_y_height, 1, 1); |
69 | 0 | if (dst_y) { |
70 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, src_y_width, |
71 | 0 | src_y_height); |
72 | 0 | } |
73 | 0 | r = ScalePlane(src_u, src_stride_u, src_uv_width, src_uv_height, dst_u, |
74 | 0 | dst_stride_u, dst_uv_width, dst_uv_height, kFilterBilinear); |
75 | 0 | if (r != 0) { |
76 | 0 | return r; |
77 | 0 | } |
78 | 0 | r = ScalePlane(src_v, src_stride_v, src_uv_width, src_uv_height, dst_v, |
79 | 0 | dst_stride_v, dst_uv_width, dst_uv_height, kFilterBilinear); |
80 | 0 | return r; |
81 | 0 | } |
82 | | |
83 | | // Copy I420 with optional vertical flipping using negative height. |
84 | | LIBYUV_API |
85 | | int I420Copy(const uint8_t* src_y, |
86 | | int src_stride_y, |
87 | | const uint8_t* src_u, |
88 | | int src_stride_u, |
89 | | const uint8_t* src_v, |
90 | | int src_stride_v, |
91 | | uint8_t* dst_y, |
92 | | int dst_stride_y, |
93 | | uint8_t* dst_u, |
94 | | int dst_stride_u, |
95 | | uint8_t* dst_v, |
96 | | int dst_stride_v, |
97 | | int width, |
98 | 0 | int height) { |
99 | 0 | int halfwidth = (width + 1) >> 1; |
100 | 0 | int halfheight = (height + 1) >> 1; |
101 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
102 | 0 | height == 0 || height == INT_MIN) { |
103 | 0 | return -1; |
104 | 0 | } |
105 | | // Negative height means invert the image. |
106 | 0 | if (height < 0) { |
107 | 0 | height = -height; |
108 | 0 | halfheight = (height + 1) >> 1; |
109 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
110 | 0 | src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u; |
111 | 0 | src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v; |
112 | 0 | src_stride_y = -src_stride_y; |
113 | 0 | src_stride_u = -src_stride_u; |
114 | 0 | src_stride_v = -src_stride_v; |
115 | 0 | } |
116 | |
|
117 | 0 | if (dst_y) { |
118 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
119 | 0 | } |
120 | | // Copy UV planes. |
121 | 0 | CopyPlane(src_u, src_stride_u, dst_u, dst_stride_u, halfwidth, halfheight); |
122 | 0 | CopyPlane(src_v, src_stride_v, dst_v, dst_stride_v, halfwidth, halfheight); |
123 | 0 | return 0; |
124 | 0 | } |
125 | | |
126 | | // Copy I010 with optional flipping. |
127 | | LIBYUV_API |
128 | | int I010Copy(const uint16_t* src_y, |
129 | | int src_stride_y, |
130 | | const uint16_t* src_u, |
131 | | int src_stride_u, |
132 | | const uint16_t* src_v, |
133 | | int src_stride_v, |
134 | | uint16_t* dst_y, |
135 | | int dst_stride_y, |
136 | | uint16_t* dst_u, |
137 | | int dst_stride_u, |
138 | | uint16_t* dst_v, |
139 | | int dst_stride_v, |
140 | | int width, |
141 | 0 | int height) { |
142 | 0 | int halfwidth = (width + 1) >> 1; |
143 | 0 | int halfheight = (height + 1) >> 1; |
144 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
145 | 0 | height == 0 || height == INT_MIN) { |
146 | 0 | return -1; |
147 | 0 | } |
148 | | // Negative height means invert the image. |
149 | 0 | if (height < 0) { |
150 | 0 | height = -height; |
151 | 0 | halfheight = (height + 1) >> 1; |
152 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
153 | 0 | src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u; |
154 | 0 | src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v; |
155 | 0 | src_stride_y = -src_stride_y; |
156 | 0 | src_stride_u = -src_stride_u; |
157 | 0 | src_stride_v = -src_stride_v; |
158 | 0 | } |
159 | |
|
160 | 0 | if (dst_y) { |
161 | 0 | CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
162 | 0 | } |
163 | | // Copy UV planes. |
164 | 0 | CopyPlane_16(src_u, src_stride_u, dst_u, dst_stride_u, halfwidth, halfheight); |
165 | 0 | CopyPlane_16(src_v, src_stride_v, dst_v, dst_stride_v, halfwidth, halfheight); |
166 | 0 | return 0; |
167 | 0 | } |
168 | | |
169 | | static int Planar16bitTo8bit(const uint16_t* src_y, |
170 | | int src_stride_y, |
171 | | const uint16_t* src_u, |
172 | | int src_stride_u, |
173 | | const uint16_t* src_v, |
174 | | int src_stride_v, |
175 | | uint8_t* dst_y, |
176 | | int dst_stride_y, |
177 | | uint8_t* dst_u, |
178 | | int dst_stride_u, |
179 | | uint8_t* dst_v, |
180 | | int dst_stride_v, |
181 | | int width, |
182 | | int height, |
183 | | int subsample_x, |
184 | | int subsample_y, |
185 | 0 | int depth) { |
186 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
187 | 0 | height == 0 || height == INT_MIN) { |
188 | 0 | return -1; |
189 | 0 | } |
190 | 0 | int uv_width = SUBSAMPLE(width, subsample_x, subsample_x); |
191 | 0 | int uv_height = SUBSAMPLE(height, subsample_y, subsample_y); |
192 | 0 | int scale = 1 << (24 - depth); |
193 | | // Negative height means invert the image. |
194 | 0 | if (height < 0) { |
195 | 0 | height = -height; |
196 | 0 | uv_height = -uv_height; |
197 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
198 | 0 | src_u = src_u + (ptrdiff_t)(uv_height - 1) * src_stride_u; |
199 | 0 | src_v = src_v + (ptrdiff_t)(uv_height - 1) * src_stride_v; |
200 | 0 | src_stride_y = -src_stride_y; |
201 | 0 | src_stride_u = -src_stride_u; |
202 | 0 | src_stride_v = -src_stride_v; |
203 | 0 | } |
204 | | |
205 | | // Convert Y plane. |
206 | 0 | if (dst_y) { |
207 | 0 | Convert16To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale, width, |
208 | 0 | height); |
209 | 0 | } |
210 | | // Convert UV planes. |
211 | 0 | Convert16To8Plane(src_u, src_stride_u, dst_u, dst_stride_u, scale, uv_width, |
212 | 0 | uv_height); |
213 | 0 | Convert16To8Plane(src_v, src_stride_v, dst_v, dst_stride_v, scale, uv_width, |
214 | 0 | uv_height); |
215 | 0 | return 0; |
216 | 0 | } |
217 | | |
218 | | static int I41xToI420(const uint16_t* src_y, |
219 | | int src_stride_y, |
220 | | const uint16_t* src_u, |
221 | | int src_stride_u, |
222 | | const uint16_t* src_v, |
223 | | int src_stride_v, |
224 | | uint8_t* dst_y, |
225 | | int dst_stride_y, |
226 | | uint8_t* dst_u, |
227 | | int dst_stride_u, |
228 | | uint8_t* dst_v, |
229 | | int dst_stride_v, |
230 | | int width, |
231 | | int height, |
232 | 0 | int depth) { |
233 | 0 | int y; |
234 | 0 | const int scale = 1 << (24 - depth); |
235 | 0 | const int uv_width = SUBSAMPLE(width, 1, 1); |
236 | 0 | void (*Convert16To8Row)(const uint16_t* src_y, uint8_t* dst_y, int scale, |
237 | 0 | int width) = Convert16To8Row_C; |
238 | 0 | void (*HalfWidthRow_16To8)(const uint16_t* src_uv, ptrdiff_t src_uv_stride, |
239 | 0 | uint8_t* dst_uv, int scale, int width) = |
240 | 0 | (width & 1) ? HalfWidthRow_16To8_Odd_C : HalfWidthRow_16To8_C; |
241 | |
|
242 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
243 | 0 | height == 0 || height == INT_MIN) { |
244 | 0 | return -1; |
245 | 0 | } |
246 | | // Negative height means invert the image. |
247 | 0 | if (height < 0) { |
248 | 0 | height = -height; |
249 | 0 | if (src_y) { |
250 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
251 | 0 | src_stride_y = -src_stride_y; |
252 | 0 | } |
253 | 0 | src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u; |
254 | 0 | src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v; |
255 | 0 | src_stride_u = -src_stride_u; |
256 | 0 | src_stride_v = -src_stride_v; |
257 | 0 | } |
258 | |
|
259 | | #if defined(HAS_CONVERT16TO8ROW_NEON) |
260 | | if (TestCpuFlag(kCpuHasNEON)) { |
261 | | Convert16To8Row = Convert16To8Row_Any_NEON; |
262 | | if (IS_ALIGNED(width, 16)) { |
263 | | Convert16To8Row = Convert16To8Row_NEON; |
264 | | } |
265 | | } |
266 | | #endif |
267 | | #if defined(HAS_CONVERT16TO8ROW_SVE2) |
268 | | if (TestCpuFlag(kCpuHasSVE2)) { |
269 | | Convert16To8Row = Convert16To8Row_SVE2; |
270 | | } |
271 | | #endif |
272 | | #if defined(HAS_CONVERT16TO8ROW_SME) |
273 | | if (TestCpuFlag(kCpuHasSME)) { |
274 | | Convert16To8Row = Convert16To8Row_SME; |
275 | | } |
276 | | #endif |
277 | 0 | #if defined(HAS_CONVERT16TO8ROW_SSSE3) |
278 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
279 | 0 | Convert16To8Row = Convert16To8Row_Any_SSSE3; |
280 | 0 | if (IS_ALIGNED(width, 16)) { |
281 | 0 | Convert16To8Row = Convert16To8Row_SSSE3; |
282 | 0 | } |
283 | 0 | } |
284 | 0 | #endif |
285 | 0 | #if defined(HAS_CONVERT16TO8ROW_AVX2) |
286 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
287 | 0 | Convert16To8Row = Convert16To8Row_Any_AVX2; |
288 | 0 | if (IS_ALIGNED(width, 32)) { |
289 | 0 | Convert16To8Row = Convert16To8Row_AVX2; |
290 | 0 | } |
291 | 0 | } |
292 | 0 | #endif |
293 | 0 | #if defined(HAS_CONVERT16TO8ROW_AVX512BW) |
294 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
295 | 0 | Convert16To8Row = Convert16To8Row_Any_AVX512BW; |
296 | 0 | if (IS_ALIGNED(width, 64)) { |
297 | 0 | Convert16To8Row = Convert16To8Row_AVX512BW; |
298 | 0 | } |
299 | 0 | } |
300 | 0 | #endif |
301 | | #if defined(HAS_CONVERT16TO8ROW_RVV) |
302 | | if (TestCpuFlag(kCpuHasRVV)) { |
303 | | Convert16To8Row = Convert16To8Row_RVV; |
304 | | } |
305 | | #endif |
306 | |
|
307 | 0 | if (IS_ALIGNED(width, 2)) { |
308 | | #if defined(HAS_HALFWIDTHROW_16TO8_NEON) |
309 | | if (TestCpuFlag(kCpuHasNEON)) { |
310 | | HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_NEON; |
311 | | if (IS_ALIGNED(uv_width, 8)) { |
312 | | HalfWidthRow_16To8 = HalfWidthRow_16To8_NEON; |
313 | | } |
314 | | } |
315 | | #endif |
316 | | #if defined(HAS_HALFWIDTHROW_16TO8_SVE2) |
317 | | if (TestCpuFlag(kCpuHasSVE2)) { |
318 | | HalfWidthRow_16To8 = HalfWidthRow_16To8_SVE2; |
319 | | } |
320 | | #endif |
321 | | #if defined(HAS_HALFWIDTHROW_16TO8_SME) |
322 | | if (TestCpuFlag(kCpuHasSME)) { |
323 | | HalfWidthRow_16To8 = HalfWidthRow_16To8_SME; |
324 | | } |
325 | | #endif |
326 | 0 | #if defined(HAS_HALFWIDTHROW_16TO8_SSSE3) |
327 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
328 | 0 | HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_SSSE3; |
329 | 0 | if (IS_ALIGNED(uv_width, 8)) { |
330 | 0 | HalfWidthRow_16To8 = HalfWidthRow_16To8_SSSE3; |
331 | 0 | } |
332 | 0 | } |
333 | 0 | #endif |
334 | 0 | #if defined(HAS_HALFWIDTHROW_16TO8_AVX2) |
335 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
336 | 0 | HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_AVX2; |
337 | 0 | if (IS_ALIGNED(uv_width, 16)) { |
338 | 0 | HalfWidthRow_16To8 = HalfWidthRow_16To8_AVX2; |
339 | 0 | } |
340 | 0 | } |
341 | 0 | #endif |
342 | 0 | #if defined(HAS_HALFWIDTHROW_16TO8_AVX512BW) |
343 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
344 | 0 | HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_AVX512BW; |
345 | 0 | if (IS_ALIGNED(uv_width, 32)) { |
346 | 0 | HalfWidthRow_16To8 = HalfWidthRow_16To8_AVX512BW; |
347 | 0 | } |
348 | 0 | } |
349 | 0 | #endif |
350 | | #if defined(HAS_HALFWIDTHROW_16TO8_RVV) |
351 | | if (TestCpuFlag(kCpuHasRVV)) { |
352 | | HalfWidthRow_16To8 = HalfWidthRow_16To8_RVV; |
353 | | } |
354 | | #endif |
355 | 0 | } |
356 | |
|
357 | 0 | for (y = 0; y < height - 1; y += 2) { |
358 | 0 | HalfWidthRow_16To8(src_u, src_stride_u, dst_u, scale, uv_width); |
359 | 0 | HalfWidthRow_16To8(src_v, src_stride_v, dst_v, scale, uv_width); |
360 | 0 | if (dst_y) { |
361 | 0 | Convert16To8Row(src_y, dst_y, scale, width); |
362 | 0 | Convert16To8Row(src_y + src_stride_y, dst_y + dst_stride_y, scale, width); |
363 | 0 | src_y += src_stride_y * 2; |
364 | 0 | dst_y += dst_stride_y * 2; |
365 | 0 | } |
366 | 0 | src_u += src_stride_u * 2; |
367 | 0 | src_v += src_stride_v * 2; |
368 | 0 | dst_u += dst_stride_u; |
369 | 0 | dst_v += dst_stride_v; |
370 | 0 | } |
371 | 0 | if (height & 1) { |
372 | 0 | HalfWidthRow_16To8(src_u, 0, dst_u, scale, uv_width); |
373 | 0 | HalfWidthRow_16To8(src_v, 0, dst_v, scale, uv_width); |
374 | 0 | if (dst_y) { |
375 | 0 | Convert16To8Row(src_y, dst_y, scale, width); |
376 | 0 | } |
377 | 0 | } |
378 | 0 | return 0; |
379 | 0 | } |
380 | | |
381 | | static int I21xToI420(const uint16_t* src_y, |
382 | | int src_stride_y, |
383 | | const uint16_t* src_u, |
384 | | int src_stride_u, |
385 | | const uint16_t* src_v, |
386 | | int src_stride_v, |
387 | | uint8_t* dst_y, |
388 | | int dst_stride_y, |
389 | | uint8_t* dst_u, |
390 | | int dst_stride_u, |
391 | | uint8_t* dst_v, |
392 | | int dst_stride_v, |
393 | | int width, |
394 | | int height, |
395 | 0 | int depth) { |
396 | 0 | int y; |
397 | 0 | const int scale = 1 << (24 - depth); |
398 | 0 | const int uv_width = SUBSAMPLE(width, 1, 1); |
399 | 0 | void (*Convert16To8Row)(const uint16_t* src_y, uint8_t* dst_y, int scale, |
400 | 0 | int width) = Convert16To8Row_C; |
401 | 0 | void (*HalfRow_16To8)(const uint16_t* src_uv, ptrdiff_t src_uv_stride, |
402 | 0 | uint8_t* dst_uv, int scale, int width) = |
403 | 0 | HalfRow_16To8_C; |
404 | |
|
405 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
406 | 0 | height == 0 || height == INT_MIN) { |
407 | 0 | return -1; |
408 | 0 | } |
409 | | // Negative height means invert the image. |
410 | 0 | if (height < 0) { |
411 | 0 | height = -height; |
412 | 0 | if (src_y) { |
413 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
414 | 0 | src_stride_y = -src_stride_y; |
415 | 0 | } |
416 | 0 | src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u; |
417 | 0 | src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v; |
418 | 0 | src_stride_u = -src_stride_u; |
419 | 0 | src_stride_v = -src_stride_v; |
420 | 0 | } |
421 | |
|
422 | | #if defined(HAS_CONVERT16TO8ROW_NEON) |
423 | | if (TestCpuFlag(kCpuHasNEON)) { |
424 | | Convert16To8Row = Convert16To8Row_Any_NEON; |
425 | | if (IS_ALIGNED(width, 16)) { |
426 | | Convert16To8Row = Convert16To8Row_NEON; |
427 | | } |
428 | | } |
429 | | #endif |
430 | | #if defined(HAS_CONVERT16TO8ROW_SVE2) |
431 | | if (TestCpuFlag(kCpuHasSVE2)) { |
432 | | Convert16To8Row = Convert16To8Row_SVE2; |
433 | | } |
434 | | #endif |
435 | | #if defined(HAS_CONVERT16TO8ROW_SME) |
436 | | if (TestCpuFlag(kCpuHasSME)) { |
437 | | Convert16To8Row = Convert16To8Row_SME; |
438 | | } |
439 | | #endif |
440 | 0 | #if defined(HAS_CONVERT16TO8ROW_SSSE3) |
441 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
442 | 0 | Convert16To8Row = Convert16To8Row_Any_SSSE3; |
443 | 0 | if (IS_ALIGNED(width, 16)) { |
444 | 0 | Convert16To8Row = Convert16To8Row_SSSE3; |
445 | 0 | } |
446 | 0 | } |
447 | 0 | #endif |
448 | 0 | #if defined(HAS_CONVERT16TO8ROW_AVX2) |
449 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
450 | 0 | Convert16To8Row = Convert16To8Row_Any_AVX2; |
451 | 0 | if (IS_ALIGNED(width, 32)) { |
452 | 0 | Convert16To8Row = Convert16To8Row_AVX2; |
453 | 0 | } |
454 | 0 | } |
455 | 0 | #endif |
456 | 0 | #if defined(HAS_CONVERT16TO8ROW_AVX512BW) |
457 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
458 | 0 | Convert16To8Row = Convert16To8Row_Any_AVX512BW; |
459 | 0 | if (IS_ALIGNED(width, 64)) { |
460 | 0 | Convert16To8Row = Convert16To8Row_AVX512BW; |
461 | 0 | } |
462 | 0 | } |
463 | 0 | #endif |
464 | | #if defined(HAS_CONVERT16TO8ROW_RVV) |
465 | | if (TestCpuFlag(kCpuHasRVV)) { |
466 | | Convert16To8Row = Convert16To8Row_RVV; |
467 | | } |
468 | | #endif |
469 | |
|
470 | | #if defined(HAS_HALFROW_16TO8_NEON) |
471 | | if (TestCpuFlag(kCpuHasNEON)) { |
472 | | HalfRow_16To8 = HalfRow_16To8_Any_NEON; |
473 | | if (IS_ALIGNED(uv_width, 16)) { |
474 | | HalfRow_16To8 = HalfRow_16To8_NEON; |
475 | | } |
476 | | } |
477 | | #endif |
478 | | #if defined(HAS_HALFROW_16TO8_SVE2) |
479 | | if (TestCpuFlag(kCpuHasSVE2)) { |
480 | | HalfRow_16To8 = HalfRow_16To8_SVE2; |
481 | | } |
482 | | #endif |
483 | | #if defined(HAS_HALFROW_16TO8_SME) |
484 | | if (TestCpuFlag(kCpuHasSME)) { |
485 | | HalfRow_16To8 = HalfRow_16To8_SME; |
486 | | } |
487 | | #endif |
488 | 0 | #if defined(HAS_HALFROW_16TO8_SSSE3) |
489 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
490 | 0 | HalfRow_16To8 = HalfRow_16To8_Any_SSSE3; |
491 | 0 | if (IS_ALIGNED(uv_width, 16)) { |
492 | 0 | HalfRow_16To8 = HalfRow_16To8_SSSE3; |
493 | 0 | } |
494 | 0 | } |
495 | 0 | #endif |
496 | 0 | #if defined(HAS_HALFROW_16TO8_AVX2) |
497 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
498 | 0 | HalfRow_16To8 = HalfRow_16To8_Any_AVX2; |
499 | 0 | if (IS_ALIGNED(uv_width, 32)) { |
500 | 0 | HalfRow_16To8 = HalfRow_16To8_AVX2; |
501 | 0 | } |
502 | 0 | } |
503 | 0 | #endif |
504 | 0 | #if defined(HAS_HALFROW_16TO8_AVX512BW) |
505 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
506 | 0 | HalfRow_16To8 = HalfRow_16To8_Any_AVX512BW; |
507 | 0 | if (IS_ALIGNED(uv_width, 64)) { |
508 | 0 | HalfRow_16To8 = HalfRow_16To8_AVX512BW; |
509 | 0 | } |
510 | 0 | } |
511 | 0 | #endif |
512 | | #if defined(HAS_HALFROW_16TO8_RVV) |
513 | | if (TestCpuFlag(kCpuHasRVV)) { |
514 | | HalfRow_16To8 = HalfRow_16To8_RVV; |
515 | | } |
516 | | #endif |
517 | |
|
518 | 0 | for (y = 0; y < height - 1; y += 2) { |
519 | 0 | HalfRow_16To8(src_u, src_stride_u, dst_u, scale, uv_width); |
520 | 0 | HalfRow_16To8(src_v, src_stride_v, dst_v, scale, uv_width); |
521 | 0 | if (dst_y) { |
522 | 0 | Convert16To8Row(src_y, dst_y, scale, width); |
523 | 0 | Convert16To8Row(src_y + src_stride_y, dst_y + dst_stride_y, scale, width); |
524 | 0 | src_y += src_stride_y * 2; |
525 | 0 | dst_y += dst_stride_y * 2; |
526 | 0 | } |
527 | 0 | src_u += src_stride_u * 2; |
528 | 0 | src_v += src_stride_v * 2; |
529 | 0 | dst_u += dst_stride_u; |
530 | 0 | dst_v += dst_stride_v; |
531 | 0 | } |
532 | 0 | if (height & 1) { |
533 | 0 | HalfRow_16To8(src_u, 0, dst_u, scale, uv_width); |
534 | 0 | HalfRow_16To8(src_v, 0, dst_v, scale, uv_width); |
535 | 0 | if (dst_y) { |
536 | 0 | Convert16To8Row(src_y, dst_y, scale, width); |
537 | 0 | } |
538 | 0 | } |
539 | 0 | return 0; |
540 | 0 | } |
541 | | |
542 | | // Convert 10 bit YUV to 8 bit. |
543 | | LIBYUV_API |
544 | | int I010ToI420(const uint16_t* src_y, |
545 | | int src_stride_y, |
546 | | const uint16_t* src_u, |
547 | | int src_stride_u, |
548 | | const uint16_t* src_v, |
549 | | int src_stride_v, |
550 | | uint8_t* dst_y, |
551 | | int dst_stride_y, |
552 | | uint8_t* dst_u, |
553 | | int dst_stride_u, |
554 | | uint8_t* dst_v, |
555 | | int dst_stride_v, |
556 | | int width, |
557 | 0 | int height) { |
558 | 0 | return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v, |
559 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, |
560 | 0 | dst_stride_u, dst_v, dst_stride_v, width, height, 1, |
561 | 0 | 1, 10); |
562 | 0 | } |
563 | | |
564 | | LIBYUV_API |
565 | | int I210ToI420(const uint16_t* src_y, |
566 | | int src_stride_y, |
567 | | const uint16_t* src_u, |
568 | | int src_stride_u, |
569 | | const uint16_t* src_v, |
570 | | int src_stride_v, |
571 | | uint8_t* dst_y, |
572 | | int dst_stride_y, |
573 | | uint8_t* dst_u, |
574 | | int dst_stride_u, |
575 | | uint8_t* dst_v, |
576 | | int dst_stride_v, |
577 | | int width, |
578 | 0 | int height) { |
579 | 0 | return I21xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, |
580 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
581 | 0 | dst_v, dst_stride_v, width, height, 10); |
582 | 0 | } |
583 | | |
584 | | LIBYUV_API |
585 | | int I210ToI422(const uint16_t* src_y, |
586 | | int src_stride_y, |
587 | | const uint16_t* src_u, |
588 | | int src_stride_u, |
589 | | const uint16_t* src_v, |
590 | | int src_stride_v, |
591 | | uint8_t* dst_y, |
592 | | int dst_stride_y, |
593 | | uint8_t* dst_u, |
594 | | int dst_stride_u, |
595 | | uint8_t* dst_v, |
596 | | int dst_stride_v, |
597 | | int width, |
598 | 0 | int height) { |
599 | 0 | return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v, |
600 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, |
601 | 0 | dst_stride_u, dst_v, dst_stride_v, width, height, 1, |
602 | 0 | 0, 10); |
603 | 0 | } |
604 | | |
605 | | LIBYUV_API |
606 | | int I410ToI420(const uint16_t* src_y, |
607 | | int src_stride_y, |
608 | | const uint16_t* src_u, |
609 | | int src_stride_u, |
610 | | const uint16_t* src_v, |
611 | | int src_stride_v, |
612 | | uint8_t* dst_y, |
613 | | int dst_stride_y, |
614 | | uint8_t* dst_u, |
615 | | int dst_stride_u, |
616 | | uint8_t* dst_v, |
617 | | int dst_stride_v, |
618 | | int width, |
619 | 0 | int height) { |
620 | 0 | return I41xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, |
621 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
622 | 0 | dst_v, dst_stride_v, width, height, 10); |
623 | 0 | } |
624 | | |
625 | | LIBYUV_API |
626 | | int I410ToI444(const uint16_t* src_y, |
627 | | int src_stride_y, |
628 | | const uint16_t* src_u, |
629 | | int src_stride_u, |
630 | | const uint16_t* src_v, |
631 | | int src_stride_v, |
632 | | uint8_t* dst_y, |
633 | | int dst_stride_y, |
634 | | uint8_t* dst_u, |
635 | | int dst_stride_u, |
636 | | uint8_t* dst_v, |
637 | | int dst_stride_v, |
638 | | int width, |
639 | 0 | int height) { |
640 | 0 | return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v, |
641 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, |
642 | 0 | dst_stride_u, dst_v, dst_stride_v, width, height, 0, |
643 | 0 | 0, 10); |
644 | 0 | } |
645 | | |
646 | | LIBYUV_API |
647 | | int I012ToI420(const uint16_t* src_y, |
648 | | int src_stride_y, |
649 | | const uint16_t* src_u, |
650 | | int src_stride_u, |
651 | | const uint16_t* src_v, |
652 | | int src_stride_v, |
653 | | uint8_t* dst_y, |
654 | | int dst_stride_y, |
655 | | uint8_t* dst_u, |
656 | | int dst_stride_u, |
657 | | uint8_t* dst_v, |
658 | | int dst_stride_v, |
659 | | int width, |
660 | 0 | int height) { |
661 | 0 | return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v, |
662 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, |
663 | 0 | dst_stride_u, dst_v, dst_stride_v, width, height, 1, |
664 | 0 | 1, 12); |
665 | 0 | } |
666 | | |
667 | | LIBYUV_API |
668 | | int I212ToI422(const uint16_t* src_y, |
669 | | int src_stride_y, |
670 | | const uint16_t* src_u, |
671 | | int src_stride_u, |
672 | | const uint16_t* src_v, |
673 | | int src_stride_v, |
674 | | uint8_t* dst_y, |
675 | | int dst_stride_y, |
676 | | uint8_t* dst_u, |
677 | | int dst_stride_u, |
678 | | uint8_t* dst_v, |
679 | | int dst_stride_v, |
680 | | int width, |
681 | 0 | int height) { |
682 | 0 | return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v, |
683 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, |
684 | 0 | dst_stride_u, dst_v, dst_stride_v, width, height, 1, |
685 | 0 | 0, 12); |
686 | 0 | } |
687 | | |
688 | | LIBYUV_API |
689 | | int I212ToI420(const uint16_t* src_y, |
690 | | int src_stride_y, |
691 | | const uint16_t* src_u, |
692 | | int src_stride_u, |
693 | | const uint16_t* src_v, |
694 | | int src_stride_v, |
695 | | uint8_t* dst_y, |
696 | | int dst_stride_y, |
697 | | uint8_t* dst_u, |
698 | | int dst_stride_u, |
699 | | uint8_t* dst_v, |
700 | | int dst_stride_v, |
701 | | int width, |
702 | 0 | int height) { |
703 | 0 | return I21xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, |
704 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
705 | 0 | dst_v, dst_stride_v, width, height, 12); |
706 | 0 | } |
707 | | |
708 | | LIBYUV_API |
709 | | int I412ToI444(const uint16_t* src_y, |
710 | | int src_stride_y, |
711 | | const uint16_t* src_u, |
712 | | int src_stride_u, |
713 | | const uint16_t* src_v, |
714 | | int src_stride_v, |
715 | | uint8_t* dst_y, |
716 | | int dst_stride_y, |
717 | | uint8_t* dst_u, |
718 | | int dst_stride_u, |
719 | | uint8_t* dst_v, |
720 | | int dst_stride_v, |
721 | | int width, |
722 | 0 | int height) { |
723 | 0 | return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v, |
724 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, |
725 | 0 | dst_stride_u, dst_v, dst_stride_v, width, height, 0, |
726 | 0 | 0, 12); |
727 | 0 | } |
728 | | |
729 | | LIBYUV_API |
730 | | int I412ToI420(const uint16_t* src_y, |
731 | | int src_stride_y, |
732 | | const uint16_t* src_u, |
733 | | int src_stride_u, |
734 | | const uint16_t* src_v, |
735 | | int src_stride_v, |
736 | | uint8_t* dst_y, |
737 | | int dst_stride_y, |
738 | | uint8_t* dst_u, |
739 | | int dst_stride_u, |
740 | | uint8_t* dst_v, |
741 | | int dst_stride_v, |
742 | | int width, |
743 | 0 | int height) { |
744 | 0 | return I41xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, |
745 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
746 | 0 | dst_v, dst_stride_v, width, height, 12); |
747 | 0 | } |
748 | | |
749 | | // Any Ix10 To I010 format |
750 | | static int Ix10ToI010(const uint16_t* src_y, |
751 | | int src_stride_y, |
752 | | const uint16_t* src_u, |
753 | | int src_stride_u, |
754 | | const uint16_t* src_v, |
755 | | int src_stride_v, |
756 | | uint16_t* dst_y, |
757 | | int dst_stride_y, |
758 | | uint16_t* dst_u, |
759 | | int dst_stride_u, |
760 | | uint16_t* dst_v, |
761 | | int dst_stride_v, |
762 | | int width, |
763 | | int height, |
764 | | int subsample_x, |
765 | 0 | int subsample_y) { |
766 | 0 | int r; |
767 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
768 | 0 | height == 0 || height == INT_MIN) { |
769 | 0 | return -1; |
770 | 0 | } |
771 | 0 | const int dst_y_width = width; |
772 | 0 | const int dst_y_height = Abs(height); |
773 | 0 | const int src_uv_width = SUBSAMPLE(width, subsample_x, subsample_x); |
774 | 0 | const int src_uv_height = SUBSAMPLE(height, subsample_y, subsample_y); |
775 | 0 | const int dst_uv_width = SUBSAMPLE(dst_y_width, 1, 1); |
776 | 0 | const int dst_uv_height = SUBSAMPLE(dst_y_height, 1, 1); |
777 | 0 | if (dst_y) { |
778 | 0 | CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
779 | 0 | } |
780 | 0 | r = ScalePlane_12(src_u, src_stride_u, src_uv_width, src_uv_height, dst_u, |
781 | 0 | dst_stride_u, dst_uv_width, dst_uv_height, kFilterBilinear); |
782 | 0 | if (r != 0) { |
783 | 0 | return r; |
784 | 0 | } |
785 | 0 | r = ScalePlane_12(src_v, src_stride_v, src_uv_width, src_uv_height, dst_v, |
786 | 0 | dst_stride_v, dst_uv_width, dst_uv_height, kFilterBilinear); |
787 | 0 | return r; |
788 | 0 | } |
789 | | |
790 | | LIBYUV_API |
791 | | int I410ToI010(const uint16_t* src_y, |
792 | | int src_stride_y, |
793 | | const uint16_t* src_u, |
794 | | int src_stride_u, |
795 | | const uint16_t* src_v, |
796 | | int src_stride_v, |
797 | | uint16_t* dst_y, |
798 | | int dst_stride_y, |
799 | | uint16_t* dst_u, |
800 | | int dst_stride_u, |
801 | | uint16_t* dst_v, |
802 | | int dst_stride_v, |
803 | | int width, |
804 | 0 | int height) { |
805 | 0 | return Ix10ToI010(src_y, src_stride_y, src_u, src_stride_u, src_v, |
806 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
807 | 0 | dst_v, dst_stride_v, width, height, 0, 0); |
808 | 0 | } |
809 | | |
810 | | LIBYUV_API |
811 | | int I210ToI010(const uint16_t* src_y, |
812 | | int src_stride_y, |
813 | | const uint16_t* src_u, |
814 | | int src_stride_u, |
815 | | const uint16_t* src_v, |
816 | | int src_stride_v, |
817 | | uint16_t* dst_y, |
818 | | int dst_stride_y, |
819 | | uint16_t* dst_u, |
820 | | int dst_stride_u, |
821 | | uint16_t* dst_v, |
822 | | int dst_stride_v, |
823 | | int width, |
824 | 0 | int height) { |
825 | 0 | return Ix10ToI010(src_y, src_stride_y, src_u, src_stride_u, src_v, |
826 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
827 | 0 | dst_v, dst_stride_v, width, height, 1, 0); |
828 | 0 | } |
829 | | |
830 | | // Any I[420]1[02] to P[420]1[02] format |
831 | | static int IxxxToPxxx(const uint16_t* src_y, |
832 | | int src_stride_y, |
833 | | const uint16_t* src_u, |
834 | | int src_stride_u, |
835 | | const uint16_t* src_v, |
836 | | int src_stride_v, |
837 | | uint16_t* dst_y, |
838 | | int dst_stride_y, |
839 | | uint16_t* dst_uv, |
840 | | int dst_stride_uv, |
841 | | int width, |
842 | | int height, |
843 | | int subsample_x, |
844 | | int subsample_y, |
845 | 0 | int depth) { |
846 | 0 | if (width <= 0 || height == 0 || height == INT_MIN) { |
847 | 0 | return -1; |
848 | 0 | } |
849 | 0 | const int uv_width = SUBSAMPLE(width, subsample_x, subsample_x); |
850 | 0 | const int uv_height = SUBSAMPLE(height, subsample_y, subsample_y); |
851 | |
|
852 | 0 | ConvertToMSBPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height, |
853 | 0 | depth); |
854 | 0 | MergeUVPlane_16(src_u, src_stride_u, src_v, src_stride_v, dst_uv, |
855 | 0 | dst_stride_uv, uv_width, uv_height, depth); |
856 | 0 | return 0; |
857 | 0 | } |
858 | | |
859 | | LIBYUV_API |
860 | | int I010ToP010(const uint16_t* src_y, |
861 | | int src_stride_y, |
862 | | const uint16_t* src_u, |
863 | | int src_stride_u, |
864 | | const uint16_t* src_v, |
865 | | int src_stride_v, |
866 | | uint16_t* dst_y, |
867 | | int dst_stride_y, |
868 | | uint16_t* dst_uv, |
869 | | int dst_stride_uv, |
870 | | int width, |
871 | 0 | int height) { |
872 | 0 | return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v, |
873 | 0 | src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv, |
874 | 0 | width, height, 1, 1, 10); |
875 | 0 | } |
876 | | |
877 | | LIBYUV_API |
878 | | int I010ToNV12(const uint16_t* src_y, |
879 | | int src_stride_y, |
880 | | const uint16_t* src_u, |
881 | | int src_stride_u, |
882 | | const uint16_t* src_v, |
883 | | int src_stride_v, |
884 | | uint8_t* dst_y, |
885 | | int dst_stride_y, |
886 | | uint8_t* dst_uv, |
887 | | int dst_stride_uv, |
888 | | int width, |
889 | 0 | int height) { |
890 | 0 | int y; |
891 | 0 | int halfwidth = (width + 1) >> 1; |
892 | 0 | int halfheight = (height + 1) >> 1; |
893 | 0 | const int scale = 16385; // 16384 for 10 bits |
894 | 0 | void (*Convert16To8Row)(const uint16_t* src_y, uint8_t* dst_y, int scale, |
895 | 0 | int width) = Convert16To8Row_C; |
896 | 0 | void (*MergeUVRow)(const uint8_t* src_u, const uint8_t* src_v, |
897 | 0 | uint8_t* dst_uv, int width) = MergeUVRow_C; |
898 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_uv || width <= 0 || |
899 | 0 | height == 0 || height == INT_MIN) { |
900 | 0 | return -1; |
901 | 0 | } |
902 | | // Negative height means invert the image. |
903 | 0 | if (height < 0) { |
904 | 0 | height = -height; |
905 | 0 | halfheight = (height + 1) >> 1; |
906 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
907 | 0 | src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u; |
908 | 0 | src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v; |
909 | 0 | src_stride_y = -src_stride_y; |
910 | 0 | src_stride_u = -src_stride_u; |
911 | 0 | src_stride_v = -src_stride_v; |
912 | 0 | } |
913 | | #if defined(HAS_CONVERT16TO8ROW_NEON) |
914 | | if (TestCpuFlag(kCpuHasNEON)) { |
915 | | Convert16To8Row = Convert16To8Row_Any_NEON; |
916 | | if (IS_ALIGNED(halfwidth, 16)) { |
917 | | Convert16To8Row = Convert16To8Row_NEON; |
918 | | } |
919 | | } |
920 | | #endif |
921 | | #if defined(HAS_CONVERT16TO8ROW_SME) |
922 | | if (TestCpuFlag(kCpuHasSME)) { |
923 | | Convert16To8Row = Convert16To8Row_SME; |
924 | | } |
925 | | #endif |
926 | 0 | #if defined(HAS_CONVERT16TO8ROW_SSSE3) |
927 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
928 | 0 | Convert16To8Row = Convert16To8Row_Any_SSSE3; |
929 | 0 | if (IS_ALIGNED(halfwidth, 16)) { |
930 | 0 | Convert16To8Row = Convert16To8Row_SSSE3; |
931 | 0 | } |
932 | 0 | } |
933 | 0 | #endif |
934 | 0 | #if defined(HAS_CONVERT16TO8ROW_AVX2) |
935 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
936 | 0 | Convert16To8Row = Convert16To8Row_Any_AVX2; |
937 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
938 | 0 | Convert16To8Row = Convert16To8Row_AVX2; |
939 | 0 | } |
940 | 0 | } |
941 | 0 | #endif |
942 | 0 | #if defined(HAS_CONVERT16TO8ROW_AVX512BW) |
943 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
944 | 0 | Convert16To8Row = Convert16To8Row_Any_AVX512BW; |
945 | 0 | if (IS_ALIGNED(halfwidth, 64)) { |
946 | 0 | Convert16To8Row = Convert16To8Row_AVX512BW; |
947 | 0 | } |
948 | 0 | } |
949 | 0 | #endif |
950 | |
|
951 | 0 | #if defined(HAS_MERGEUVROW_SSE2) |
952 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
953 | 0 | MergeUVRow = MergeUVRow_Any_SSE2; |
954 | 0 | if (IS_ALIGNED(halfwidth, 16)) { |
955 | 0 | MergeUVRow = MergeUVRow_SSE2; |
956 | 0 | } |
957 | 0 | } |
958 | 0 | #endif |
959 | 0 | #if defined(HAS_MERGEUVROW_AVX2) |
960 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
961 | 0 | MergeUVRow = MergeUVRow_Any_AVX2; |
962 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
963 | 0 | MergeUVRow = MergeUVRow_AVX2; |
964 | 0 | } |
965 | 0 | } |
966 | 0 | #endif |
967 | 0 | #if defined(HAS_MERGEUVROW_AVX512BW) |
968 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
969 | 0 | MergeUVRow = MergeUVRow_Any_AVX512BW; |
970 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
971 | 0 | MergeUVRow = MergeUVRow_AVX512BW; |
972 | 0 | } |
973 | 0 | } |
974 | 0 | #endif |
975 | | #if defined(HAS_MERGEUVROW_NEON) |
976 | | if (TestCpuFlag(kCpuHasNEON)) { |
977 | | MergeUVRow = MergeUVRow_Any_NEON; |
978 | | if (IS_ALIGNED(halfwidth, 16)) { |
979 | | MergeUVRow = MergeUVRow_NEON; |
980 | | } |
981 | | } |
982 | | #endif |
983 | | #if defined(HAS_MERGEUVROW_SVE2) |
984 | | if (TestCpuFlag(kCpuHasSVE2)) { |
985 | | MergeUVRow = MergeUVRow_SVE2; |
986 | | } |
987 | | #endif |
988 | | #if defined(HAS_MERGEUVROW_SME) |
989 | | if (TestCpuFlag(kCpuHasSME)) { |
990 | | MergeUVRow = MergeUVRow_SME; |
991 | | } |
992 | | #endif |
993 | | #if defined(HAS_MERGEUVROW_LSX) |
994 | | if (TestCpuFlag(kCpuHasLSX)) { |
995 | | MergeUVRow = MergeUVRow_Any_LSX; |
996 | | if (IS_ALIGNED(halfwidth, 16)) { |
997 | | MergeUVRow = MergeUVRow_LSX; |
998 | | } |
999 | | } |
1000 | | #endif |
1001 | | #if defined(HAS_MERGEUVROW_RVV) |
1002 | | if (TestCpuFlag(kCpuHasRVV)) { |
1003 | | MergeUVRow = MergeUVRow_RVV; |
1004 | | } |
1005 | | #endif |
1006 | | |
1007 | | // Convert Y plane. |
1008 | 0 | if (dst_y) { |
1009 | 0 | Convert16To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale, width, |
1010 | 0 | height); |
1011 | 0 | } |
1012 | |
|
1013 | 0 | { |
1014 | | // Allocate a row of uv. |
1015 | 0 | align_buffer_64(row_u, ((halfwidth + 31) & ~31) * 2); |
1016 | 0 | uint8_t* row_v = row_u + ((halfwidth + 31) & ~31); |
1017 | 0 | if (!row_u) |
1018 | 0 | return 1; |
1019 | | |
1020 | 0 | for (y = 0; y < halfheight; ++y) { |
1021 | 0 | Convert16To8Row(src_u, row_u, scale, halfwidth); |
1022 | 0 | Convert16To8Row(src_v, row_v, scale, halfwidth); |
1023 | 0 | MergeUVRow(row_u, row_v, dst_uv, halfwidth); |
1024 | 0 | src_u += src_stride_u; |
1025 | 0 | src_v += src_stride_v; |
1026 | 0 | dst_uv += dst_stride_uv; |
1027 | 0 | } |
1028 | 0 | free_aligned_buffer_64(row_u); |
1029 | 0 | } |
1030 | 0 | return 0; |
1031 | 0 | } |
1032 | | |
1033 | | LIBYUV_API |
1034 | | int I210ToP210(const uint16_t* src_y, |
1035 | | int src_stride_y, |
1036 | | const uint16_t* src_u, |
1037 | | int src_stride_u, |
1038 | | const uint16_t* src_v, |
1039 | | int src_stride_v, |
1040 | | uint16_t* dst_y, |
1041 | | int dst_stride_y, |
1042 | | uint16_t* dst_uv, |
1043 | | int dst_stride_uv, |
1044 | | int width, |
1045 | 0 | int height) { |
1046 | 0 | return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1047 | 0 | src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv, |
1048 | 0 | width, height, 1, 0, 10); |
1049 | 0 | } |
1050 | | |
1051 | | LIBYUV_API |
1052 | | int I012ToP012(const uint16_t* src_y, |
1053 | | int src_stride_y, |
1054 | | const uint16_t* src_u, |
1055 | | int src_stride_u, |
1056 | | const uint16_t* src_v, |
1057 | | int src_stride_v, |
1058 | | uint16_t* dst_y, |
1059 | | int dst_stride_y, |
1060 | | uint16_t* dst_uv, |
1061 | | int dst_stride_uv, |
1062 | | int width, |
1063 | 0 | int height) { |
1064 | 0 | return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1065 | 0 | src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv, |
1066 | 0 | width, height, 1, 1, 12); |
1067 | 0 | } |
1068 | | |
1069 | | LIBYUV_API |
1070 | | int I212ToP212(const uint16_t* src_y, |
1071 | | int src_stride_y, |
1072 | | const uint16_t* src_u, |
1073 | | int src_stride_u, |
1074 | | const uint16_t* src_v, |
1075 | | int src_stride_v, |
1076 | | uint16_t* dst_y, |
1077 | | int dst_stride_y, |
1078 | | uint16_t* dst_uv, |
1079 | | int dst_stride_uv, |
1080 | | int width, |
1081 | 0 | int height) { |
1082 | 0 | return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1083 | 0 | src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv, |
1084 | 0 | width, height, 1, 0, 12); |
1085 | 0 | } |
1086 | | |
1087 | | // 422 chroma is 1/2 width, 1x height |
1088 | | // 420 chroma is 1/2 width, 1/2 height |
1089 | | LIBYUV_API |
1090 | | int I422ToI420(const uint8_t* src_y, |
1091 | | int src_stride_y, |
1092 | | const uint8_t* src_u, |
1093 | | int src_stride_u, |
1094 | | const uint8_t* src_v, |
1095 | | int src_stride_v, |
1096 | | uint8_t* dst_y, |
1097 | | int dst_stride_y, |
1098 | | uint8_t* dst_u, |
1099 | | int dst_stride_u, |
1100 | | uint8_t* dst_v, |
1101 | | int dst_stride_v, |
1102 | | int width, |
1103 | 0 | int height) { |
1104 | 0 | const int src_uv_width = SUBSAMPLE(width, 1, 1); |
1105 | 0 | return I4xxToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1106 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
1107 | 0 | dst_v, dst_stride_v, width, height, src_uv_width, height); |
1108 | 0 | } |
1109 | | |
1110 | | LIBYUV_API |
1111 | | int I422ToI210(const uint8_t* src_y, |
1112 | | int src_stride_y, |
1113 | | const uint8_t* src_u, |
1114 | | int src_stride_u, |
1115 | | const uint8_t* src_v, |
1116 | | int src_stride_v, |
1117 | | uint16_t* dst_y, |
1118 | | int dst_stride_y, |
1119 | | uint16_t* dst_u, |
1120 | | int dst_stride_u, |
1121 | | uint16_t* dst_v, |
1122 | | int dst_stride_v, |
1123 | | int width, |
1124 | 0 | int height) { |
1125 | 0 | int halfwidth = (width + 1) >> 1; |
1126 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
1127 | 0 | height == 0 || height == INT_MIN) { |
1128 | 0 | return -1; |
1129 | 0 | } |
1130 | | // Negative height means invert the image. |
1131 | 0 | if (height < 0) { |
1132 | 0 | height = -height; |
1133 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
1134 | 0 | src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u; |
1135 | 0 | src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v; |
1136 | 0 | src_stride_y = -src_stride_y; |
1137 | 0 | src_stride_u = -src_stride_u; |
1138 | 0 | src_stride_v = -src_stride_v; |
1139 | 0 | } |
1140 | | |
1141 | | // Convert Y plane. |
1142 | 0 | Convert8To16Plane(src_y, src_stride_y, dst_y, dst_stride_y, 1024, width, |
1143 | 0 | height); |
1144 | | // Convert UV planes. |
1145 | 0 | Convert8To16Plane(src_u, src_stride_u, dst_u, dst_stride_u, 1024, halfwidth, |
1146 | 0 | height); |
1147 | 0 | Convert8To16Plane(src_v, src_stride_v, dst_v, dst_stride_v, 1024, halfwidth, |
1148 | 0 | height); |
1149 | 0 | return 0; |
1150 | 0 | } |
1151 | | |
1152 | | // TODO(fbarchard): Implement row conversion. |
1153 | | LIBYUV_API |
1154 | | int I422ToNV21(const uint8_t* src_y, |
1155 | | int src_stride_y, |
1156 | | const uint8_t* src_u, |
1157 | | int src_stride_u, |
1158 | | const uint8_t* src_v, |
1159 | | int src_stride_v, |
1160 | | uint8_t* dst_y, |
1161 | | int dst_stride_y, |
1162 | | uint8_t* dst_vu, |
1163 | | int dst_stride_vu, |
1164 | | int width, |
1165 | 0 | int height) { |
1166 | 0 | int r; |
1167 | 0 | if (width <= 0 || height == 0 || height == INT_MIN) { |
1168 | 0 | return -1; |
1169 | 0 | } |
1170 | 0 | int halfwidth = (width + 1) >> 1; |
1171 | 0 | int halfheight = (height + 1) >> 1; |
1172 | | // Negative height means invert the image. |
1173 | 0 | if (height < 0) { |
1174 | 0 | height = -height; |
1175 | 0 | halfheight = (height + 1) >> 1; |
1176 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
1177 | 0 | src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u; |
1178 | 0 | src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v; |
1179 | 0 | src_stride_y = -src_stride_y; |
1180 | 0 | src_stride_u = -src_stride_u; |
1181 | 0 | src_stride_v = -src_stride_v; |
1182 | 0 | } |
1183 | | |
1184 | | // Allocate u and v buffers |
1185 | 0 | const uint64_t plane_size = (uint64_t)halfwidth * halfheight; |
1186 | 0 | if (plane_size > SIZE_MAX / 2) |
1187 | 0 | return 1; |
1188 | 0 | align_buffer_64(plane_u, (size_t)plane_size * 2); |
1189 | 0 | if (!plane_u) |
1190 | 0 | return 1; |
1191 | 0 | uint8_t* plane_v = plane_u + (size_t)plane_size; |
1192 | |
|
1193 | 0 | r = I422ToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
1194 | 0 | dst_y, dst_stride_y, plane_u, halfwidth, plane_v, halfwidth, |
1195 | 0 | width, height); |
1196 | 0 | if (r != 0) { |
1197 | 0 | return r; |
1198 | 0 | } |
1199 | 0 | MergeUVPlane(plane_v, halfwidth, plane_u, halfwidth, dst_vu, dst_stride_vu, |
1200 | 0 | halfwidth, halfheight); |
1201 | 0 | free_aligned_buffer_64(plane_u); |
1202 | 0 | return 0; |
1203 | 0 | } |
1204 | | |
1205 | | LIBYUV_API |
1206 | | int MM21ToNV12(const uint8_t* src_y, |
1207 | | int src_stride_y, |
1208 | | const uint8_t* src_uv, |
1209 | | int src_stride_uv, |
1210 | | uint8_t* dst_y, |
1211 | | int dst_stride_y, |
1212 | | uint8_t* dst_uv, |
1213 | | int dst_stride_uv, |
1214 | | int width, |
1215 | 0 | int height) { |
1216 | 0 | if (!src_uv || !dst_uv || width <= 0) { |
1217 | 0 | return -1; |
1218 | 0 | } |
1219 | | |
1220 | 0 | int sign = height < 0 ? -1 : 1; |
1221 | |
|
1222 | 0 | if (dst_y) { |
1223 | 0 | DetilePlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height, 32); |
1224 | 0 | } |
1225 | 0 | DetilePlane(src_uv, src_stride_uv, dst_uv, dst_stride_uv, (width + 1) & ~1, |
1226 | 0 | (height + sign) / 2, 16); |
1227 | |
|
1228 | 0 | return 0; |
1229 | 0 | } |
1230 | | |
1231 | | LIBYUV_API |
1232 | | int MM21ToI420(const uint8_t* src_y, |
1233 | | int src_stride_y, |
1234 | | const uint8_t* src_uv, |
1235 | | int src_stride_uv, |
1236 | | uint8_t* dst_y, |
1237 | | int dst_stride_y, |
1238 | | uint8_t* dst_u, |
1239 | | int dst_stride_u, |
1240 | | uint8_t* dst_v, |
1241 | | int dst_stride_v, |
1242 | | int width, |
1243 | 0 | int height) { |
1244 | 0 | int sign = height < 0 ? -1 : 1; |
1245 | |
|
1246 | 0 | if (!src_uv || !dst_u || !dst_v || width <= 0) { |
1247 | 0 | return -1; |
1248 | 0 | } |
1249 | | |
1250 | 0 | if (dst_y) { |
1251 | 0 | DetilePlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height, 32); |
1252 | 0 | } |
1253 | 0 | DetileSplitUVPlane(src_uv, src_stride_uv, dst_u, dst_stride_u, dst_v, |
1254 | 0 | dst_stride_v, (width + 1) & ~1, (height + sign) / 2, 16); |
1255 | |
|
1256 | 0 | return 0; |
1257 | 0 | } |
1258 | | |
1259 | | LIBYUV_API |
1260 | | int MM21ToYUY2(const uint8_t* src_y, |
1261 | | int src_stride_y, |
1262 | | const uint8_t* src_uv, |
1263 | | int src_stride_uv, |
1264 | | uint8_t* dst_yuy2, |
1265 | | int dst_stride_yuy2, |
1266 | | int width, |
1267 | 0 | int height) { |
1268 | 0 | if (!src_y || !src_uv || !dst_yuy2 || width <= 0) { |
1269 | 0 | return -1; |
1270 | 0 | } |
1271 | | |
1272 | 0 | DetileToYUY2(src_y, src_stride_y, src_uv, src_stride_uv, dst_yuy2, |
1273 | 0 | dst_stride_yuy2, width, height, 32); |
1274 | |
|
1275 | 0 | return 0; |
1276 | 0 | } |
1277 | | |
1278 | | // Convert MT2T into P010. See tinyurl.com/mtk-10bit-video-format for format |
1279 | | // documentation. |
1280 | | // TODO(greenjustin): Add an MT2T to I420 conversion. |
1281 | | LIBYUV_API |
1282 | | int MT2TToP010(const uint8_t* src_y, |
1283 | | int src_stride_y, |
1284 | | const uint8_t* src_uv, |
1285 | | int src_stride_uv, |
1286 | | uint16_t* dst_y, |
1287 | | int dst_stride_y, |
1288 | | uint16_t* dst_uv, |
1289 | | int dst_stride_uv, |
1290 | | int width, |
1291 | 0 | int height) { |
1292 | 0 | if (width <= 0 || height == 0 || height == INT_MIN || !src_uv || !dst_uv) { |
1293 | 0 | return -1; |
1294 | 0 | } |
1295 | | |
1296 | 0 | { |
1297 | 0 | int uv_width = (width + 1) & ~1; |
1298 | 0 | int uv_height = (height + 1) / 2; |
1299 | 0 | int y = 0; |
1300 | 0 | const int tile_width = 16; |
1301 | 0 | const int y_tile_height = 32; |
1302 | 0 | const int uv_tile_height = 16; |
1303 | 0 | int padded_width = (width + tile_width - 1) & ~(tile_width - 1); |
1304 | 0 | int y_tile_row_size = padded_width * y_tile_height * 10 / 8; |
1305 | 0 | int uv_tile_row_size = padded_width * uv_tile_height * 10 / 8; |
1306 | 0 | size_t row_buf_size = padded_width * y_tile_height * sizeof(uint16_t); |
1307 | 0 | void (*UnpackMT2T)(const uint8_t* src, uint16_t* dst, size_t size) = |
1308 | 0 | UnpackMT2T_C; |
1309 | 0 | align_buffer_64(row_buf, row_buf_size); |
1310 | 0 | if (!row_buf) |
1311 | 0 | return 1; |
1312 | | |
1313 | | #if defined(HAS_UNPACKMT2T_NEON) |
1314 | | if (TestCpuFlag(kCpuHasNEON)) { |
1315 | | UnpackMT2T = UnpackMT2T_NEON; |
1316 | | } |
1317 | | #endif |
1318 | | // Negative height means invert the image. |
1319 | 0 | if (height < 0) { |
1320 | 0 | height = -height; |
1321 | 0 | uv_height = (height + 1) / 2; |
1322 | 0 | if (dst_y) { |
1323 | 0 | dst_y = dst_y + (ptrdiff_t)(height - 1) * dst_stride_y; |
1324 | 0 | dst_stride_y = -dst_stride_y; |
1325 | 0 | } |
1326 | 0 | dst_uv = dst_uv + (ptrdiff_t)(uv_height - 1) * dst_stride_uv; |
1327 | 0 | dst_stride_uv = -dst_stride_uv; |
1328 | 0 | } |
1329 | | |
1330 | | // Unpack and detile Y in rows of tiles |
1331 | 0 | if (src_y && dst_y) { |
1332 | 0 | for (y = 0; y < (height & ~(y_tile_height - 1)); y += y_tile_height) { |
1333 | 0 | UnpackMT2T(src_y, (uint16_t*)row_buf, y_tile_row_size); |
1334 | 0 | DetilePlane_16((uint16_t*)row_buf, padded_width, dst_y, dst_stride_y, |
1335 | 0 | width, y_tile_height, y_tile_height); |
1336 | 0 | src_y += src_stride_y * y_tile_height; |
1337 | 0 | dst_y += dst_stride_y * y_tile_height; |
1338 | 0 | } |
1339 | 0 | if (height & (y_tile_height - 1)) { |
1340 | 0 | UnpackMT2T(src_y, (uint16_t*)row_buf, y_tile_row_size); |
1341 | 0 | DetilePlane_16((uint16_t*)row_buf, padded_width, dst_y, dst_stride_y, |
1342 | 0 | width, height & (y_tile_height - 1), y_tile_height); |
1343 | 0 | } |
1344 | 0 | } |
1345 | | |
1346 | | // Unpack and detile UV plane |
1347 | 0 | for (y = 0; y < (uv_height & ~(uv_tile_height - 1)); y += uv_tile_height) { |
1348 | 0 | UnpackMT2T(src_uv, (uint16_t*)row_buf, uv_tile_row_size); |
1349 | 0 | DetilePlane_16((uint16_t*)row_buf, padded_width, dst_uv, dst_stride_uv, |
1350 | 0 | uv_width, uv_tile_height, uv_tile_height); |
1351 | 0 | src_uv += src_stride_uv * uv_tile_height; |
1352 | 0 | dst_uv += dst_stride_uv * uv_tile_height; |
1353 | 0 | } |
1354 | 0 | if (uv_height & (uv_tile_height - 1)) { |
1355 | 0 | UnpackMT2T(src_uv, (uint16_t*)row_buf, uv_tile_row_size); |
1356 | 0 | DetilePlane_16((uint16_t*)row_buf, padded_width, dst_uv, dst_stride_uv, |
1357 | 0 | uv_width, uv_height & (uv_tile_height - 1), |
1358 | 0 | uv_tile_height); |
1359 | 0 | } |
1360 | 0 | free_aligned_buffer_64(row_buf); |
1361 | 0 | } |
1362 | 0 | return 0; |
1363 | 0 | } |
1364 | | |
1365 | | #ifdef I422TONV21_ROW_VERSION |
1366 | | // Unittest fails for this version. |
1367 | | // 422 chroma is 1/2 width, 1x height |
1368 | | // 420 chroma is 1/2 width, 1/2 height |
1369 | | // Swap src_u and src_v to implement I422ToNV12 |
1370 | | LIBYUV_API |
1371 | | int I422ToNV21(const uint8_t* src_y, |
1372 | | int src_stride_y, |
1373 | | const uint8_t* src_u, |
1374 | | int src_stride_u, |
1375 | | const uint8_t* src_v, |
1376 | | int src_stride_v, |
1377 | | uint8_t* dst_y, |
1378 | | int dst_stride_y, |
1379 | | uint8_t* dst_vu, |
1380 | | int dst_stride_vu, |
1381 | | int width, |
1382 | | int height) { |
1383 | | int y; |
1384 | | void (*MergeUVRow)(const uint8_t* src_u, const uint8_t* src_v, |
1385 | | uint8_t* dst_uv, int width) = MergeUVRow_C; |
1386 | | void (*InterpolateRow)(uint8_t* dst_ptr, const uint8_t* src_ptr, |
1387 | | ptrdiff_t src_stride, int dst_width, |
1388 | | int source_y_fraction) = InterpolateRow_C; |
1389 | | int halfwidth = (width + 1) >> 1; |
1390 | | int halfheight = (height + 1) >> 1; |
1391 | | if ((!src_y && dst_y) || !src_u || !src_v || !dst_vu || width <= 0 || |
1392 | | height == 0 || height == INT_MIN) { |
1393 | | return -1; |
1394 | | } |
1395 | | // Negative height means invert the image. |
1396 | | if (height < 0) { |
1397 | | height = -height; |
1398 | | halfheight = (height + 1) >> 1; |
1399 | | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
1400 | | src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u; |
1401 | | src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v; |
1402 | | src_stride_y = -src_stride_y; |
1403 | | src_stride_u = -src_stride_u; |
1404 | | src_stride_v = -src_stride_v; |
1405 | | } |
1406 | | #if defined(HAS_MERGEUVROW_SSE2) |
1407 | | if (TestCpuFlag(kCpuHasSSE2)) { |
1408 | | MergeUVRow = MergeUVRow_Any_SSE2; |
1409 | | if (IS_ALIGNED(halfwidth, 16)) { |
1410 | | MergeUVRow = MergeUVRow_SSE2; |
1411 | | } |
1412 | | } |
1413 | | #endif |
1414 | | #if defined(HAS_MERGEUVROW_AVX2) |
1415 | | if (TestCpuFlag(kCpuHasAVX2)) { |
1416 | | MergeUVRow = MergeUVRow_Any_AVX2; |
1417 | | if (IS_ALIGNED(halfwidth, 32)) { |
1418 | | MergeUVRow = MergeUVRow_AVX2; |
1419 | | } |
1420 | | } |
1421 | | #endif |
1422 | | #if defined(HAS_MERGEUVROW_AVX512BW) |
1423 | | if (TestCpuFlag(kCpuHasAVX512BW)) { |
1424 | | MergeUVRow = MergeUVRow_Any_AVX512BW; |
1425 | | if (IS_ALIGNED(halfwidth, 32)) { |
1426 | | MergeUVRow = MergeUVRow_AVX512BW; |
1427 | | } |
1428 | | } |
1429 | | #endif |
1430 | | #if defined(HAS_MERGEUVROW_NEON) |
1431 | | if (TestCpuFlag(kCpuHasNEON)) { |
1432 | | MergeUVRow = MergeUVRow_Any_NEON; |
1433 | | if (IS_ALIGNED(halfwidth, 16)) { |
1434 | | MergeUVRow = MergeUVRow_NEON; |
1435 | | } |
1436 | | } |
1437 | | #endif |
1438 | | #if defined(HAS_MERGEUVROW_SVE2) |
1439 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1440 | | MergeUVRow = MergeUVRow_SVE2; |
1441 | | } |
1442 | | #endif |
1443 | | #if defined(HAS_MERGEUVROW_SME) |
1444 | | if (TestCpuFlag(kCpuHasSME)) { |
1445 | | MergeUVRow = MergeUVRow_SME; |
1446 | | } |
1447 | | #endif |
1448 | | #if defined(HAS_MERGEUVROW_LSX) |
1449 | | if (TestCpuFlag(kCpuHasLSX)) { |
1450 | | MergeUVRow = MergeUVRow_Any_LSX; |
1451 | | if (IS_ALIGNED(halfwidth, 16)) { |
1452 | | MergeUVRow = MergeUVRow_LSX; |
1453 | | } |
1454 | | } |
1455 | | #endif |
1456 | | #if defined(HAS_MERGEUVROW_RVV) |
1457 | | if (TestCpuFlag(kCpuHasRVV)) { |
1458 | | MergeUVRow = MergeUVRow_RVV; |
1459 | | } |
1460 | | #endif |
1461 | | #if defined(HAS_INTERPOLATEROW_AVX2) |
1462 | | if (TestCpuFlag(kCpuHasAVX2)) { |
1463 | | InterpolateRow = InterpolateRow_Any_AVX2; |
1464 | | if (IS_ALIGNED(width, 32)) { |
1465 | | InterpolateRow = InterpolateRow_AVX2; |
1466 | | } |
1467 | | } |
1468 | | #endif |
1469 | | #if defined(HAS_INTERPOLATEROW_NEON) |
1470 | | if (TestCpuFlag(kCpuHasNEON)) { |
1471 | | InterpolateRow = InterpolateRow_Any_NEON; |
1472 | | if (IS_ALIGNED(width, 16)) { |
1473 | | InterpolateRow = InterpolateRow_NEON; |
1474 | | } |
1475 | | } |
1476 | | #endif |
1477 | | #if defined(HAS_INTERPOLATEROW_SVE2) |
1478 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1479 | | InterpolateRow = InterpolateRow_SVE2; |
1480 | | } |
1481 | | #endif |
1482 | | #if defined(HAS_INTERPOLATEROW_SME) |
1483 | | if (TestCpuFlag(kCpuHasSME)) { |
1484 | | InterpolateRow = InterpolateRow_SME; |
1485 | | } |
1486 | | #endif |
1487 | | #if defined(HAS_INTERPOLATEROW_LSX) |
1488 | | if (TestCpuFlag(kCpuHasLSX)) { |
1489 | | InterpolateRow = InterpolateRow_Any_LSX; |
1490 | | if (IS_ALIGNED(width, 32)) { |
1491 | | InterpolateRow = InterpolateRow_LSX; |
1492 | | } |
1493 | | } |
1494 | | #endif |
1495 | | #if defined(HAS_INTERPOLATEROW_RVV) |
1496 | | if (TestCpuFlag(kCpuHasRVV)) { |
1497 | | InterpolateRow = InterpolateRow_RVV; |
1498 | | } |
1499 | | #endif |
1500 | | |
1501 | | if (dst_y) { |
1502 | | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, halfwidth, height); |
1503 | | } |
1504 | | { |
1505 | | // Allocate 2 rows of vu. |
1506 | | int awidth = halfwidth * 2; |
1507 | | align_buffer_64(row_vu_0, awidth * 2); |
1508 | | uint8_t* row_vu_1 = row_vu_0 + awidth; |
1509 | | if (!row_vu_0) |
1510 | | return 1; |
1511 | | |
1512 | | for (y = 0; y < height - 1; y += 2) { |
1513 | | MergeUVRow(src_v, src_u, row_vu_0, halfwidth); |
1514 | | MergeUVRow(src_v + src_stride_v, src_u + src_stride_u, row_vu_1, |
1515 | | halfwidth); |
1516 | | InterpolateRow(dst_vu, row_vu_0, awidth, awidth, 128); |
1517 | | src_u += src_stride_u * 2; |
1518 | | src_v += src_stride_v * 2; |
1519 | | dst_vu += dst_stride_vu; |
1520 | | } |
1521 | | if (height & 1) { |
1522 | | MergeUVRow(src_v, src_u, dst_vu, halfwidth); |
1523 | | } |
1524 | | free_aligned_buffer_64(row_vu_0); |
1525 | | } |
1526 | | return 0; |
1527 | | } |
1528 | | #endif // I422TONV21_ROW_VERSION |
1529 | | |
1530 | | // 444 chroma is 1x width, 1x height |
1531 | | // 420 chroma is 1/2 width, 1/2 height |
1532 | | LIBYUV_API |
1533 | | int I444ToI420(const uint8_t* src_y, |
1534 | | int src_stride_y, |
1535 | | const uint8_t* src_u, |
1536 | | int src_stride_u, |
1537 | | const uint8_t* src_v, |
1538 | | int src_stride_v, |
1539 | | uint8_t* dst_y, |
1540 | | int dst_stride_y, |
1541 | | uint8_t* dst_u, |
1542 | | int dst_stride_u, |
1543 | | uint8_t* dst_v, |
1544 | | int dst_stride_v, |
1545 | | int width, |
1546 | 0 | int height) { |
1547 | 0 | return I4xxToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1548 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u, |
1549 | 0 | dst_v, dst_stride_v, width, height, width, height); |
1550 | 0 | } |
1551 | | |
1552 | | LIBYUV_API |
1553 | | int I444ToNV12(const uint8_t* src_y, |
1554 | | int src_stride_y, |
1555 | | const uint8_t* src_u, |
1556 | | int src_stride_u, |
1557 | | const uint8_t* src_v, |
1558 | | int src_stride_v, |
1559 | | uint8_t* dst_y, |
1560 | | int dst_stride_y, |
1561 | | uint8_t* dst_uv, |
1562 | | int dst_stride_uv, |
1563 | | int width, |
1564 | 0 | int height) { |
1565 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_uv || width <= 0 || |
1566 | 0 | height == 0 || height == INT_MIN) { |
1567 | 0 | return -1; |
1568 | 0 | } |
1569 | | // Negative height means invert the image. |
1570 | 0 | if (height < 0) { |
1571 | 0 | height = -height; |
1572 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
1573 | 0 | src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u; |
1574 | 0 | src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v; |
1575 | 0 | src_stride_y = -src_stride_y; |
1576 | 0 | src_stride_u = -src_stride_u; |
1577 | 0 | src_stride_v = -src_stride_v; |
1578 | 0 | } |
1579 | 0 | if (dst_y) { |
1580 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1581 | 0 | } |
1582 | 0 | HalfMergeUVPlane(src_u, src_stride_u, src_v, src_stride_v, dst_uv, |
1583 | 0 | dst_stride_uv, width, height); |
1584 | 0 | return 0; |
1585 | 0 | } |
1586 | | |
1587 | | LIBYUV_API |
1588 | | int I444ToNV21(const uint8_t* src_y, |
1589 | | int src_stride_y, |
1590 | | const uint8_t* src_u, |
1591 | | int src_stride_u, |
1592 | | const uint8_t* src_v, |
1593 | | int src_stride_v, |
1594 | | uint8_t* dst_y, |
1595 | | int dst_stride_y, |
1596 | | uint8_t* dst_vu, |
1597 | | int dst_stride_vu, |
1598 | | int width, |
1599 | 0 | int height) { |
1600 | 0 | return I444ToNV12(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1601 | 0 | src_stride_u, dst_y, dst_stride_y, dst_vu, dst_stride_vu, |
1602 | 0 | width, height); |
1603 | 0 | } |
1604 | | |
1605 | | // I400 is greyscale typically used in MJPG |
1606 | | LIBYUV_API |
1607 | | int I400ToI420(const uint8_t* src_y, |
1608 | | int src_stride_y, |
1609 | | uint8_t* dst_y, |
1610 | | int dst_stride_y, |
1611 | | uint8_t* dst_u, |
1612 | | int dst_stride_u, |
1613 | | uint8_t* dst_v, |
1614 | | int dst_stride_v, |
1615 | | int width, |
1616 | 0 | int height) { |
1617 | 0 | int halfwidth = (width + 1) >> 1; |
1618 | 0 | int halfheight = (height + 1) >> 1; |
1619 | 0 | if ((!src_y && dst_y) || !dst_u || !dst_v || width <= 0 || height == 0 || |
1620 | 0 | height == INT_MIN) { |
1621 | 0 | return -1; |
1622 | 0 | } |
1623 | | // Negative height means invert the image. |
1624 | 0 | if (height < 0) { |
1625 | 0 | height = -height; |
1626 | 0 | halfheight = (height + 1) >> 1; |
1627 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
1628 | 0 | src_stride_y = -src_stride_y; |
1629 | 0 | } |
1630 | 0 | if (dst_y) { |
1631 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1632 | 0 | } |
1633 | 0 | SetPlane(dst_u, dst_stride_u, halfwidth, halfheight, 128); |
1634 | 0 | SetPlane(dst_v, dst_stride_v, halfwidth, halfheight, 128); |
1635 | 0 | return 0; |
1636 | 0 | } |
1637 | | |
1638 | | // I400 is greyscale typically used in MJPG |
1639 | | LIBYUV_API |
1640 | | int I400ToNV21(const uint8_t* src_y, |
1641 | | int src_stride_y, |
1642 | | uint8_t* dst_y, |
1643 | | int dst_stride_y, |
1644 | | uint8_t* dst_vu, |
1645 | | int dst_stride_vu, |
1646 | | int width, |
1647 | 0 | int height) { |
1648 | 0 | int halfwidth = (width + 1) >> 1; |
1649 | 0 | int halfheight = (height + 1) >> 1; |
1650 | 0 | if ((!src_y && dst_y) || !dst_vu || width <= 0 || height == 0 || |
1651 | 0 | height == INT_MIN) { |
1652 | 0 | return -1; |
1653 | 0 | } |
1654 | | // Negative height means invert the image. |
1655 | 0 | if (height < 0) { |
1656 | 0 | height = -height; |
1657 | 0 | halfheight = (height + 1) >> 1; |
1658 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
1659 | 0 | src_stride_y = -src_stride_y; |
1660 | 0 | } |
1661 | 0 | if (dst_y) { |
1662 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1663 | 0 | } |
1664 | 0 | SetPlane(dst_vu, dst_stride_vu, halfwidth * 2, halfheight, 128); |
1665 | 0 | return 0; |
1666 | 0 | } |
1667 | | |
1668 | | // Convert NV12 to I420. |
1669 | | // TODO(fbarchard): Consider inverting destination. Faster on ARM with prfm. |
1670 | | LIBYUV_API |
1671 | | int NV12ToI420(const uint8_t* src_y, |
1672 | | int src_stride_y, |
1673 | | const uint8_t* src_uv, |
1674 | | int src_stride_uv, |
1675 | | uint8_t* dst_y, |
1676 | | int dst_stride_y, |
1677 | | uint8_t* dst_u, |
1678 | | int dst_stride_u, |
1679 | | uint8_t* dst_v, |
1680 | | int dst_stride_v, |
1681 | | int width, |
1682 | 0 | int height) { |
1683 | 0 | int halfwidth = (width + 1) >> 1; |
1684 | 0 | int halfheight = (height + 1) >> 1; |
1685 | 0 | if ((!src_y && dst_y) || !src_uv || !dst_u || !dst_v || width <= 0 || |
1686 | 0 | height == 0 || height == INT_MIN) { |
1687 | 0 | return -1; |
1688 | 0 | } |
1689 | | // Negative height means invert the image. |
1690 | 0 | if (height < 0) { |
1691 | 0 | height = -height; |
1692 | 0 | halfheight = (height + 1) >> 1; |
1693 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
1694 | 0 | src_uv = src_uv + (ptrdiff_t)(halfheight - 1) * src_stride_uv; |
1695 | 0 | src_stride_y = -src_stride_y; |
1696 | 0 | src_stride_uv = -src_stride_uv; |
1697 | 0 | } |
1698 | | // Coalesce rows. |
1699 | 0 | if (src_stride_y == width && dst_stride_y == width && |
1700 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1701 | 0 | width *= height; |
1702 | 0 | height = 1; |
1703 | 0 | src_stride_y = dst_stride_y = 0; |
1704 | 0 | } |
1705 | | // Coalesce rows. |
1706 | 0 | if (src_stride_uv == halfwidth * 2 && dst_stride_u == halfwidth && |
1707 | 0 | dst_stride_v == halfwidth && |
1708 | 0 | (ptrdiff_t)halfwidth * halfheight <= INT_MAX) { |
1709 | 0 | halfwidth *= halfheight; |
1710 | 0 | halfheight = 1; |
1711 | 0 | src_stride_uv = dst_stride_u = dst_stride_v = 0; |
1712 | 0 | } |
1713 | |
|
1714 | 0 | if (dst_y) { |
1715 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1716 | 0 | } |
1717 | | |
1718 | | // Split UV plane - NV12 / NV21 |
1719 | 0 | SplitUVPlane(src_uv, src_stride_uv, dst_u, dst_stride_u, dst_v, dst_stride_v, |
1720 | 0 | halfwidth, halfheight); |
1721 | |
|
1722 | 0 | return 0; |
1723 | 0 | } |
1724 | | |
1725 | | // Convert NV21 to I420. Same as NV12 but u and v pointers swapped. |
1726 | | LIBYUV_API |
1727 | | int NV21ToI420(const uint8_t* src_y, |
1728 | | int src_stride_y, |
1729 | | const uint8_t* src_vu, |
1730 | | int src_stride_vu, |
1731 | | uint8_t* dst_y, |
1732 | | int dst_stride_y, |
1733 | | uint8_t* dst_u, |
1734 | | int dst_stride_u, |
1735 | | uint8_t* dst_v, |
1736 | | int dst_stride_v, |
1737 | | int width, |
1738 | 0 | int height) { |
1739 | 0 | return NV12ToI420(src_y, src_stride_y, src_vu, src_stride_vu, dst_y, |
1740 | 0 | dst_stride_y, dst_v, dst_stride_v, dst_u, dst_stride_u, |
1741 | 0 | width, height); |
1742 | 0 | } |
1743 | | |
1744 | | LIBYUV_API |
1745 | | int NV12ToNV24(const uint8_t* src_y, |
1746 | | int src_stride_y, |
1747 | | const uint8_t* src_uv, |
1748 | | int src_stride_uv, |
1749 | | uint8_t* dst_y, |
1750 | | int dst_stride_y, |
1751 | | uint8_t* dst_uv, |
1752 | | int dst_stride_uv, |
1753 | | int width, |
1754 | 0 | int height) { |
1755 | 0 | int r; |
1756 | 0 | if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 || |
1757 | 0 | height == INT_MIN) { |
1758 | 0 | return -1; |
1759 | 0 | } |
1760 | | |
1761 | 0 | if (dst_y) { |
1762 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1763 | 0 | } |
1764 | 0 | r = UVScale(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1), |
1765 | 0 | SUBSAMPLE(height, 1, 1), dst_uv, dst_stride_uv, Abs(width), |
1766 | 0 | Abs(height), kFilterBilinear); |
1767 | 0 | return r; |
1768 | 0 | } |
1769 | | |
1770 | | LIBYUV_API |
1771 | | int NV16ToNV24(const uint8_t* src_y, |
1772 | | int src_stride_y, |
1773 | | const uint8_t* src_uv, |
1774 | | int src_stride_uv, |
1775 | | uint8_t* dst_y, |
1776 | | int dst_stride_y, |
1777 | | uint8_t* dst_uv, |
1778 | | int dst_stride_uv, |
1779 | | int width, |
1780 | 0 | int height) { |
1781 | 0 | int r; |
1782 | 0 | if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 || |
1783 | 0 | height == INT_MIN) { |
1784 | 0 | return -1; |
1785 | 0 | } |
1786 | | |
1787 | 0 | if (dst_y) { |
1788 | 0 | CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1789 | 0 | } |
1790 | 0 | r = UVScale(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1), height, dst_uv, |
1791 | 0 | dst_stride_uv, Abs(width), Abs(height), kFilterBilinear); |
1792 | 0 | return r; |
1793 | 0 | } |
1794 | | |
1795 | | // Any P[420]1[02] to I[420]1[02] format |
1796 | | static int PxxxToIxxx(const uint16_t* src_y, |
1797 | | int src_stride_y, |
1798 | | const uint16_t* src_uv, |
1799 | | int src_stride_uv, |
1800 | | uint16_t* dst_y, |
1801 | | int dst_stride_y, |
1802 | | uint16_t* dst_u, |
1803 | | int dst_stride_u, |
1804 | | uint16_t* dst_v, |
1805 | | int dst_stride_v, |
1806 | | int width, |
1807 | | int height, |
1808 | | int subsample_x, |
1809 | | int subsample_y, |
1810 | 0 | int depth) { |
1811 | 0 | const int uv_width = SUBSAMPLE(width, subsample_x, subsample_x); |
1812 | 0 | const int uv_height = SUBSAMPLE(height, subsample_y, subsample_y); |
1813 | 0 | if (!src_y || !dst_y || !src_uv || !dst_u || !dst_v || width <= 0 || |
1814 | 0 | height == 0 || height == INT_MIN) { |
1815 | 0 | return -1; |
1816 | 0 | } |
1817 | 0 | ConvertToLSBPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height, |
1818 | 0 | depth); |
1819 | 0 | SplitUVPlane_16(src_uv, src_stride_uv, dst_u, dst_stride_u, dst_v, |
1820 | 0 | dst_stride_v, uv_width, uv_height, depth); |
1821 | 0 | return 0; |
1822 | 0 | } |
1823 | | |
1824 | | LIBYUV_API |
1825 | | int P010ToI010(const uint16_t* src_y, |
1826 | | int src_stride_y, |
1827 | | const uint16_t* src_uv, |
1828 | | int src_stride_uv, |
1829 | | uint16_t* dst_y, |
1830 | | int dst_stride_y, |
1831 | | uint16_t* dst_u, |
1832 | | int dst_stride_u, |
1833 | | uint16_t* dst_v, |
1834 | | int dst_stride_v, |
1835 | | int width, |
1836 | 0 | int height) { |
1837 | 0 | return PxxxToIxxx(src_y, src_stride_y, src_uv, src_stride_uv, dst_y, |
1838 | 0 | dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v, |
1839 | 0 | width, height, 1, 1, 10); |
1840 | 0 | } |
1841 | | |
1842 | | LIBYUV_API |
1843 | | int P012ToI012(const uint16_t* src_y, |
1844 | | int src_stride_y, |
1845 | | const uint16_t* src_uv, |
1846 | | int src_stride_uv, |
1847 | | uint16_t* dst_y, |
1848 | | int dst_stride_y, |
1849 | | uint16_t* dst_u, |
1850 | | int dst_stride_u, |
1851 | | uint16_t* dst_v, |
1852 | | int dst_stride_v, |
1853 | | int width, |
1854 | 0 | int height) { |
1855 | 0 | return PxxxToIxxx(src_y, src_stride_y, src_uv, src_stride_uv, dst_y, |
1856 | 0 | dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v, |
1857 | 0 | width, height, 1, 1, 12); |
1858 | 0 | } |
1859 | | |
1860 | | LIBYUV_API |
1861 | | int P010ToP410(const uint16_t* src_y, |
1862 | | int src_stride_y, |
1863 | | const uint16_t* src_uv, |
1864 | | int src_stride_uv, |
1865 | | uint16_t* dst_y, |
1866 | | int dst_stride_y, |
1867 | | uint16_t* dst_uv, |
1868 | | int dst_stride_uv, |
1869 | | int width, |
1870 | 0 | int height) { |
1871 | 0 | int r; |
1872 | 0 | if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 || |
1873 | 0 | height == INT_MIN) { |
1874 | 0 | return -1; |
1875 | 0 | } |
1876 | | |
1877 | 0 | if (dst_y) { |
1878 | 0 | CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1879 | 0 | } |
1880 | 0 | r = UVScale_16(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1), |
1881 | 0 | SUBSAMPLE(height, 1, 1), dst_uv, dst_stride_uv, Abs(width), |
1882 | 0 | Abs(height), kFilterBilinear); |
1883 | 0 | return r; |
1884 | 0 | } |
1885 | | |
1886 | | LIBYUV_API |
1887 | | int P210ToP410(const uint16_t* src_y, |
1888 | | int src_stride_y, |
1889 | | const uint16_t* src_uv, |
1890 | | int src_stride_uv, |
1891 | | uint16_t* dst_y, |
1892 | | int dst_stride_y, |
1893 | | uint16_t* dst_uv, |
1894 | | int dst_stride_uv, |
1895 | | int width, |
1896 | 0 | int height) { |
1897 | 0 | int r; |
1898 | 0 | if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 || |
1899 | 0 | height == INT_MIN) { |
1900 | 0 | return -1; |
1901 | 0 | } |
1902 | | |
1903 | 0 | if (dst_y) { |
1904 | 0 | CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height); |
1905 | 0 | } |
1906 | 0 | r = UVScale_16(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1), height, dst_uv, |
1907 | 0 | dst_stride_uv, Abs(width), Abs(height), kFilterBilinear); |
1908 | 0 | return r; |
1909 | 0 | } |
1910 | | |
1911 | | // Convert YUY2 to I420. |
1912 | | LIBYUV_API |
1913 | | int YUY2ToI420(const uint8_t* src_yuy2, |
1914 | | int src_stride_yuy2, |
1915 | | uint8_t* dst_y, |
1916 | | int dst_stride_y, |
1917 | | uint8_t* dst_u, |
1918 | | int dst_stride_u, |
1919 | | uint8_t* dst_v, |
1920 | | int dst_stride_v, |
1921 | | int width, |
1922 | 0 | int height) { |
1923 | 0 | int y; |
1924 | 0 | void (*YUY2ToUVRow)(const uint8_t* src_yuy2, int src_stride_yuy2, |
1925 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width) = |
1926 | 0 | YUY2ToUVRow_C; |
1927 | 0 | void (*YUY2ToYRow)(const uint8_t* src_yuy2, uint8_t* dst_y, int width) = |
1928 | 0 | YUY2ToYRow_C; |
1929 | 0 | if (width <= 0 || height == 0 || height == INT_MIN) { |
1930 | 0 | return -1; |
1931 | 0 | } |
1932 | | // Negative height means invert the image. |
1933 | 0 | if (height < 0) { |
1934 | 0 | height = -height; |
1935 | 0 | src_yuy2 = src_yuy2 + (ptrdiff_t)(height - 1) * src_stride_yuy2; |
1936 | 0 | src_stride_yuy2 = -src_stride_yuy2; |
1937 | 0 | } |
1938 | 0 | #if defined(HAS_YUY2TOYROW_SSE2) |
1939 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
1940 | 0 | YUY2ToUVRow = YUY2ToUVRow_Any_SSE2; |
1941 | 0 | YUY2ToYRow = YUY2ToYRow_Any_SSE2; |
1942 | 0 | if (IS_ALIGNED(width, 16)) { |
1943 | 0 | YUY2ToUVRow = YUY2ToUVRow_SSE2; |
1944 | 0 | YUY2ToYRow = YUY2ToYRow_SSE2; |
1945 | 0 | } |
1946 | 0 | } |
1947 | 0 | #endif |
1948 | 0 | #if defined(HAS_YUY2TOYROW_AVX2) |
1949 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1950 | 0 | YUY2ToUVRow = YUY2ToUVRow_Any_AVX2; |
1951 | 0 | YUY2ToYRow = YUY2ToYRow_Any_AVX2; |
1952 | 0 | if (IS_ALIGNED(width, 32)) { |
1953 | 0 | YUY2ToUVRow = YUY2ToUVRow_AVX2; |
1954 | 0 | YUY2ToYRow = YUY2ToYRow_AVX2; |
1955 | 0 | } |
1956 | 0 | } |
1957 | 0 | #endif |
1958 | | #if defined(HAS_YUY2TOYROW_NEON) |
1959 | | if (TestCpuFlag(kCpuHasNEON)) { |
1960 | | YUY2ToYRow = YUY2ToYRow_Any_NEON; |
1961 | | YUY2ToUVRow = YUY2ToUVRow_Any_NEON; |
1962 | | if (IS_ALIGNED(width, 16)) { |
1963 | | YUY2ToYRow = YUY2ToYRow_NEON; |
1964 | | YUY2ToUVRow = YUY2ToUVRow_NEON; |
1965 | | } |
1966 | | } |
1967 | | #endif |
1968 | | #if defined(HAS_YUY2TOYROW_SVE2) && defined(HAS_YUY2TOUVROW_SVE2) |
1969 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1970 | | YUY2ToYRow = YUY2ToYRow_SVE2; |
1971 | | YUY2ToUVRow = YUY2ToUVRow_SVE2; |
1972 | | } |
1973 | | #endif |
1974 | | #if defined(HAS_YUY2TOYROW_LSX) && defined(HAS_YUY2TOUVROW_LSX) |
1975 | | if (TestCpuFlag(kCpuHasLSX)) { |
1976 | | YUY2ToYRow = YUY2ToYRow_Any_LSX; |
1977 | | YUY2ToUVRow = YUY2ToUVRow_Any_LSX; |
1978 | | if (IS_ALIGNED(width, 16)) { |
1979 | | YUY2ToYRow = YUY2ToYRow_LSX; |
1980 | | YUY2ToUVRow = YUY2ToUVRow_LSX; |
1981 | | } |
1982 | | } |
1983 | | #endif |
1984 | | #if defined(HAS_YUY2TOYROW_LASX) && defined(HAS_YUY2TOUVROW_LASX) |
1985 | | if (TestCpuFlag(kCpuHasLASX)) { |
1986 | | YUY2ToYRow = YUY2ToYRow_Any_LASX; |
1987 | | YUY2ToUVRow = YUY2ToUVRow_Any_LASX; |
1988 | | if (IS_ALIGNED(width, 32)) { |
1989 | | YUY2ToYRow = YUY2ToYRow_LASX; |
1990 | | YUY2ToUVRow = YUY2ToUVRow_LASX; |
1991 | | } |
1992 | | } |
1993 | | #endif |
1994 | |
|
1995 | 0 | for (y = 0; y < height - 1; y += 2) { |
1996 | 0 | YUY2ToUVRow(src_yuy2, src_stride_yuy2, dst_u, dst_v, width); |
1997 | 0 | YUY2ToYRow(src_yuy2, dst_y, width); |
1998 | 0 | YUY2ToYRow(src_yuy2 + src_stride_yuy2, dst_y + dst_stride_y, width); |
1999 | 0 | src_yuy2 += src_stride_yuy2 * 2; |
2000 | 0 | dst_y += dst_stride_y * 2; |
2001 | 0 | dst_u += dst_stride_u; |
2002 | 0 | dst_v += dst_stride_v; |
2003 | 0 | } |
2004 | 0 | if (height & 1) { |
2005 | 0 | YUY2ToUVRow(src_yuy2, 0, dst_u, dst_v, width); |
2006 | 0 | YUY2ToYRow(src_yuy2, dst_y, width); |
2007 | 0 | } |
2008 | 0 | return 0; |
2009 | 0 | } |
2010 | | |
2011 | | // Convert UYVY to I420. |
2012 | | LIBYUV_API |
2013 | | int UYVYToI420(const uint8_t* src_uyvy, |
2014 | | int src_stride_uyvy, |
2015 | | uint8_t* dst_y, |
2016 | | int dst_stride_y, |
2017 | | uint8_t* dst_u, |
2018 | | int dst_stride_u, |
2019 | | uint8_t* dst_v, |
2020 | | int dst_stride_v, |
2021 | | int width, |
2022 | 0 | int height) { |
2023 | 0 | int y; |
2024 | 0 | void (*UYVYToUVRow)(const uint8_t* src_uyvy, int src_stride_uyvy, |
2025 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width) = |
2026 | 0 | UYVYToUVRow_C; |
2027 | 0 | void (*UYVYToYRow)(const uint8_t* src_uyvy, uint8_t* dst_y, int width) = |
2028 | 0 | UYVYToYRow_C; |
2029 | 0 | if (width <= 0 || height == 0 || height == INT_MIN) { |
2030 | 0 | return -1; |
2031 | 0 | } |
2032 | | // Negative height means invert the image. |
2033 | 0 | if (height < 0) { |
2034 | 0 | height = -height; |
2035 | 0 | src_uyvy = src_uyvy + (ptrdiff_t)(height - 1) * src_stride_uyvy; |
2036 | 0 | src_stride_uyvy = -src_stride_uyvy; |
2037 | 0 | } |
2038 | 0 | #if defined(HAS_UYVYTOYROW_SSE2) |
2039 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
2040 | 0 | UYVYToUVRow = UYVYToUVRow_Any_SSE2; |
2041 | 0 | UYVYToYRow = UYVYToYRow_Any_SSE2; |
2042 | 0 | if (IS_ALIGNED(width, 16)) { |
2043 | 0 | UYVYToUVRow = UYVYToUVRow_SSE2; |
2044 | 0 | UYVYToYRow = UYVYToYRow_SSE2; |
2045 | 0 | } |
2046 | 0 | } |
2047 | 0 | #endif |
2048 | 0 | #if defined(HAS_UYVYTOYROW_AVX2) |
2049 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2050 | 0 | UYVYToUVRow = UYVYToUVRow_Any_AVX2; |
2051 | 0 | UYVYToYRow = UYVYToYRow_Any_AVX2; |
2052 | 0 | if (IS_ALIGNED(width, 32)) { |
2053 | 0 | UYVYToUVRow = UYVYToUVRow_AVX2; |
2054 | 0 | UYVYToYRow = UYVYToYRow_AVX2; |
2055 | 0 | } |
2056 | 0 | } |
2057 | 0 | #endif |
2058 | | #if defined(HAS_UYVYTOYROW_NEON) |
2059 | | if (TestCpuFlag(kCpuHasNEON)) { |
2060 | | UYVYToYRow = UYVYToYRow_Any_NEON; |
2061 | | UYVYToUVRow = UYVYToUVRow_Any_NEON; |
2062 | | if (IS_ALIGNED(width, 16)) { |
2063 | | UYVYToYRow = UYVYToYRow_NEON; |
2064 | | UYVYToUVRow = UYVYToUVRow_NEON; |
2065 | | } |
2066 | | } |
2067 | | #endif |
2068 | | #if defined(HAS_UYVYTOYROW_SVE2) && defined(HAS_UYVYTOUVROW_SVE2) |
2069 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2070 | | UYVYToYRow = UYVYToYRow_SVE2; |
2071 | | UYVYToUVRow = UYVYToUVRow_SVE2; |
2072 | | } |
2073 | | #endif |
2074 | | #if defined(HAS_UYVYTOYROW_LSX) |
2075 | | if (TestCpuFlag(kCpuHasLSX)) { |
2076 | | UYVYToYRow = UYVYToYRow_Any_LSX; |
2077 | | UYVYToUVRow = UYVYToUVRow_Any_LSX; |
2078 | | if (IS_ALIGNED(width, 16)) { |
2079 | | UYVYToYRow = UYVYToYRow_LSX; |
2080 | | UYVYToUVRow = UYVYToUVRow_LSX; |
2081 | | } |
2082 | | } |
2083 | | #endif |
2084 | | #if defined(HAS_UYVYTOYROW_LSX) |
2085 | | if (TestCpuFlag(kCpuHasLSX)) { |
2086 | | UYVYToYRow = UYVYToYRow_Any_LSX; |
2087 | | UYVYToUVRow = UYVYToUVRow_Any_LSX; |
2088 | | if (IS_ALIGNED(width, 16)) { |
2089 | | UYVYToYRow = UYVYToYRow_LSX; |
2090 | | UYVYToUVRow = UYVYToUVRow_LSX; |
2091 | | } |
2092 | | } |
2093 | | #endif |
2094 | | #if defined(HAS_UYVYTOYROW_LASX) |
2095 | | if (TestCpuFlag(kCpuHasLASX)) { |
2096 | | UYVYToYRow = UYVYToYRow_Any_LASX; |
2097 | | UYVYToUVRow = UYVYToUVRow_Any_LASX; |
2098 | | if (IS_ALIGNED(width, 32)) { |
2099 | | UYVYToYRow = UYVYToYRow_LASX; |
2100 | | UYVYToUVRow = UYVYToUVRow_LASX; |
2101 | | } |
2102 | | } |
2103 | | #endif |
2104 | |
|
2105 | 0 | for (y = 0; y < height - 1; y += 2) { |
2106 | 0 | UYVYToUVRow(src_uyvy, src_stride_uyvy, dst_u, dst_v, width); |
2107 | 0 | UYVYToYRow(src_uyvy, dst_y, width); |
2108 | 0 | UYVYToYRow(src_uyvy + src_stride_uyvy, dst_y + dst_stride_y, width); |
2109 | 0 | src_uyvy += src_stride_uyvy * 2; |
2110 | 0 | dst_y += dst_stride_y * 2; |
2111 | 0 | dst_u += dst_stride_u; |
2112 | 0 | dst_v += dst_stride_v; |
2113 | 0 | } |
2114 | 0 | if (height & 1) { |
2115 | 0 | UYVYToUVRow(src_uyvy, 0, dst_u, dst_v, width); |
2116 | 0 | UYVYToYRow(src_uyvy, dst_y, width); |
2117 | 0 | } |
2118 | 0 | return 0; |
2119 | 0 | } |
2120 | | |
2121 | | // Convert AYUV to NV12. |
2122 | | LIBYUV_API |
2123 | | int AYUVToNV12(const uint8_t* src_ayuv, |
2124 | | int src_stride_ayuv, |
2125 | | uint8_t* dst_y, |
2126 | | int dst_stride_y, |
2127 | | uint8_t* dst_uv, |
2128 | | int dst_stride_uv, |
2129 | | int width, |
2130 | 0 | int height) { |
2131 | 0 | int y; |
2132 | 0 | void (*AYUVToUVRow)(const uint8_t* src_ayuv, int src_stride_ayuv, |
2133 | 0 | uint8_t* dst_uv, int width) = AYUVToUVRow_C; |
2134 | 0 | void (*AYUVToYRow)(const uint8_t* src_ayuv, uint8_t* dst_y, int width) = |
2135 | 0 | AYUVToYRow_C; |
2136 | 0 | if (width <= 0 || height == 0 || height == INT_MIN) { |
2137 | 0 | return -1; |
2138 | 0 | } |
2139 | | // Negative height means invert the image. |
2140 | 0 | if (height < 0) { |
2141 | 0 | height = -height; |
2142 | 0 | src_ayuv = src_ayuv + (ptrdiff_t)(height - 1) * src_stride_ayuv; |
2143 | 0 | src_stride_ayuv = -src_stride_ayuv; |
2144 | 0 | } |
2145 | | // place holders for future intel code |
2146 | | #if defined(HAS_AYUVTOYROW_SSE2) |
2147 | | if (TestCpuFlag(kCpuHasSSE2)) { |
2148 | | AYUVToUVRow = AYUVToUVRow_Any_SSE2; |
2149 | | AYUVToYRow = AYUVToYRow_Any_SSE2; |
2150 | | if (IS_ALIGNED(width, 16)) { |
2151 | | AYUVToUVRow = AYUVToUVRow_SSE2; |
2152 | | AYUVToYRow = AYUVToYRow_SSE2; |
2153 | | } |
2154 | | } |
2155 | | #endif |
2156 | | #if defined(HAS_AYUVTOYROW_AVX2) |
2157 | | if (TestCpuFlag(kCpuHasAVX2)) { |
2158 | | AYUVToUVRow = AYUVToUVRow_Any_AVX2; |
2159 | | AYUVToYRow = AYUVToYRow_Any_AVX2; |
2160 | | if (IS_ALIGNED(width, 32)) { |
2161 | | AYUVToUVRow = AYUVToUVRow_AVX2; |
2162 | | AYUVToYRow = AYUVToYRow_AVX2; |
2163 | | } |
2164 | | } |
2165 | | #endif |
2166 | |
|
2167 | | #if defined(HAS_AYUVTOYROW_NEON) |
2168 | | if (TestCpuFlag(kCpuHasNEON)) { |
2169 | | AYUVToYRow = AYUVToYRow_Any_NEON; |
2170 | | AYUVToUVRow = AYUVToUVRow_Any_NEON; |
2171 | | if (IS_ALIGNED(width, 16)) { |
2172 | | AYUVToYRow = AYUVToYRow_NEON; |
2173 | | AYUVToUVRow = AYUVToUVRow_NEON; |
2174 | | } |
2175 | | } |
2176 | | #endif |
2177 | | #if defined(HAS_AYUVTOUVROW_SVE2) |
2178 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2179 | | AYUVToUVRow = AYUVToUVRow_Any_SVE2; |
2180 | | if (IS_ALIGNED(width, 2)) { |
2181 | | AYUVToUVRow = AYUVToUVRow_SVE2; |
2182 | | } |
2183 | | } |
2184 | | #endif |
2185 | |
|
2186 | 0 | for (y = 0; y < height - 1; y += 2) { |
2187 | 0 | AYUVToUVRow(src_ayuv, src_stride_ayuv, dst_uv, width); |
2188 | 0 | AYUVToYRow(src_ayuv, dst_y, width); |
2189 | 0 | AYUVToYRow(src_ayuv + src_stride_ayuv, dst_y + dst_stride_y, width); |
2190 | 0 | src_ayuv += src_stride_ayuv * 2; |
2191 | 0 | dst_y += dst_stride_y * 2; |
2192 | 0 | dst_uv += dst_stride_uv; |
2193 | 0 | } |
2194 | 0 | if (height & 1) { |
2195 | 0 | AYUVToUVRow(src_ayuv, 0, dst_uv, width); |
2196 | 0 | AYUVToYRow(src_ayuv, dst_y, width); |
2197 | 0 | } |
2198 | 0 | return 0; |
2199 | 0 | } |
2200 | | |
2201 | | // Convert AYUV to NV21. |
2202 | | LIBYUV_API |
2203 | | int AYUVToNV21(const uint8_t* src_ayuv, |
2204 | | int src_stride_ayuv, |
2205 | | uint8_t* dst_y, |
2206 | | int dst_stride_y, |
2207 | | uint8_t* dst_vu, |
2208 | | int dst_stride_vu, |
2209 | | int width, |
2210 | 0 | int height) { |
2211 | 0 | int y; |
2212 | 0 | void (*AYUVToVURow)(const uint8_t* src_ayuv, int src_stride_ayuv, |
2213 | 0 | uint8_t* dst_vu, int width) = AYUVToVURow_C; |
2214 | 0 | void (*AYUVToYRow)(const uint8_t* src_ayuv, uint8_t* dst_y, int width) = |
2215 | 0 | AYUVToYRow_C; |
2216 | 0 | if (width <= 0 || height == 0 || height == INT_MIN) { |
2217 | 0 | return -1; |
2218 | 0 | } |
2219 | | // Negative height means invert the image. |
2220 | 0 | if (height < 0) { |
2221 | 0 | height = -height; |
2222 | 0 | src_ayuv = src_ayuv + (ptrdiff_t)(height - 1) * src_stride_ayuv; |
2223 | 0 | src_stride_ayuv = -src_stride_ayuv; |
2224 | 0 | } |
2225 | | // place holders for future intel code |
2226 | | #if defined(HAS_AYUVTOYROW_SSE2) |
2227 | | if (TestCpuFlag(kCpuHasSSE2)) { |
2228 | | AYUVToVURow = AYUVToVURow_Any_SSE2; |
2229 | | AYUVToYRow = AYUVToYRow_Any_SSE2; |
2230 | | if (IS_ALIGNED(width, 16)) { |
2231 | | AYUVToVURow = AYUVToVURow_SSE2; |
2232 | | AYUVToYRow = AYUVToYRow_SSE2; |
2233 | | } |
2234 | | } |
2235 | | #endif |
2236 | | #if defined(HAS_AYUVTOYROW_AVX2) |
2237 | | if (TestCpuFlag(kCpuHasAVX2)) { |
2238 | | AYUVToVURow = AYUVToVURow_Any_AVX2; |
2239 | | AYUVToYRow = AYUVToYRow_Any_AVX2; |
2240 | | if (IS_ALIGNED(width, 32)) { |
2241 | | AYUVToVURow = AYUVToVURow_AVX2; |
2242 | | AYUVToYRow = AYUVToYRow_AVX2; |
2243 | | } |
2244 | | } |
2245 | | #endif |
2246 | |
|
2247 | | #if defined(HAS_AYUVTOYROW_NEON) |
2248 | | if (TestCpuFlag(kCpuHasNEON)) { |
2249 | | AYUVToYRow = AYUVToYRow_Any_NEON; |
2250 | | AYUVToVURow = AYUVToVURow_Any_NEON; |
2251 | | if (IS_ALIGNED(width, 16)) { |
2252 | | AYUVToYRow = AYUVToYRow_NEON; |
2253 | | AYUVToVURow = AYUVToVURow_NEON; |
2254 | | } |
2255 | | } |
2256 | | #endif |
2257 | | #if defined(HAS_AYUVTOVUROW_SVE2) |
2258 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2259 | | AYUVToVURow = AYUVToVURow_Any_SVE2; |
2260 | | if (IS_ALIGNED(width, 2)) { |
2261 | | AYUVToVURow = AYUVToVURow_SVE2; |
2262 | | } |
2263 | | } |
2264 | | #endif |
2265 | |
|
2266 | 0 | for (y = 0; y < height - 1; y += 2) { |
2267 | 0 | AYUVToVURow(src_ayuv, src_stride_ayuv, dst_vu, width); |
2268 | 0 | AYUVToYRow(src_ayuv, dst_y, width); |
2269 | 0 | AYUVToYRow(src_ayuv + src_stride_ayuv, dst_y + dst_stride_y, width); |
2270 | 0 | src_ayuv += src_stride_ayuv * 2; |
2271 | 0 | dst_y += dst_stride_y * 2; |
2272 | 0 | dst_vu += dst_stride_vu; |
2273 | 0 | } |
2274 | 0 | if (height & 1) { |
2275 | 0 | AYUVToVURow(src_ayuv, 0, dst_vu, width); |
2276 | 0 | AYUVToYRow(src_ayuv, dst_y, width); |
2277 | 0 | } |
2278 | 0 | return 0; |
2279 | 0 | } |
2280 | | |
2281 | | // Convert ARGB to I420. |
2282 | | LIBYUV_API |
2283 | | int ARGBToI420(const uint8_t* src_argb, |
2284 | | int src_stride_argb, |
2285 | | uint8_t* dst_y, |
2286 | | int dst_stride_y, |
2287 | | uint8_t* dst_u, |
2288 | | int dst_stride_u, |
2289 | | uint8_t* dst_v, |
2290 | | int dst_stride_v, |
2291 | | int width, |
2292 | 0 | int height) { |
2293 | 0 | return ARGBToI420Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u, |
2294 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2295 | 0 | &kArgbI601Constants, width, height); |
2296 | 0 | } |
2297 | | |
2298 | | LIBYUV_API |
2299 | | int ARGBToI420Matrix(const uint8_t* src_argb, |
2300 | | int src_stride_argb, |
2301 | | uint8_t* dst_y, |
2302 | | int dst_stride_y, |
2303 | | uint8_t* dst_u, |
2304 | | int dst_stride_u, |
2305 | | uint8_t* dst_v, |
2306 | | int dst_stride_v, |
2307 | | const struct ArgbConstants* argbconstants, |
2308 | | int width, |
2309 | 5 | int height) { |
2310 | 5 | int y; |
2311 | 5 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
2312 | 5 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
2313 | 5 | void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb, |
2314 | 5 | uint8_t* dst_u, uint8_t* dst_v, int width, |
2315 | 5 | const struct ArgbConstants* c) = |
2316 | 5 | ARGBToUVMatrixRow_C; |
2317 | | |
2318 | 5 | #if defined(HAS_ARGBTOYMATRIXROW_SSSE3) |
2319 | 5 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2320 | 5 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3; |
2321 | 5 | if (IS_ALIGNED(width, 16)) { |
2322 | 2 | ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3; |
2323 | 2 | } |
2324 | 5 | } |
2325 | 5 | #endif |
2326 | 5 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
2327 | 5 | if (TestCpuFlag(kCpuHasAVX2)) { |
2328 | 5 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
2329 | 5 | if (IS_ALIGNED(width, 32)) { |
2330 | 2 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
2331 | 2 | } |
2332 | 5 | } |
2333 | 5 | #endif |
2334 | 5 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
2335 | 5 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2336 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
2337 | 0 | if (IS_ALIGNED(width, 64)) { |
2338 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
2339 | 0 | } |
2340 | 0 | } |
2341 | 5 | #endif |
2342 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
2343 | | if (TestCpuFlag(kCpuHasNEON)) { |
2344 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
2345 | | if (IS_ALIGNED(width, 16)) { |
2346 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
2347 | | } |
2348 | | } |
2349 | | #endif |
2350 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
2351 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
2352 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
2353 | | if (IS_ALIGNED(width, 16)) { |
2354 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
2355 | | } |
2356 | | } |
2357 | | #endif |
2358 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
2359 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2360 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
2361 | | } |
2362 | | #endif |
2363 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
2364 | | if (TestCpuFlag(kCpuHasSME)) { |
2365 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
2366 | | } |
2367 | | #endif |
2368 | | #if defined(HAS_ARGBTOYMATRIXROW_LSX) |
2369 | | if (TestCpuFlag(kCpuHasLSX)) { |
2370 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX; |
2371 | | if (IS_ALIGNED(width, 16)) { |
2372 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LSX; |
2373 | | } |
2374 | | } |
2375 | | #endif |
2376 | | #if defined(HAS_ARGBTOYMATRIXROW_LASX) |
2377 | | if (TestCpuFlag(kCpuHasLASX)) { |
2378 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX; |
2379 | | if (IS_ALIGNED(width, 32)) { |
2380 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LASX; |
2381 | | } |
2382 | | } |
2383 | | #endif |
2384 | | #if defined(HAS_ARGBTOYMATRIXROW_RVV) |
2385 | | if (TestCpuFlag(kCpuHasRVV)) { |
2386 | | ARGBToYMatrixRow = ARGBToYMatrixRow_RVV; |
2387 | | } |
2388 | | #endif |
2389 | | |
2390 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON) |
2391 | | if (TestCpuFlag(kCpuHasNEON)) { |
2392 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON; |
2393 | | if (IS_ALIGNED(width, 16)) { |
2394 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON; |
2395 | | } |
2396 | | } |
2397 | | #endif |
2398 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM) |
2399 | | if (TestCpuFlag(kCpuHasNEON) && TestCpuFlag(kCpuHasNeonI8MM)) { |
2400 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM; |
2401 | | if (IS_ALIGNED(width, 16)) { |
2402 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM; |
2403 | | } |
2404 | | } |
2405 | | #endif |
2406 | | #if defined(HAS_ARGBTOUVMATRIXROW_SVE2) |
2407 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2408 | | if (IS_ALIGNED(width, 2)) { |
2409 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2; |
2410 | | } |
2411 | | } |
2412 | | #endif |
2413 | | #if defined(HAS_ARGBTOUVMATRIXROW_SME) |
2414 | | if (TestCpuFlag(kCpuHasSME)) { |
2415 | | if (IS_ALIGNED(width, 2)) { |
2416 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME; |
2417 | | } |
2418 | | } |
2419 | | #endif |
2420 | 5 | #if defined(HAS_ARGBTOUVMATRIXROW_SSSE3) |
2421 | 5 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2422 | 5 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3; |
2423 | 5 | if (IS_ALIGNED(width, 8)) { |
2424 | 2 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3; |
2425 | 2 | } |
2426 | 5 | } |
2427 | 5 | #endif |
2428 | 5 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX2) |
2429 | 5 | if (TestCpuFlag(kCpuHasAVX2)) { |
2430 | 5 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2; |
2431 | 5 | if (IS_ALIGNED(width, 32)) { |
2432 | 2 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2; |
2433 | 2 | } |
2434 | 5 | } |
2435 | 5 | #endif |
2436 | 5 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW) |
2437 | 5 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2438 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX512BW; |
2439 | 0 | if (IS_ALIGNED(width, 64)) { |
2440 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW; |
2441 | 0 | } |
2442 | 0 | } |
2443 | 5 | #endif |
2444 | | #if defined(HAS_ARGBTOUVMATRIXROW_RVV) |
2445 | | if (TestCpuFlag(kCpuHasRVV)) { |
2446 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV; |
2447 | | } |
2448 | | #endif |
2449 | 5 | if (!src_argb || !dst_y || !dst_u || !dst_v || !argbconstants || width <= 0 || |
2450 | 5 | height == 0 || height == INT_MIN) { |
2451 | 0 | return -1; |
2452 | 0 | } |
2453 | | // Negative height means invert the image. |
2454 | 5 | if (height < 0) { |
2455 | 0 | height = -height; |
2456 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
2457 | 0 | src_stride_argb = -src_stride_argb; |
2458 | 0 | } |
2459 | | |
2460 | 454 | for (y = 0; y < height - 1; y += 2) { |
2461 | 449 | ARGBToUVMatrixRow(src_argb, src_stride_argb, dst_u, dst_v, width, |
2462 | 449 | argbconstants); |
2463 | 449 | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
2464 | 449 | ARGBToYMatrixRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width, |
2465 | 449 | argbconstants); |
2466 | 449 | src_argb += src_stride_argb * 2; |
2467 | 449 | dst_y += dst_stride_y * 2; |
2468 | 449 | dst_u += dst_stride_u; |
2469 | 449 | dst_v += dst_stride_v; |
2470 | 449 | } |
2471 | 5 | if (height & 1) { |
2472 | 1 | ARGBToUVMatrixRow(src_argb, 0, dst_u, dst_v, width, argbconstants); |
2473 | 1 | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
2474 | 1 | } |
2475 | 5 | return 0; |
2476 | 5 | } |
2477 | | |
2478 | | #ifdef USE_EXTRACTALPHA |
2479 | | // Convert ARGB to I420 with Alpha |
2480 | | // The following version calls ARGBExtractAlpha on the full image. |
2481 | | LIBYUV_API |
2482 | | int ARGBToI420Alpha(const uint8_t* src_argb, |
2483 | | int src_stride_argb, |
2484 | | uint8_t* dst_y, |
2485 | | int dst_stride_y, |
2486 | | uint8_t* dst_u, |
2487 | | int dst_stride_u, |
2488 | | uint8_t* dst_v, |
2489 | | int dst_stride_v, |
2490 | | uint8_t* dst_a, |
2491 | | int dst_stride_a, |
2492 | | int width, |
2493 | | int height) { |
2494 | | int r = ARGBToI420(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u, |
2495 | | dst_stride_u, dst_v, dst_stride_v, width, height); |
2496 | | if (r == 0) { |
2497 | | r = ARGBExtractAlpha(src_argb, src_stride_argb, dst_a, dst_stride_a, width, |
2498 | | height); |
2499 | | } |
2500 | | return r; |
2501 | | } |
2502 | | #else // USE_EXTRACTALPHA |
2503 | | // Convert ARGB to I420 with Alpha |
2504 | | LIBYUV_API |
2505 | | int ARGBToI420Alpha(const uint8_t* src_argb, |
2506 | | int src_stride_argb, |
2507 | | uint8_t* dst_y, |
2508 | | int dst_stride_y, |
2509 | | uint8_t* dst_u, |
2510 | | int dst_stride_u, |
2511 | | uint8_t* dst_v, |
2512 | | int dst_stride_v, |
2513 | | uint8_t* dst_a, |
2514 | | int dst_stride_a, |
2515 | | int width, |
2516 | 0 | int height) { |
2517 | 0 | int y; |
2518 | 0 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
2519 | 0 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
2520 | 0 | void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb, |
2521 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
2522 | 0 | const struct ArgbConstants* c) = |
2523 | 0 | ARGBToUVMatrixRow_C; |
2524 | 0 | void (*ARGBExtractAlphaRow)(const uint8_t* src_argb, uint8_t* dst_a, |
2525 | 0 | int width) = ARGBExtractAlphaRow_C; |
2526 | 0 | if (!src_argb || !dst_y || !dst_u || !dst_v || !dst_a || width <= 0 || |
2527 | 0 | height == 0 || height == INT_MIN) { |
2528 | 0 | return -1; |
2529 | 0 | } |
2530 | | // Negative height means invert the image. |
2531 | 0 | if (height < 0) { |
2532 | 0 | height = -height; |
2533 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
2534 | 0 | src_stride_argb = -src_stride_argb; |
2535 | 0 | } |
2536 | |
|
2537 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_SSSE3) |
2538 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2539 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3; |
2540 | 0 | if (IS_ALIGNED(width, 16)) { |
2541 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3; |
2542 | 0 | } |
2543 | 0 | } |
2544 | 0 | #endif |
2545 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
2546 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2547 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
2548 | 0 | if (IS_ALIGNED(width, 32)) { |
2549 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
2550 | 0 | } |
2551 | 0 | } |
2552 | 0 | #endif |
2553 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
2554 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2555 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
2556 | 0 | if (IS_ALIGNED(width, 64)) { |
2557 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
2558 | 0 | } |
2559 | 0 | } |
2560 | 0 | #endif |
2561 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
2562 | | if (TestCpuFlag(kCpuHasNEON)) { |
2563 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
2564 | | if (IS_ALIGNED(width, 16)) { |
2565 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
2566 | | } |
2567 | | } |
2568 | | #endif |
2569 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
2570 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
2571 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
2572 | | if (IS_ALIGNED(width, 16)) { |
2573 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
2574 | | } |
2575 | | } |
2576 | | #endif |
2577 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
2578 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2579 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
2580 | | } |
2581 | | #endif |
2582 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
2583 | | if (TestCpuFlag(kCpuHasSME)) { |
2584 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
2585 | | } |
2586 | | #endif |
2587 | | #if defined(HAS_ARGBTOYMATRIXROW_LSX) |
2588 | | if (TestCpuFlag(kCpuHasLSX)) { |
2589 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX; |
2590 | | if (IS_ALIGNED(width, 16)) { |
2591 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LSX; |
2592 | | } |
2593 | | } |
2594 | | #endif |
2595 | | #if defined(HAS_ARGBTOYMATRIXROW_LASX) |
2596 | | if (TestCpuFlag(kCpuHasLASX)) { |
2597 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX; |
2598 | | if (IS_ALIGNED(width, 32)) { |
2599 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LASX; |
2600 | | } |
2601 | | } |
2602 | | #endif |
2603 | | #if defined(HAS_ARGBTOYMATRIXROW_RVV) |
2604 | | if (TestCpuFlag(kCpuHasRVV)) { |
2605 | | ARGBToYMatrixRow = ARGBToYMatrixRow_RVV; |
2606 | | } |
2607 | | #endif |
2608 | |
|
2609 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON) |
2610 | | if (TestCpuFlag(kCpuHasNEON)) { |
2611 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON; |
2612 | | if (IS_ALIGNED(width, 16)) { |
2613 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON; |
2614 | | } |
2615 | | } |
2616 | | #endif |
2617 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM) |
2618 | | if (TestCpuFlag(kCpuHasNEON) && TestCpuFlag(kCpuHasNeonI8MM)) { |
2619 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM; |
2620 | | if (IS_ALIGNED(width, 16)) { |
2621 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM; |
2622 | | } |
2623 | | } |
2624 | | #endif |
2625 | | #if defined(HAS_ARGBTOUVMATRIXROW_SVE2) |
2626 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2627 | | if (IS_ALIGNED(width, 2)) { |
2628 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2; |
2629 | | } |
2630 | | } |
2631 | | #endif |
2632 | | #if defined(HAS_ARGBTOUVMATRIXROW_SME) |
2633 | | if (TestCpuFlag(kCpuHasSME)) { |
2634 | | if (IS_ALIGNED(width, 2)) { |
2635 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME; |
2636 | | } |
2637 | | } |
2638 | | #endif |
2639 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_SSSE3) |
2640 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2641 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3; |
2642 | 0 | if (IS_ALIGNED(width, 8)) { |
2643 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3; |
2644 | 0 | } |
2645 | 0 | } |
2646 | 0 | #endif |
2647 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX2) |
2648 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2649 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2; |
2650 | 0 | if (IS_ALIGNED(width, 32)) { |
2651 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2; |
2652 | 0 | } |
2653 | 0 | } |
2654 | 0 | #endif |
2655 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW) |
2656 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2657 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX512BW; |
2658 | 0 | if (IS_ALIGNED(width, 64)) { |
2659 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW; |
2660 | 0 | } |
2661 | 0 | } |
2662 | 0 | #endif |
2663 | | #if defined(HAS_ARGBTOUVMATRIXROW_RVV) |
2664 | | if (TestCpuFlag(kCpuHasRVV)) { |
2665 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV; |
2666 | | } |
2667 | | #endif |
2668 | |
|
2669 | 0 | #if defined(HAS_ARGBEXTRACTALPHAROW_SSE2) |
2670 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
2671 | 0 | ARGBExtractAlphaRow = IS_ALIGNED(width, 8) ? ARGBExtractAlphaRow_SSE2 |
2672 | 0 | : ARGBExtractAlphaRow_Any_SSE2; |
2673 | 0 | } |
2674 | 0 | #endif |
2675 | 0 | #if defined(HAS_ARGBEXTRACTALPHAROW_AVX2) |
2676 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2677 | 0 | ARGBExtractAlphaRow = IS_ALIGNED(width, 32) ? ARGBExtractAlphaRow_AVX2 |
2678 | 0 | : ARGBExtractAlphaRow_Any_AVX2; |
2679 | 0 | } |
2680 | 0 | #endif |
2681 | | #if defined(HAS_ARGBEXTRACTALPHAROW_NEON) |
2682 | | if (TestCpuFlag(kCpuHasNEON)) { |
2683 | | ARGBExtractAlphaRow = IS_ALIGNED(width, 16) ? ARGBExtractAlphaRow_NEON |
2684 | | : ARGBExtractAlphaRow_Any_NEON; |
2685 | | } |
2686 | | #endif |
2687 | | #if defined(HAS_ARGBEXTRACTALPHAROW_LSX) |
2688 | | if (TestCpuFlag(kCpuHasLSX)) { |
2689 | | ARGBExtractAlphaRow = IS_ALIGNED(width, 16) ? ARGBExtractAlphaRow_LSX |
2690 | | : ARGBExtractAlphaRow_Any_LSX; |
2691 | | } |
2692 | | #endif |
2693 | | #if defined(HAS_ARGBEXTRACTALPHAROW_RVV) |
2694 | | if (TestCpuFlag(kCpuHasRVV)) { |
2695 | | ARGBExtractAlphaRow = ARGBExtractAlphaRow_RVV; |
2696 | | } |
2697 | | #endif |
2698 | |
|
2699 | 0 | for (y = 0; y < height - 1; y += 2) { |
2700 | 0 | ARGBToUVMatrixRow(src_argb, src_stride_argb, dst_u, dst_v, width, |
2701 | 0 | &kArgbI601Constants); |
2702 | 0 | ARGBToYMatrixRow(src_argb, dst_y, width, &kArgbI601Constants); |
2703 | 0 | ARGBToYMatrixRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width, |
2704 | 0 | &kArgbI601Constants); |
2705 | 0 | ARGBExtractAlphaRow(src_argb, dst_a, width); |
2706 | 0 | ARGBExtractAlphaRow(src_argb + src_stride_argb, dst_a + dst_stride_a, |
2707 | 0 | width); |
2708 | 0 | src_argb += src_stride_argb * 2; |
2709 | 0 | dst_y += dst_stride_y * 2; |
2710 | 0 | dst_u += dst_stride_u; |
2711 | 0 | dst_v += dst_stride_v; |
2712 | 0 | dst_a += dst_stride_a * 2; |
2713 | 0 | } |
2714 | 0 | if (height & 1) { |
2715 | 0 | ARGBToUVMatrixRow(src_argb, 0, dst_u, dst_v, width, &kArgbI601Constants); |
2716 | 0 | ARGBToYMatrixRow(src_argb, dst_y, width, &kArgbI601Constants); |
2717 | 0 | ARGBExtractAlphaRow(src_argb, dst_a, width); |
2718 | 0 | } |
2719 | 0 | return 0; |
2720 | 0 | } |
2721 | | #endif // USE_EXTRACTALPHA |
2722 | | |
2723 | | // Convert BGRA to I420. |
2724 | | LIBYUV_API |
2725 | | int BGRAToI420(const uint8_t* src_bgra, |
2726 | | int src_stride_bgra, |
2727 | | uint8_t* dst_y, |
2728 | | int dst_stride_y, |
2729 | | uint8_t* dst_u, |
2730 | | int dst_stride_u, |
2731 | | uint8_t* dst_v, |
2732 | | int dst_stride_v, |
2733 | | int width, |
2734 | 0 | int height) { |
2735 | 0 | return ARGBToI420Matrix(src_bgra, src_stride_bgra, dst_y, dst_stride_y, dst_u, |
2736 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2737 | 0 | &kBgraI601Constants, width, height); |
2738 | 0 | } |
2739 | | |
2740 | | // Convert BGRA to I422. |
2741 | | LIBYUV_API |
2742 | | int BGRAToI422(const uint8_t* src_bgra, |
2743 | | int src_stride_bgra, |
2744 | | uint8_t* dst_y, |
2745 | | int dst_stride_y, |
2746 | | uint8_t* dst_u, |
2747 | | int dst_stride_u, |
2748 | | uint8_t* dst_v, |
2749 | | int dst_stride_v, |
2750 | | int width, |
2751 | 0 | int height) { |
2752 | 0 | return ARGBToI422Matrix(src_bgra, src_stride_bgra, dst_y, dst_stride_y, dst_u, |
2753 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2754 | 0 | &kBgraI601Constants, width, height); |
2755 | 0 | } |
2756 | | |
2757 | | // Convert ABGR to I422. |
2758 | | LIBYUV_API |
2759 | | int ABGRToI422(const uint8_t* src_abgr, |
2760 | | int src_stride_abgr, |
2761 | | uint8_t* dst_y, |
2762 | | int dst_stride_y, |
2763 | | uint8_t* dst_u, |
2764 | | int dst_stride_u, |
2765 | | uint8_t* dst_v, |
2766 | | int dst_stride_v, |
2767 | | int width, |
2768 | 0 | int height) { |
2769 | 0 | return ARGBToI422Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, dst_u, |
2770 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2771 | 0 | &kAbgrI601Constants, width, height); |
2772 | 0 | } |
2773 | | |
2774 | | // Convert RGBA to I422. |
2775 | | LIBYUV_API |
2776 | | int RGBAToI422(const uint8_t* src_rgba, |
2777 | | int src_stride_rgba, |
2778 | | uint8_t* dst_y, |
2779 | | int dst_stride_y, |
2780 | | uint8_t* dst_u, |
2781 | | int dst_stride_u, |
2782 | | uint8_t* dst_v, |
2783 | | int dst_stride_v, |
2784 | | int width, |
2785 | 0 | int height) { |
2786 | 0 | return ARGBToI422Matrix(src_rgba, src_stride_rgba, dst_y, dst_stride_y, dst_u, |
2787 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2788 | 0 | &kRgbaI601Constants, width, height); |
2789 | 0 | } |
2790 | | |
2791 | | // Convert ABGR to I420. |
2792 | | LIBYUV_API |
2793 | | int ABGRToI420(const uint8_t* src_abgr, |
2794 | | int src_stride_abgr, |
2795 | | uint8_t* dst_y, |
2796 | | int dst_stride_y, |
2797 | | uint8_t* dst_u, |
2798 | | int dst_stride_u, |
2799 | | uint8_t* dst_v, |
2800 | | int dst_stride_v, |
2801 | | int width, |
2802 | 0 | int height) { |
2803 | 0 | return ARGBToI420Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, dst_u, |
2804 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2805 | 0 | &kAbgrI601Constants, width, height); |
2806 | 0 | } |
2807 | | |
2808 | | // Convert RGBA to I420. |
2809 | | LIBYUV_API |
2810 | | int RGBAToI420(const uint8_t* src_rgba, |
2811 | | int src_stride_rgba, |
2812 | | uint8_t* dst_y, |
2813 | | int dst_stride_y, |
2814 | | uint8_t* dst_u, |
2815 | | int dst_stride_u, |
2816 | | uint8_t* dst_v, |
2817 | | int dst_stride_v, |
2818 | | int width, |
2819 | 0 | int height) { |
2820 | 0 | return ARGBToI420Matrix(src_rgba, src_stride_rgba, dst_y, dst_stride_y, dst_u, |
2821 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2822 | 0 | &kRgbaI601Constants, width, height); |
2823 | 0 | } |
2824 | | |
2825 | | // Enabled if 1 pass is available |
2826 | | #if (defined(HAS_RGB24TOYROW_NEON) || defined(HAS_RGB24TOYROW_LSX) || \ |
2827 | | defined(HAS_RGB24TOYROW_RVV)) |
2828 | | #define HAS_RGB24TOYROW |
2829 | | #endif |
2830 | | |
2831 | | // Convert RGB24 to I420. |
2832 | | LIBYUV_API |
2833 | | int RGB24ToI420(const uint8_t* src_rgb24, |
2834 | | int src_stride_rgb24, |
2835 | | uint8_t* dst_y, |
2836 | | int dst_stride_y, |
2837 | | uint8_t* dst_u, |
2838 | | int dst_stride_u, |
2839 | | uint8_t* dst_v, |
2840 | | int dst_stride_v, |
2841 | | int width, |
2842 | 0 | int height) { |
2843 | 0 | int y; |
2844 | 0 | void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb, |
2845 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
2846 | 0 | const struct ArgbConstants* c) = RGBToUVMatrixRow_C; |
2847 | 0 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
2848 | 0 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
2849 | |
|
2850 | 0 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
2851 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2852 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
2853 | 0 | if (IS_ALIGNED(width, 16)) { |
2854 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
2855 | 0 | } |
2856 | 0 | } |
2857 | 0 | #endif |
2858 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
2859 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2860 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
2861 | 0 | if (IS_ALIGNED(width, 32)) { |
2862 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
2863 | 0 | } |
2864 | 0 | } |
2865 | 0 | #endif |
2866 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
2867 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2868 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
2869 | 0 | if (IS_ALIGNED(width, 64)) { |
2870 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
2871 | 0 | } |
2872 | 0 | } |
2873 | 0 | #endif |
2874 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_SSSE3) |
2875 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2876 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3; |
2877 | 0 | if (IS_ALIGNED(width, 16)) { |
2878 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3; |
2879 | 0 | } |
2880 | 0 | } |
2881 | 0 | #endif |
2882 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX2) |
2883 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2884 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2; |
2885 | 0 | if (IS_ALIGNED(width, 32)) { |
2886 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2; |
2887 | 0 | } |
2888 | 0 | } |
2889 | 0 | #endif |
2890 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX512BW) |
2891 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2892 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX512BW; |
2893 | 0 | if (IS_ALIGNED(width, 64)) { |
2894 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW; |
2895 | 0 | } |
2896 | 0 | } |
2897 | 0 | #endif |
2898 | | #if defined(HAS_RGBTOUVMATRIXROW_NEON) |
2899 | | if (TestCpuFlag(kCpuHasNEON)) { |
2900 | | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON; |
2901 | | if (IS_ALIGNED(width, 16)) { |
2902 | | RGBToUVMatrixRow = RGBToUVMatrixRow_NEON; |
2903 | | } |
2904 | | } |
2905 | | #endif |
2906 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
2907 | | if (TestCpuFlag(kCpuHasNEON)) { |
2908 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
2909 | | if (IS_ALIGNED(width, 16)) { |
2910 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
2911 | | } |
2912 | | } |
2913 | | #endif |
2914 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
2915 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
2916 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
2917 | | if (IS_ALIGNED(width, 16)) { |
2918 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
2919 | | } |
2920 | | } |
2921 | | #endif |
2922 | | #if defined(HAS_RGBTOUVMATRIXROW_RVV) |
2923 | | if (TestCpuFlag(kCpuHasRVV)) { |
2924 | | RGBToUVMatrixRow = RGBToUVMatrixRow_RVV; |
2925 | | } |
2926 | | #endif |
2927 | | #if defined(HAS_RGBTOYMATRIXROW_RVV) |
2928 | | if (TestCpuFlag(kCpuHasRVV)) { |
2929 | | RGBToYMatrixRow = RGBToYMatrixRow_RVV; |
2930 | | } |
2931 | | #endif |
2932 | |
|
2933 | 0 | if (!src_rgb24 || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 || |
2934 | 0 | height == INT_MIN) { |
2935 | 0 | return -1; |
2936 | 0 | } |
2937 | | // Negative height means invert the image. |
2938 | 0 | if (height < 0) { |
2939 | 0 | height = -height; |
2940 | 0 | src_rgb24 = src_rgb24 + (ptrdiff_t)(height - 1) * src_stride_rgb24; |
2941 | 0 | src_stride_rgb24 = -src_stride_rgb24; |
2942 | 0 | } |
2943 | |
|
2944 | 0 | for (y = 0; y < height - 1; y += 2) { |
2945 | 0 | RGBToUVMatrixRow(src_rgb24, src_stride_rgb24, dst_u, dst_v, width, |
2946 | 0 | &kArgbI601Constants); |
2947 | 0 | RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbI601Constants); |
2948 | 0 | RGBToYMatrixRow(src_rgb24 + src_stride_rgb24, dst_y + dst_stride_y, width, |
2949 | 0 | &kArgbI601Constants); |
2950 | 0 | src_rgb24 += src_stride_rgb24 * 2; |
2951 | 0 | dst_y += dst_stride_y * 2; |
2952 | 0 | dst_u += dst_stride_u; |
2953 | 0 | dst_v += dst_stride_v; |
2954 | 0 | } |
2955 | 0 | if (height & 1) { |
2956 | 0 | RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbI601Constants); |
2957 | 0 | RGBToUVMatrixRow(src_rgb24, 0, dst_u, dst_v, width, &kArgbI601Constants); |
2958 | 0 | } |
2959 | 0 | return 0; |
2960 | 0 | } |
2961 | | #undef HAS_RGB24TOYROW |
2962 | | |
2963 | | /// Convert RGB24 to J420. |
2964 | | LIBYUV_API |
2965 | | int RGB24ToJ420(const uint8_t* src_rgb24, |
2966 | | int src_stride_rgb24, |
2967 | | uint8_t* dst_y, |
2968 | | int dst_stride_y, |
2969 | | uint8_t* dst_u, |
2970 | | int dst_stride_u, |
2971 | | uint8_t* dst_v, |
2972 | | int dst_stride_v, |
2973 | | int width, |
2974 | 0 | int height) { |
2975 | 0 | int y; |
2976 | 0 | void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb, |
2977 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
2978 | 0 | const struct ArgbConstants* c) = RGBToUVMatrixRow_C; |
2979 | 0 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
2980 | 0 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
2981 | |
|
2982 | 0 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
2983 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2984 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
2985 | 0 | if (IS_ALIGNED(width, 16)) { |
2986 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
2987 | 0 | } |
2988 | 0 | } |
2989 | 0 | #endif |
2990 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
2991 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2992 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
2993 | 0 | if (IS_ALIGNED(width, 32)) { |
2994 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
2995 | 0 | } |
2996 | 0 | } |
2997 | 0 | #endif |
2998 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
2999 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3000 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
3001 | 0 | if (IS_ALIGNED(width, 64)) { |
3002 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
3003 | 0 | } |
3004 | 0 | } |
3005 | 0 | #endif |
3006 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_SSSE3) |
3007 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3008 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3; |
3009 | 0 | if (IS_ALIGNED(width, 16)) { |
3010 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3; |
3011 | 0 | } |
3012 | 0 | } |
3013 | 0 | #endif |
3014 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX2) |
3015 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3016 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2; |
3017 | 0 | if (IS_ALIGNED(width, 32)) { |
3018 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2; |
3019 | 0 | } |
3020 | 0 | } |
3021 | 0 | #endif |
3022 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX512BW) |
3023 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3024 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX512BW; |
3025 | 0 | if (IS_ALIGNED(width, 64)) { |
3026 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW; |
3027 | 0 | } |
3028 | 0 | } |
3029 | 0 | #endif |
3030 | | #if defined(HAS_RGBTOUVMATRIXROW_NEON) |
3031 | | if (TestCpuFlag(kCpuHasNEON)) { |
3032 | | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON; |
3033 | | if (IS_ALIGNED(width, 16)) { |
3034 | | RGBToUVMatrixRow = RGBToUVMatrixRow_NEON; |
3035 | | } |
3036 | | } |
3037 | | #endif |
3038 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
3039 | | if (TestCpuFlag(kCpuHasNEON)) { |
3040 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
3041 | | if (IS_ALIGNED(width, 16)) { |
3042 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
3043 | | } |
3044 | | } |
3045 | | #endif |
3046 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
3047 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
3048 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
3049 | | if (IS_ALIGNED(width, 16)) { |
3050 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
3051 | | } |
3052 | | } |
3053 | | #endif |
3054 | | #if defined(HAS_RGBTOUVMATRIXROW_RVV) |
3055 | | if (TestCpuFlag(kCpuHasRVV)) { |
3056 | | RGBToUVMatrixRow = RGBToUVMatrixRow_RVV; |
3057 | | } |
3058 | | #endif |
3059 | | #if defined(HAS_RGBTOYMATRIXROW_RVV) |
3060 | | if (TestCpuFlag(kCpuHasRVV)) { |
3061 | | RGBToYMatrixRow = RGBToYMatrixRow_RVV; |
3062 | | } |
3063 | | #endif |
3064 | |
|
3065 | 0 | if (!src_rgb24 || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 || |
3066 | 0 | height == INT_MIN) { |
3067 | 0 | return -1; |
3068 | 0 | } |
3069 | | // Negative height means invert the image. |
3070 | 0 | if (height < 0) { |
3071 | 0 | height = -height; |
3072 | 0 | src_rgb24 = src_rgb24 + (ptrdiff_t)(height - 1) * src_stride_rgb24; |
3073 | 0 | src_stride_rgb24 = -src_stride_rgb24; |
3074 | 0 | } |
3075 | |
|
3076 | 0 | for (y = 0; y < height - 1; y += 2) { |
3077 | 0 | RGBToUVMatrixRow(src_rgb24, src_stride_rgb24, dst_u, dst_v, width, |
3078 | 0 | &kArgbJPEGConstants); |
3079 | 0 | RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbJPEGConstants); |
3080 | 0 | RGBToYMatrixRow(src_rgb24 + src_stride_rgb24, dst_y + dst_stride_y, width, |
3081 | 0 | &kArgbJPEGConstants); |
3082 | 0 | src_rgb24 += src_stride_rgb24 * 2; |
3083 | 0 | dst_y += dst_stride_y * 2; |
3084 | 0 | dst_u += dst_stride_u; |
3085 | 0 | dst_v += dst_stride_v; |
3086 | 0 | } |
3087 | 0 | if (height & 1) { |
3088 | 0 | RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbJPEGConstants); |
3089 | 0 | RGBToUVMatrixRow(src_rgb24, 0, dst_u, dst_v, width, &kArgbJPEGConstants); |
3090 | 0 | } |
3091 | 0 | return 0; |
3092 | 0 | } |
3093 | | |
3094 | | // Enabled if 1 pass is available |
3095 | | #if (defined(HAS_RAWTOYROW_NEON) || defined(HAS_RAWTOYROW_LSX) || \ |
3096 | | defined(HAS_RAWTOYROW_RVV)) |
3097 | | #define HAS_RAWTOYROW |
3098 | | #endif |
3099 | | |
3100 | | // Convert RAW to I420. |
3101 | | LIBYUV_API |
3102 | | int RAWToI420(const uint8_t* src_raw, |
3103 | | int src_stride_raw, |
3104 | | uint8_t* dst_y, |
3105 | | int dst_stride_y, |
3106 | | uint8_t* dst_u, |
3107 | | int dst_stride_u, |
3108 | | uint8_t* dst_v, |
3109 | | int dst_stride_v, |
3110 | | int width, |
3111 | 0 | int height) { |
3112 | 0 | int y; |
3113 | 0 | void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb, |
3114 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
3115 | 0 | const struct ArgbConstants* c) = RGBToUVMatrixRow_C; |
3116 | 0 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
3117 | 0 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
3118 | |
|
3119 | 0 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
3120 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3121 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
3122 | 0 | if (IS_ALIGNED(width, 16)) { |
3123 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
3124 | 0 | } |
3125 | 0 | } |
3126 | 0 | #endif |
3127 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
3128 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3129 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
3130 | 0 | if (IS_ALIGNED(width, 32)) { |
3131 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
3132 | 0 | } |
3133 | 0 | } |
3134 | 0 | #endif |
3135 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
3136 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3137 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
3138 | 0 | if (IS_ALIGNED(width, 64)) { |
3139 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
3140 | 0 | } |
3141 | 0 | } |
3142 | 0 | #endif |
3143 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_SSSE3) |
3144 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3145 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3; |
3146 | 0 | if (IS_ALIGNED(width, 16)) { |
3147 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3; |
3148 | 0 | } |
3149 | 0 | } |
3150 | 0 | #endif |
3151 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX2) |
3152 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3153 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2; |
3154 | 0 | if (IS_ALIGNED(width, 32)) { |
3155 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2; |
3156 | 0 | } |
3157 | 0 | } |
3158 | 0 | #endif |
3159 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX512BW) |
3160 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3161 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX512BW; |
3162 | 0 | if (IS_ALIGNED(width, 64)) { |
3163 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW; |
3164 | 0 | } |
3165 | 0 | } |
3166 | 0 | #endif |
3167 | | #if defined(HAS_RGBTOUVMATRIXROW_NEON) |
3168 | | if (TestCpuFlag(kCpuHasNEON)) { |
3169 | | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON; |
3170 | | if (IS_ALIGNED(width, 16)) { |
3171 | | RGBToUVMatrixRow = RGBToUVMatrixRow_NEON; |
3172 | | } |
3173 | | } |
3174 | | #endif |
3175 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
3176 | | if (TestCpuFlag(kCpuHasNEON)) { |
3177 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
3178 | | if (IS_ALIGNED(width, 16)) { |
3179 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
3180 | | } |
3181 | | } |
3182 | | #endif |
3183 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
3184 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
3185 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
3186 | | if (IS_ALIGNED(width, 16)) { |
3187 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
3188 | | } |
3189 | | } |
3190 | | #endif |
3191 | |
|
3192 | 0 | if (!src_raw || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 || |
3193 | 0 | height == INT_MIN) { |
3194 | 0 | return -1; |
3195 | 0 | } |
3196 | | // Negative height means invert the image. |
3197 | 0 | if (height < 0) { |
3198 | 0 | height = -height; |
3199 | 0 | src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw; |
3200 | 0 | src_stride_raw = -src_stride_raw; |
3201 | 0 | } |
3202 | |
|
3203 | 0 | for (y = 0; y < height - 1; y += 2) { |
3204 | 0 | RGBToUVMatrixRow(src_raw, src_stride_raw, dst_u, dst_v, width, |
3205 | 0 | &kAbgrI601Constants); |
3206 | 0 | RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrI601Constants); |
3207 | 0 | RGBToYMatrixRow(src_raw + src_stride_raw, dst_y + dst_stride_y, width, |
3208 | 0 | &kAbgrI601Constants); |
3209 | 0 | src_raw += src_stride_raw * 2; |
3210 | 0 | dst_y += dst_stride_y * 2; |
3211 | 0 | dst_u += dst_stride_u; |
3212 | 0 | dst_v += dst_stride_v; |
3213 | 0 | } |
3214 | 0 | if (height & 1) { |
3215 | 0 | RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrI601Constants); |
3216 | 0 | RGBToUVMatrixRow(src_raw, 0, dst_u, dst_v, width, &kAbgrI601Constants); |
3217 | 0 | } |
3218 | 0 | return 0; |
3219 | 0 | } |
3220 | | #undef HAS_RAWTOYROW |
3221 | | |
3222 | | // Convert RAW to J420. |
3223 | | LIBYUV_API |
3224 | | int RAWToJ420(const uint8_t* src_raw, |
3225 | | int src_stride_raw, |
3226 | | uint8_t* dst_y, |
3227 | | int dst_stride_y, |
3228 | | uint8_t* dst_u, |
3229 | | int dst_stride_u, |
3230 | | uint8_t* dst_v, |
3231 | | int dst_stride_v, |
3232 | | int width, |
3233 | 96 | int height) { |
3234 | 96 | int y; |
3235 | 96 | void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb, |
3236 | 96 | uint8_t* dst_u, uint8_t* dst_v, int width, |
3237 | 96 | const struct ArgbConstants* c) = RGBToUVMatrixRow_C; |
3238 | 96 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
3239 | 96 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
3240 | | |
3241 | 96 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
3242 | 96 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3243 | 96 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
3244 | 96 | if (IS_ALIGNED(width, 16)) { |
3245 | 9 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
3246 | 9 | } |
3247 | 96 | } |
3248 | 96 | #endif |
3249 | 96 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
3250 | 96 | if (TestCpuFlag(kCpuHasAVX2)) { |
3251 | 96 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
3252 | 96 | if (IS_ALIGNED(width, 32)) { |
3253 | 7 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
3254 | 7 | } |
3255 | 96 | } |
3256 | 96 | #endif |
3257 | 96 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
3258 | 96 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3259 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
3260 | 0 | if (IS_ALIGNED(width, 64)) { |
3261 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
3262 | 0 | } |
3263 | 0 | } |
3264 | 96 | #endif |
3265 | 96 | #if defined(HAS_RGBTOUVMATRIXROW_SSSE3) |
3266 | 96 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3267 | 96 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3; |
3268 | 96 | if (IS_ALIGNED(width, 16)) { |
3269 | 9 | RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3; |
3270 | 9 | } |
3271 | 96 | } |
3272 | 96 | #endif |
3273 | 96 | #if defined(HAS_RGBTOUVMATRIXROW_AVX2) |
3274 | 96 | if (TestCpuFlag(kCpuHasAVX2)) { |
3275 | 96 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2; |
3276 | 96 | if (IS_ALIGNED(width, 32)) { |
3277 | 7 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2; |
3278 | 7 | } |
3279 | 96 | } |
3280 | 96 | #endif |
3281 | 96 | #if defined(HAS_RGBTOUVMATRIXROW_AVX512BW) |
3282 | 96 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3283 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX512BW; |
3284 | 0 | if (IS_ALIGNED(width, 64)) { |
3285 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW; |
3286 | 0 | } |
3287 | 0 | } |
3288 | 96 | #endif |
3289 | | #if defined(HAS_RGBTOUVMATRIXROW_NEON) |
3290 | | if (TestCpuFlag(kCpuHasNEON)) { |
3291 | | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON; |
3292 | | if (IS_ALIGNED(width, 16)) { |
3293 | | RGBToUVMatrixRow = RGBToUVMatrixRow_NEON; |
3294 | | } |
3295 | | } |
3296 | | #endif |
3297 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
3298 | | if (TestCpuFlag(kCpuHasNEON)) { |
3299 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
3300 | | if (IS_ALIGNED(width, 16)) { |
3301 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
3302 | | } |
3303 | | } |
3304 | | #endif |
3305 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
3306 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
3307 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
3308 | | if (IS_ALIGNED(width, 16)) { |
3309 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
3310 | | } |
3311 | | } |
3312 | | #endif |
3313 | | |
3314 | 96 | if (!src_raw || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 || |
3315 | 96 | height == INT_MIN) { |
3316 | 0 | return -1; |
3317 | 0 | } |
3318 | | // Negative height means invert the image. |
3319 | 96 | if (height < 0) { |
3320 | 0 | height = -height; |
3321 | 0 | src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw; |
3322 | 0 | src_stride_raw = -src_stride_raw; |
3323 | 0 | } |
3324 | | |
3325 | 495k | for (y = 0; y < height - 1; y += 2) { |
3326 | 495k | RGBToUVMatrixRow(src_raw, src_stride_raw, dst_u, dst_v, width, |
3327 | 495k | &kAbgrJPEGConstants); |
3328 | 495k | RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrJPEGConstants); |
3329 | 495k | RGBToYMatrixRow(src_raw + src_stride_raw, dst_y + dst_stride_y, width, |
3330 | 495k | &kAbgrJPEGConstants); |
3331 | 495k | src_raw += src_stride_raw * 2; |
3332 | 495k | dst_y += dst_stride_y * 2; |
3333 | 495k | dst_u += dst_stride_u; |
3334 | 495k | dst_v += dst_stride_v; |
3335 | 495k | } |
3336 | 96 | if (height & 1) { |
3337 | 46 | RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrJPEGConstants); |
3338 | 46 | RGBToUVMatrixRow(src_raw, 0, dst_u, dst_v, width, &kAbgrJPEGConstants); |
3339 | 46 | } |
3340 | 96 | return 0; |
3341 | 96 | } |
3342 | | |
3343 | | // Convert RAW to I444 with matrix. |
3344 | | static int RAWToI444Matrix(const uint8_t* src_raw, |
3345 | | int src_stride_raw, |
3346 | | uint8_t* dst_y, |
3347 | | int dst_stride_y, |
3348 | | uint8_t* dst_u, |
3349 | | int dst_stride_u, |
3350 | | uint8_t* dst_v, |
3351 | | int dst_stride_v, |
3352 | | const struct ArgbConstants* argbconstants, |
3353 | | int width, |
3354 | 87 | int height) { |
3355 | 87 | int y; |
3356 | 87 | void (*RGBToUV444MatrixRow)(const uint8_t* src_rgb, uint8_t* dst_u, |
3357 | 87 | uint8_t* dst_v, int width, |
3358 | 87 | const struct ArgbConstants* c) = |
3359 | 87 | RGBToUV444MatrixRow_C; |
3360 | 87 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
3361 | 87 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
3362 | | |
3363 | 87 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
3364 | 87 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3365 | 87 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
3366 | 87 | if (IS_ALIGNED(width, 16)) { |
3367 | 3 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
3368 | 3 | } |
3369 | 87 | } |
3370 | 87 | #endif |
3371 | 87 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
3372 | 87 | if (TestCpuFlag(kCpuHasAVX2)) { |
3373 | 87 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
3374 | 87 | if (IS_ALIGNED(width, 32)) { |
3375 | 2 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
3376 | 2 | } |
3377 | 87 | } |
3378 | 87 | #endif |
3379 | 87 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
3380 | 87 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3381 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
3382 | 0 | if (IS_ALIGNED(width, 64)) { |
3383 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
3384 | 0 | } |
3385 | 0 | } |
3386 | 87 | #endif |
3387 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
3388 | | if (TestCpuFlag(kCpuHasNEON)) { |
3389 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
3390 | | if (IS_ALIGNED(width, 16)) { |
3391 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
3392 | | } |
3393 | | } |
3394 | | #endif |
3395 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
3396 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
3397 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
3398 | | if (IS_ALIGNED(width, 16)) { |
3399 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
3400 | | } |
3401 | | } |
3402 | | #endif |
3403 | | #if defined(HAS_RGBTOYMATRIXROW_LASX) |
3404 | | if (TestCpuFlag(kCpuHasLASX)) { |
3405 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX; |
3406 | | if (IS_ALIGNED(width, 32)) { |
3407 | | RGBToYMatrixRow = RGBToYMatrixRow_LASX; |
3408 | | } |
3409 | | } |
3410 | | #endif |
3411 | | #if defined(HAS_RGBTOYMATRIXROW_RVV) |
3412 | | if (TestCpuFlag(kCpuHasRVV)) { |
3413 | | RGBToYMatrixRow = RGBToYMatrixRow_RVV; |
3414 | | } |
3415 | | #endif |
3416 | | |
3417 | 87 | #if defined(HAS_RGBTOUV444MATRIXROW_SSSE3) |
3418 | 87 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3419 | 87 | RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_SSSE3; |
3420 | 87 | if (IS_ALIGNED(width, 16)) { |
3421 | 3 | RGBToUV444MatrixRow = RGBToUV444MatrixRow_SSSE3; |
3422 | 3 | } |
3423 | 87 | } |
3424 | 87 | #endif |
3425 | 87 | #if defined(HAS_RGBTOUV444MATRIXROW_AVX2) |
3426 | 87 | if (TestCpuFlag(kCpuHasAVX2)) { |
3427 | 87 | RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_AVX2; |
3428 | 87 | if (IS_ALIGNED(width, 32)) { |
3429 | 2 | RGBToUV444MatrixRow = RGBToUV444MatrixRow_AVX2; |
3430 | 2 | } |
3431 | 87 | } |
3432 | 87 | #endif |
3433 | 87 | #if defined(HAS_RGBTOUV444MATRIXROW_AVX512BW) |
3434 | 87 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3435 | 0 | RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_AVX512BW; |
3436 | 0 | if (IS_ALIGNED(width, 64)) { |
3437 | 0 | RGBToUV444MatrixRow = RGBToUV444MatrixRow_AVX512BW; |
3438 | 0 | } |
3439 | 0 | } |
3440 | 87 | #endif |
3441 | | #if defined(HAS_RGBTOUV444MATRIXROW_NEON) |
3442 | | if (TestCpuFlag(kCpuHasNEON)) { |
3443 | | RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_NEON; |
3444 | | if (IS_ALIGNED(width, 8)) { |
3445 | | RGBToUV444MatrixRow = RGBToUV444MatrixRow_NEON; |
3446 | | } |
3447 | | } |
3448 | | #endif |
3449 | | #if defined(HAS_RGBTOUV444MATRIXROW_RVV) |
3450 | | if (TestCpuFlag(kCpuHasRVV)) { |
3451 | | RGBToUV444MatrixRow = RGBToUV444MatrixRow_RVV; |
3452 | | } |
3453 | | #endif |
3454 | | |
3455 | 87 | if (!src_raw || !dst_y || !dst_u || !dst_v || !argbconstants || width <= 0 || |
3456 | 87 | height == 0 || height == INT_MIN) { |
3457 | 0 | return -1; |
3458 | 0 | } |
3459 | 87 | if (height < 0) { |
3460 | 0 | height = -height; |
3461 | 0 | src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw; |
3462 | 0 | src_stride_raw = -src_stride_raw; |
3463 | 0 | } |
3464 | | |
3465 | 981k | for (y = 0; y < height; ++y) { |
3466 | 981k | RGBToYMatrixRow(src_raw, dst_y, width, argbconstants); |
3467 | 981k | RGBToUV444MatrixRow(src_raw, dst_u, dst_v, width, argbconstants); |
3468 | 981k | src_raw += src_stride_raw; |
3469 | 981k | dst_y += dst_stride_y; |
3470 | 981k | dst_u += dst_stride_u; |
3471 | 981k | dst_v += dst_stride_v; |
3472 | 981k | } |
3473 | 87 | return 0; |
3474 | 87 | } |
3475 | | |
3476 | | // Convert RAW to I444. |
3477 | | LIBYUV_API |
3478 | | int RAWToI444(const uint8_t* src_raw, |
3479 | | int src_stride_raw, |
3480 | | uint8_t* dst_y, |
3481 | | int dst_stride_y, |
3482 | | uint8_t* dst_u, |
3483 | | int dst_stride_u, |
3484 | | uint8_t* dst_v, |
3485 | | int dst_stride_v, |
3486 | | int width, |
3487 | 0 | int height) { |
3488 | 0 | return RAWToI444Matrix(src_raw, src_stride_raw, dst_y, dst_stride_y, dst_u, |
3489 | 0 | dst_stride_u, dst_v, dst_stride_v, |
3490 | 0 | &kAbgrI601Constants, width, height); |
3491 | 0 | } |
3492 | | |
3493 | | // Convert RAW to J444. |
3494 | | LIBYUV_API |
3495 | | int RAWToJ444(const uint8_t* src_raw, |
3496 | | int src_stride_raw, |
3497 | | uint8_t* dst_y, |
3498 | | int dst_stride_y, |
3499 | | uint8_t* dst_u, |
3500 | | int dst_stride_u, |
3501 | | uint8_t* dst_v, |
3502 | | int dst_stride_v, |
3503 | | int width, |
3504 | 87 | int height) { |
3505 | 87 | return RAWToI444Matrix(src_raw, src_stride_raw, dst_y, dst_stride_y, dst_u, |
3506 | 87 | dst_stride_u, dst_v, dst_stride_v, |
3507 | 87 | &kAbgrJPEGConstants, width, height); |
3508 | 87 | } |
3509 | | |
3510 | | // Convert RGB565 to I420. |
3511 | | LIBYUV_API |
3512 | | int RGB565ToI420(const uint8_t* src_rgb565, |
3513 | | int src_stride_rgb565, |
3514 | | uint8_t* dst_y, |
3515 | | int dst_stride_y, |
3516 | | uint8_t* dst_u, |
3517 | | int dst_stride_u, |
3518 | | uint8_t* dst_v, |
3519 | | int dst_stride_v, |
3520 | | int width, |
3521 | 0 | int height) { |
3522 | 0 | int y; |
3523 | 0 | void (*RGB565ToUVMatrixRow)(const uint8_t* src_rgb565, int src_stride_rgb565, |
3524 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
3525 | 0 | const struct ArgbConstants* c) = |
3526 | 0 | RGB565ToUVMatrixRow_C; |
3527 | 0 | void (*RGB565ToYMatrixRow)(const uint8_t* src_rgb565, uint8_t* dst_y, |
3528 | 0 | int width, const struct ArgbConstants* c) = |
3529 | 0 | RGB565ToYMatrixRow_C; |
3530 | |
|
3531 | 0 | #if defined(HAS_RGB565TOYMATRIXROW_AVX2) |
3532 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3533 | 0 | RGB565ToYMatrixRow = RGB565ToYMatrixRow_Any_AVX2; |
3534 | 0 | if (IS_ALIGNED(width, 32)) { |
3535 | 0 | RGB565ToYMatrixRow = RGB565ToYMatrixRow_AVX2; |
3536 | 0 | } |
3537 | 0 | } |
3538 | 0 | #endif |
3539 | 0 | #if defined(HAS_RGB565TOUVMATRIXROW_AVX2) |
3540 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3541 | 0 | RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_Any_AVX2; |
3542 | 0 | if (IS_ALIGNED(width, 16)) { |
3543 | 0 | RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_AVX2; |
3544 | 0 | } |
3545 | 0 | } |
3546 | 0 | #endif |
3547 | | #if defined(HAS_RGB565TOUVMATRIXROW_NEON) |
3548 | | if (TestCpuFlag(kCpuHasNEON)) { |
3549 | | RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_Any_NEON; |
3550 | | if (IS_ALIGNED(width, 16)) { |
3551 | | RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_NEON; |
3552 | | } |
3553 | | } |
3554 | | #endif |
3555 | | #if defined(HAS_RGB565TOYMATRIXROW_NEON) |
3556 | | if (TestCpuFlag(kCpuHasNEON)) { |
3557 | | RGB565ToYMatrixRow = RGB565ToYMatrixRow_Any_NEON; |
3558 | | if (IS_ALIGNED(width, 16)) { |
3559 | | RGB565ToYMatrixRow = RGB565ToYMatrixRow_NEON; |
3560 | | } |
3561 | | } |
3562 | | #endif |
3563 | |
|
3564 | 0 | if (!src_rgb565 || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 || |
3565 | 0 | height == INT_MIN) { |
3566 | 0 | return -1; |
3567 | 0 | } |
3568 | | // Negative height means invert the image. |
3569 | 0 | if (height < 0) { |
3570 | 0 | height = -height; |
3571 | 0 | src_rgb565 = src_rgb565 + (ptrdiff_t)(height - 1) * src_stride_rgb565; |
3572 | 0 | src_stride_rgb565 = -src_stride_rgb565; |
3573 | 0 | } |
3574 | |
|
3575 | 0 | for (y = 0; y < height - 1; y += 2) { |
3576 | 0 | RGB565ToUVMatrixRow(src_rgb565, src_stride_rgb565, dst_u, dst_v, width, |
3577 | 0 | &kArgbI601Constants); |
3578 | 0 | RGB565ToYMatrixRow(src_rgb565, dst_y, width, &kArgbI601Constants); |
3579 | 0 | RGB565ToYMatrixRow(src_rgb565 + src_stride_rgb565, dst_y + dst_stride_y, |
3580 | 0 | width, &kArgbI601Constants); |
3581 | 0 | src_rgb565 += src_stride_rgb565 * 2; |
3582 | 0 | dst_y += dst_stride_y * 2; |
3583 | 0 | dst_u += dst_stride_u; |
3584 | 0 | dst_v += dst_stride_v; |
3585 | 0 | } |
3586 | 0 | if (height & 1) { |
3587 | 0 | RGB565ToYMatrixRow(src_rgb565, dst_y, width, &kArgbI601Constants); |
3588 | 0 | RGB565ToUVMatrixRow(src_rgb565, 0, dst_u, dst_v, width, |
3589 | 0 | &kArgbI601Constants); |
3590 | 0 | } |
3591 | 0 | return 0; |
3592 | 0 | } |
3593 | | // Convert ARGB1555 to I420. |
3594 | | LIBYUV_API |
3595 | | int ARGB1555ToI420(const uint8_t* src_argb1555, |
3596 | | int src_stride_argb1555, |
3597 | | uint8_t* dst_y, |
3598 | | int dst_stride_y, |
3599 | | uint8_t* dst_u, |
3600 | | int dst_stride_u, |
3601 | | uint8_t* dst_v, |
3602 | | int dst_stride_v, |
3603 | | int width, |
3604 | 0 | int height) { |
3605 | 0 | int y; |
3606 | 0 | void (*ARGB1555ToUVMatrixRow)( |
3607 | 0 | const uint8_t* src_argb1555, int src_stride_argb1555, uint8_t* dst_u, |
3608 | 0 | uint8_t* dst_v, int width, const struct ArgbConstants* c) = |
3609 | 0 | ARGB1555ToUVMatrixRow_C; |
3610 | 0 | void (*ARGB1555ToYMatrixRow)(const uint8_t* src_argb1555, uint8_t* dst_y, |
3611 | 0 | int width, const struct ArgbConstants* c) = |
3612 | 0 | ARGB1555ToYMatrixRow_C; |
3613 | |
|
3614 | 0 | #if defined(HAS_ARGB1555TOYMATRIXROW_AVX2) |
3615 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3616 | 0 | ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_Any_AVX2; |
3617 | 0 | if (IS_ALIGNED(width, 32)) { |
3618 | 0 | ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_AVX2; |
3619 | 0 | } |
3620 | 0 | } |
3621 | 0 | #endif |
3622 | 0 | #if defined(HAS_ARGB1555TOUVMATRIXROW_AVX2) |
3623 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3624 | 0 | ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_Any_AVX2; |
3625 | 0 | if (IS_ALIGNED(width, 16)) { |
3626 | 0 | ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_AVX2; |
3627 | 0 | } |
3628 | 0 | } |
3629 | 0 | #endif |
3630 | | #if defined(HAS_ARGB1555TOUVMATRIXROW_NEON) |
3631 | | if (TestCpuFlag(kCpuHasNEON)) { |
3632 | | ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_Any_NEON; |
3633 | | if (IS_ALIGNED(width, 16)) { |
3634 | | ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_NEON; |
3635 | | } |
3636 | | } |
3637 | | #endif |
3638 | | #if defined(HAS_ARGB1555TOYMATRIXROW_NEON) |
3639 | | if (TestCpuFlag(kCpuHasNEON)) { |
3640 | | ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_Any_NEON; |
3641 | | if (IS_ALIGNED(width, 16)) { |
3642 | | ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_NEON; |
3643 | | } |
3644 | | } |
3645 | | #endif |
3646 | |
|
3647 | 0 | if (!src_argb1555 || !dst_y || !dst_u || !dst_v || width <= 0 || |
3648 | 0 | height == 0 || height == INT_MIN) { |
3649 | 0 | return -1; |
3650 | 0 | } |
3651 | | // Negative height means invert the image. |
3652 | 0 | if (height < 0) { |
3653 | 0 | height = -height; |
3654 | 0 | src_argb1555 = src_argb1555 + (ptrdiff_t)(height - 1) * src_stride_argb1555; |
3655 | 0 | src_stride_argb1555 = -src_stride_argb1555; |
3656 | 0 | } |
3657 | |
|
3658 | 0 | for (y = 0; y < height - 1; y += 2) { |
3659 | 0 | ARGB1555ToUVMatrixRow(src_argb1555, src_stride_argb1555, dst_u, dst_v, |
3660 | 0 | width, &kArgbI601Constants); |
3661 | 0 | ARGB1555ToYMatrixRow(src_argb1555, dst_y, width, &kArgbI601Constants); |
3662 | 0 | ARGB1555ToYMatrixRow(src_argb1555 + src_stride_argb1555, |
3663 | 0 | dst_y + dst_stride_y, width, &kArgbI601Constants); |
3664 | 0 | src_argb1555 += src_stride_argb1555 * 2; |
3665 | 0 | dst_y += dst_stride_y * 2; |
3666 | 0 | dst_u += dst_stride_u; |
3667 | 0 | dst_v += dst_stride_v; |
3668 | 0 | } |
3669 | 0 | if (height & 1) { |
3670 | 0 | ARGB1555ToYMatrixRow(src_argb1555, dst_y, width, &kArgbI601Constants); |
3671 | 0 | ARGB1555ToUVMatrixRow(src_argb1555, 0, dst_u, dst_v, width, |
3672 | 0 | &kArgbI601Constants); |
3673 | 0 | } |
3674 | 0 | return 0; |
3675 | 0 | } |
3676 | | // Convert ARGB4444 to I420. |
3677 | | LIBYUV_API |
3678 | | int ARGB4444ToI420(const uint8_t* src_argb4444, |
3679 | | int src_stride_argb4444, |
3680 | | uint8_t* dst_y, |
3681 | | int dst_stride_y, |
3682 | | uint8_t* dst_u, |
3683 | | int dst_stride_u, |
3684 | | uint8_t* dst_v, |
3685 | | int dst_stride_v, |
3686 | | int width, |
3687 | 0 | int height) { |
3688 | 0 | int y; |
3689 | 0 | void (*ARGB4444ToUVMatrixRow)( |
3690 | 0 | const uint8_t* src_argb4444, int src_stride_argb4444, uint8_t* dst_u, |
3691 | 0 | uint8_t* dst_v, int width, const struct ArgbConstants* c) = |
3692 | 0 | ARGB4444ToUVMatrixRow_C; |
3693 | 0 | void (*ARGB4444ToYMatrixRow)(const uint8_t* src_argb4444, uint8_t* dst_y, |
3694 | 0 | int width, const struct ArgbConstants* c) = |
3695 | 0 | ARGB4444ToYMatrixRow_C; |
3696 | |
|
3697 | 0 | #if defined(HAS_ARGB4444TOYMATRIXROW_AVX2) |
3698 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3699 | 0 | ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_Any_AVX2; |
3700 | 0 | if (IS_ALIGNED(width, 32)) { |
3701 | 0 | ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_AVX2; |
3702 | 0 | } |
3703 | 0 | } |
3704 | 0 | #endif |
3705 | 0 | #if defined(HAS_ARGB4444TOUVMATRIXROW_AVX2) |
3706 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3707 | 0 | ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_Any_AVX2; |
3708 | 0 | if (IS_ALIGNED(width, 16)) { |
3709 | 0 | ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_AVX2; |
3710 | 0 | } |
3711 | 0 | } |
3712 | 0 | #endif |
3713 | | #if defined(HAS_ARGB4444TOUVMATRIXROW_NEON) |
3714 | | if (TestCpuFlag(kCpuHasNEON)) { |
3715 | | ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_Any_NEON; |
3716 | | if (IS_ALIGNED(width, 16)) { |
3717 | | ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_NEON; |
3718 | | } |
3719 | | } |
3720 | | #endif |
3721 | | #if defined(HAS_ARGB4444TOYMATRIXROW_NEON) |
3722 | | if (TestCpuFlag(kCpuHasNEON)) { |
3723 | | ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_Any_NEON; |
3724 | | if (IS_ALIGNED(width, 16)) { |
3725 | | ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_NEON; |
3726 | | } |
3727 | | } |
3728 | | #endif |
3729 | |
|
3730 | 0 | if (!src_argb4444 || !dst_y || !dst_u || !dst_v || width <= 0 || |
3731 | 0 | height == 0 || height == INT_MIN) { |
3732 | 0 | return -1; |
3733 | 0 | } |
3734 | | // Negative height means invert the image. |
3735 | 0 | if (height < 0) { |
3736 | 0 | height = -height; |
3737 | 0 | src_argb4444 = src_argb4444 + (ptrdiff_t)(height - 1) * src_stride_argb4444; |
3738 | 0 | src_stride_argb4444 = -src_stride_argb4444; |
3739 | 0 | } |
3740 | |
|
3741 | 0 | for (y = 0; y < height - 1; y += 2) { |
3742 | 0 | ARGB4444ToUVMatrixRow(src_argb4444, src_stride_argb4444, dst_u, dst_v, |
3743 | 0 | width, &kArgbI601Constants); |
3744 | 0 | ARGB4444ToYMatrixRow(src_argb4444, dst_y, width, &kArgbI601Constants); |
3745 | 0 | ARGB4444ToYMatrixRow(src_argb4444 + src_stride_argb4444, |
3746 | 0 | dst_y + dst_stride_y, width, &kArgbI601Constants); |
3747 | 0 | src_argb4444 += src_stride_argb4444 * 2; |
3748 | 0 | dst_y += dst_stride_y * 2; |
3749 | 0 | dst_u += dst_stride_u; |
3750 | 0 | dst_v += dst_stride_v; |
3751 | 0 | } |
3752 | 0 | if (height & 1) { |
3753 | 0 | ARGB4444ToYMatrixRow(src_argb4444, dst_y, width, &kArgbI601Constants); |
3754 | 0 | ARGB4444ToUVMatrixRow(src_argb4444, 0, dst_u, dst_v, width, |
3755 | 0 | &kArgbI601Constants); |
3756 | 0 | } |
3757 | 0 | return 0; |
3758 | 0 | } |
3759 | | // Convert RGB24 to J400. |
3760 | | LIBYUV_API |
3761 | | int RGB24ToJ400(const uint8_t* src_rgb24, |
3762 | | int src_stride_rgb24, |
3763 | | uint8_t* dst_yj, |
3764 | | int dst_stride_yj, |
3765 | | int width, |
3766 | 0 | int height) { |
3767 | 0 | int y; |
3768 | 0 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
3769 | 0 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
3770 | 0 | if (!src_rgb24 || !dst_yj || width <= 0 || height == 0 || height == INT_MIN) { |
3771 | 0 | return -1; |
3772 | 0 | } |
3773 | 0 | if (height < 0) { |
3774 | 0 | height = -height; |
3775 | 0 | src_rgb24 = src_rgb24 + (ptrdiff_t)(height - 1) * src_stride_rgb24; |
3776 | 0 | src_stride_rgb24 = -src_stride_rgb24; |
3777 | 0 | } |
3778 | | // Coalesce rows. |
3779 | 0 | if (src_stride_rgb24 == width * 3 && dst_stride_yj == width && |
3780 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
3781 | 0 | width *= height; |
3782 | 0 | height = 1; |
3783 | 0 | src_stride_rgb24 = dst_stride_yj = 0; |
3784 | 0 | } |
3785 | |
|
3786 | 0 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
3787 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3788 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
3789 | 0 | if (IS_ALIGNED(width, 16)) { |
3790 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
3791 | 0 | } |
3792 | 0 | } |
3793 | 0 | #endif |
3794 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
3795 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
3796 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
3797 | 0 | if (IS_ALIGNED(width, 32)) { |
3798 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
3799 | 0 | } |
3800 | 0 | } |
3801 | 0 | #endif |
3802 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
3803 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3804 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
3805 | 0 | if (IS_ALIGNED(width, 64)) { |
3806 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
3807 | 0 | } |
3808 | 0 | } |
3809 | 0 | #endif |
3810 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
3811 | | if (TestCpuFlag(kCpuHasNEON)) { |
3812 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
3813 | | if (IS_ALIGNED(width, 16)) { |
3814 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
3815 | | } |
3816 | | } |
3817 | | #endif |
3818 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
3819 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
3820 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
3821 | | if (IS_ALIGNED(width, 16)) { |
3822 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
3823 | | } |
3824 | | } |
3825 | | #endif |
3826 | | #if defined(HAS_RGBTOYMATRIXROW_LASX) |
3827 | | if (TestCpuFlag(kCpuHasLASX)) { |
3828 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX; |
3829 | | if (IS_ALIGNED(width, 32)) { |
3830 | | RGBToYMatrixRow = RGBToYMatrixRow_LASX; |
3831 | | } |
3832 | | } |
3833 | | #endif |
3834 | | #if defined(HAS_RGBTOYMATRIXROW_RVV) |
3835 | | if (TestCpuFlag(kCpuHasRVV)) { |
3836 | | RGBToYMatrixRow = RGBToYMatrixRow_RVV; |
3837 | | } |
3838 | | #endif |
3839 | |
|
3840 | 0 | for (y = 0; y < height; ++y) { |
3841 | 0 | RGBToYMatrixRow(src_rgb24, dst_yj, width, &kArgbJPEGConstants); |
3842 | 0 | src_rgb24 += src_stride_rgb24; |
3843 | 0 | dst_yj += dst_stride_yj; |
3844 | 0 | } |
3845 | 0 | return 0; |
3846 | 0 | } |
3847 | | |
3848 | | // Convert RAW to J400. |
3849 | | LIBYUV_API |
3850 | | int RAWToJ400(const uint8_t* src_raw, |
3851 | | int src_stride_raw, |
3852 | | uint8_t* dst_yj, |
3853 | | int dst_stride_yj, |
3854 | | int width, |
3855 | 44 | int height) { |
3856 | 44 | int y; |
3857 | 44 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
3858 | 44 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
3859 | 44 | if (!src_raw || !dst_yj || width <= 0 || height == 0 || height == INT_MIN) { |
3860 | 0 | return -1; |
3861 | 0 | } |
3862 | 44 | if (height < 0) { |
3863 | 0 | height = -height; |
3864 | 0 | src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw; |
3865 | 0 | src_stride_raw = -src_stride_raw; |
3866 | 0 | } |
3867 | | // Coalesce rows. |
3868 | 44 | if (src_stride_raw == width * 3 && dst_stride_yj == width && |
3869 | 44 | (ptrdiff_t)width * height <= INT_MAX) { |
3870 | 44 | width *= height; |
3871 | 44 | height = 1; |
3872 | 44 | src_stride_raw = dst_stride_yj = 0; |
3873 | 44 | } |
3874 | | |
3875 | 44 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
3876 | 44 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3877 | 44 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
3878 | 44 | if (IS_ALIGNED(width, 16)) { |
3879 | 10 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
3880 | 10 | } |
3881 | 44 | } |
3882 | 44 | #endif |
3883 | 44 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
3884 | 44 | if (TestCpuFlag(kCpuHasAVX2)) { |
3885 | 44 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
3886 | 44 | if (IS_ALIGNED(width, 32)) { |
3887 | 5 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
3888 | 5 | } |
3889 | 44 | } |
3890 | 44 | #endif |
3891 | 44 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
3892 | 44 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
3893 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
3894 | 0 | if (IS_ALIGNED(width, 64)) { |
3895 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
3896 | 0 | } |
3897 | 0 | } |
3898 | 44 | #endif |
3899 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
3900 | | if (TestCpuFlag(kCpuHasNEON)) { |
3901 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
3902 | | if (IS_ALIGNED(width, 16)) { |
3903 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
3904 | | } |
3905 | | } |
3906 | | #endif |
3907 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
3908 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
3909 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
3910 | | if (IS_ALIGNED(width, 16)) { |
3911 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
3912 | | } |
3913 | | } |
3914 | | #endif |
3915 | | #if defined(HAS_RGBTOYMATRIXROW_LASX) |
3916 | | if (TestCpuFlag(kCpuHasLASX)) { |
3917 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX; |
3918 | | if (IS_ALIGNED(width, 32)) { |
3919 | | RGBToYMatrixRow = RGBToYMatrixRow_LASX; |
3920 | | } |
3921 | | } |
3922 | | #endif |
3923 | | #if defined(HAS_RGBTOYMATRIXROW_RVV) |
3924 | | if (TestCpuFlag(kCpuHasRVV)) { |
3925 | | RGBToYMatrixRow = RGBToYMatrixRow_RVV; |
3926 | | } |
3927 | | #endif |
3928 | | |
3929 | 88 | for (y = 0; y < height; ++y) { |
3930 | 44 | RGBToYMatrixRow(src_raw, dst_yj, width, &kAbgrJPEGConstants); |
3931 | 44 | src_raw += src_stride_raw; |
3932 | 44 | dst_yj += dst_stride_yj; |
3933 | 44 | } |
3934 | 44 | return 0; |
3935 | 44 | } |
3936 | | |
3937 | | // Convert Android420 to I420. |
3938 | | LIBYUV_API |
3939 | | int Android420ToI420(const uint8_t* src_y, |
3940 | | int src_stride_y, |
3941 | | const uint8_t* src_u, |
3942 | | int src_stride_u, |
3943 | | const uint8_t* src_v, |
3944 | | int src_stride_v, |
3945 | | int src_pixel_stride_uv, |
3946 | | uint8_t* dst_y, |
3947 | | int dst_stride_y, |
3948 | | uint8_t* dst_u, |
3949 | | int dst_stride_u, |
3950 | | uint8_t* dst_v, |
3951 | | int dst_stride_v, |
3952 | | int width, |
3953 | 0 | int height) { |
3954 | 0 | return Android420ToI420Rotate(src_y, src_stride_y, src_u, src_stride_u, src_v, |
3955 | 0 | src_stride_v, src_pixel_stride_uv, dst_y, |
3956 | 0 | dst_stride_y, dst_u, dst_stride_u, dst_v, |
3957 | 0 | dst_stride_v, width, height, kRotate0); |
3958 | 0 | } |
3959 | | |
3960 | | // depth is source bits measured from lsb; For msb use 16 |
3961 | | static int Biplanar16bitTo8bit(const uint16_t* src_y, |
3962 | | int src_stride_y, |
3963 | | const uint16_t* src_uv, |
3964 | | int src_stride_uv, |
3965 | | uint8_t* dst_y, |
3966 | | int dst_stride_y, |
3967 | | uint8_t* dst_uv, |
3968 | | int dst_stride_uv, |
3969 | | int width, |
3970 | | int height, |
3971 | | int subsample_x, |
3972 | | int subsample_y, |
3973 | 0 | int depth) { |
3974 | 0 | if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 || |
3975 | 0 | height == INT_MIN) { |
3976 | 0 | return -1; |
3977 | 0 | } |
3978 | 0 | int uv_width = SUBSAMPLE(width, subsample_x, subsample_x); |
3979 | 0 | int uv_height = SUBSAMPLE(height, subsample_y, subsample_y); |
3980 | 0 | int scale = 1 << (24 - depth); |
3981 | | // Negative height means invert the image. |
3982 | 0 | if (height < 0) { |
3983 | 0 | height = -height; |
3984 | 0 | uv_height = -uv_height; |
3985 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
3986 | 0 | src_uv = src_uv + (ptrdiff_t)(uv_height - 1) * src_stride_uv; |
3987 | 0 | src_stride_y = -src_stride_y; |
3988 | 0 | src_stride_uv = -src_stride_uv; |
3989 | 0 | } |
3990 | | |
3991 | | // Convert Y plane. |
3992 | 0 | if (dst_y) { |
3993 | 0 | Convert16To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale, width, |
3994 | 0 | height); |
3995 | 0 | } |
3996 | | // Convert UV planes. |
3997 | 0 | Convert16To8Plane(src_uv, src_stride_uv, dst_uv, dst_stride_uv, scale, |
3998 | 0 | uv_width * 2, uv_height); |
3999 | 0 | return 0; |
4000 | 0 | } |
4001 | | |
4002 | | // Convert 10 bit P010 to 8 bit NV12. |
4003 | | // Depth set to 16 because P010 uses 10 msb and this function keeps the upper 8 |
4004 | | // bits of the specified number of bits. |
4005 | | LIBYUV_API |
4006 | | int P010ToNV12(const uint16_t* src_y, |
4007 | | int src_stride_y, |
4008 | | const uint16_t* src_uv, |
4009 | | int src_stride_uv, |
4010 | | uint8_t* dst_y, |
4011 | | int dst_stride_y, |
4012 | | uint8_t* dst_uv, |
4013 | | int dst_stride_uv, |
4014 | | int width, |
4015 | 0 | int height) { |
4016 | 0 | return Biplanar16bitTo8bit(src_y, src_stride_y, src_uv, src_stride_uv, dst_y, |
4017 | 0 | dst_stride_y, dst_uv, dst_stride_uv, width, height, |
4018 | 0 | 1, 1, 16); |
4019 | 0 | } |
4020 | | |
4021 | | static int Planar8bitTo8bit(const uint8_t* src_y, |
4022 | | int src_stride_y, |
4023 | | const uint8_t* src_u, |
4024 | | int src_stride_u, |
4025 | | const uint8_t* src_v, |
4026 | | int src_stride_v, |
4027 | | uint8_t* dst_y, |
4028 | | int dst_stride_y, |
4029 | | uint8_t* dst_u, |
4030 | | int dst_stride_u, |
4031 | | uint8_t* dst_v, |
4032 | | int dst_stride_v, |
4033 | | int width, |
4034 | | int height, |
4035 | | int subsample_x, |
4036 | | int subsample_y, |
4037 | | int scale_y, |
4038 | | int bias_y, |
4039 | | int scale_uv, |
4040 | 0 | int bias_uv) { |
4041 | 0 | if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 || |
4042 | 0 | height == 0 || height == INT_MIN) { |
4043 | 0 | return -1; |
4044 | 0 | } |
4045 | 0 | int uv_width = SUBSAMPLE(width, subsample_x, subsample_x); |
4046 | 0 | int uv_height = SUBSAMPLE(height, subsample_y, subsample_y); |
4047 | | // Negative height means invert the image. |
4048 | 0 | if (height < 0) { |
4049 | 0 | height = -height; |
4050 | 0 | uv_height = -uv_height; |
4051 | 0 | src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y; |
4052 | 0 | src_u = src_u + (ptrdiff_t)(uv_height - 1) * src_stride_u; |
4053 | 0 | src_v = src_v + (ptrdiff_t)(uv_height - 1) * src_stride_v; |
4054 | 0 | src_stride_y = -src_stride_y; |
4055 | 0 | src_stride_u = -src_stride_u; |
4056 | 0 | src_stride_v = -src_stride_v; |
4057 | 0 | } |
4058 | | |
4059 | | // Convert Y plane. |
4060 | 0 | if (dst_y) { |
4061 | 0 | Convert8To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale_y, bias_y, |
4062 | 0 | width, height); |
4063 | 0 | } |
4064 | | // Convert UV planes. |
4065 | 0 | Convert8To8Plane(src_u, src_stride_u, dst_u, dst_stride_u, scale_uv, bias_uv, |
4066 | 0 | uv_width, uv_height); |
4067 | 0 | Convert8To8Plane(src_v, src_stride_v, dst_v, dst_stride_v, scale_uv, bias_uv, |
4068 | 0 | uv_width, uv_height); |
4069 | 0 | return 0; |
4070 | 0 | } |
4071 | | |
4072 | | LIBYUV_API |
4073 | | int J420ToI420(const uint8_t* src_y, |
4074 | | int src_stride_y, |
4075 | | const uint8_t* src_u, |
4076 | | int src_stride_u, |
4077 | | const uint8_t* src_v, |
4078 | | int src_stride_v, |
4079 | | uint8_t* dst_y, |
4080 | | int dst_stride_y, |
4081 | | uint8_t* dst_u, |
4082 | | int dst_stride_u, |
4083 | | uint8_t* dst_v, |
4084 | | int dst_stride_v, |
4085 | | int width, |
4086 | 0 | int height) { |
4087 | 0 | return Planar8bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v, |
4088 | 0 | src_stride_v, dst_y, dst_stride_y, dst_u, |
4089 | 0 | dst_stride_u, dst_v, dst_stride_v, width, height, 1, |
4090 | 0 | 1, 220, 16, 225, 16); |
4091 | 0 | } |
4092 | | |
4093 | | #ifdef __cplusplus |
4094 | | } // extern "C" |
4095 | | } // namespace libyuv |
4096 | | #endif |