2 * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 #ifndef SWSCALE_SWSCALE_INTERNAL_H
22 #define SWSCALE_SWSCALE_INTERNAL_H
32 #include "libavutil/avassert.h"
33 #include "libavutil/avutil.h"
34 #include "libavutil/common.h"
35 #include "libavutil/intreadwrite.h"
36 #include "libavutil/log.h"
37 #include "libavutil/pixfmt.h"
38 #include "libavutil/pixdesc.h"
40 #define STR(s) AV_TOSTRING(s) // AV_STRINGIFY is too long
42 #define YUVRGB_TABLE_HEADROOM 256
44 #define MAX_FILTER_SIZE SWS_MAX_FILTER_SIZE
49 #define ALT32_CORR (-1)
64 #define RETCODE_USE_CASCADE -12345
68 typedef enum SwsDither {
78 typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
79 int srcStride[], int srcSliceY, int srcSliceH,
80 uint8_t *dst[], int dstStride[]);
83 * Write one line of horizontally scaled data to planar output
84 * without any additional vertical scaling (or point-scaling).
86 * @param src scaled source data, 15bit for 8-10bit output,
87 * 19-bit for 16bit output (in int32_t)
88 * @param dest pointer to the output plane. For >8bit
89 * output, this is in uint16_t
90 * @param dstW width of destination in pixels
91 * @param dither ordered dither array of type int16_t and size 8
92 * @param offset Dither offset
94 typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
95 const uint8_t *dither, int offset);
98 * Write one line of horizontally scaled data to planar output
99 * with multi-point vertical scaling between input pixels.
101 * @param filter vertical luma/alpha scaling coefficients, 12bit [0,4096]
102 * @param src scaled luma (Y) or alpha (A) source data, 15bit for 8-10bit output,
103 * 19-bit for 16bit output (in int32_t)
104 * @param filterSize number of vertical input lines to scale
105 * @param dest pointer to output plane. For >8bit
106 * output, this is in uint16_t
107 * @param dstW width of destination pixels
108 * @param offset Dither offset
110 typedef void (*yuv2planarX_fn)(const int16_t *filter, int filterSize,
111 const int16_t **src, uint8_t *dest, int dstW,
112 const uint8_t *dither, int offset);
115 * Write one line of horizontally scaled chroma to interleaved output
116 * with multi-point vertical scaling between input pixels.
118 * @param c SWS scaling context
119 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
120 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
121 * 19-bit for 16bit output (in int32_t)
122 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
123 * 19-bit for 16bit output (in int32_t)
124 * @param chrFilterSize number of vertical chroma input lines to scale
125 * @param dest pointer to the output plane. For >8bit
126 * output, this is in uint16_t
127 * @param dstW width of chroma planes
129 typedef void (*yuv2interleavedX_fn)(struct SwsContext *c,
130 const int16_t *chrFilter,
132 const int16_t **chrUSrc,
133 const int16_t **chrVSrc,
134 uint8_t *dest, int dstW);
137 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
138 * output without any additional vertical scaling (or point-scaling). Note
139 * that this function may do chroma scaling, see the "uvalpha" argument.
141 * @param c SWS scaling context
142 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
143 * 19-bit for 16bit output (in int32_t)
144 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
145 * 19-bit for 16bit output (in int32_t)
146 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
147 * 19-bit for 16bit output (in int32_t)
148 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
149 * 19-bit for 16bit output (in int32_t)
150 * @param dest pointer to the output plane. For 16bit output, this is
152 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
153 * to write into dest[]
154 * @param uvalpha chroma scaling coefficient for the second line of chroma
155 * pixels, either 2048 or 0. If 0, one chroma input is used
156 * for 2 output pixels (or if the SWS_FLAG_FULL_CHR_INT flag
157 * is set, it generates 1 output pixel). If 2048, two chroma
158 * input pixels should be averaged for 2 output pixels (this
159 * only happens if SWS_FLAG_FULL_CHR_INT is not set)
160 * @param y vertical line number for this output. This does not need
161 * to be used to calculate the offset in the destination,
162 * but can be used to generate comfort noise using dithering
163 * for some output formats.
165 typedef void (*yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc,
166 const int16_t *chrUSrc[2],
167 const int16_t *chrVSrc[2],
168 const int16_t *alpSrc, uint8_t *dest,
169 int dstW, int uvalpha, int y);
171 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
172 * output by doing bilinear scaling between two input lines.
174 * @param c SWS scaling context
175 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
176 * 19-bit for 16bit output (in int32_t)
177 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
178 * 19-bit for 16bit output (in int32_t)
179 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
180 * 19-bit for 16bit output (in int32_t)
181 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
182 * 19-bit for 16bit output (in int32_t)
183 * @param dest pointer to the output plane. For 16bit output, this is
185 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
186 * to write into dest[]
187 * @param yalpha luma/alpha scaling coefficients for the second input line.
188 * The first line's coefficients can be calculated by using
190 * @param uvalpha chroma scaling coefficient for the second input line. The
191 * first line's coefficients can be calculated by using
193 * @param y vertical line number for this output. This does not need
194 * to be used to calculate the offset in the destination,
195 * but can be used to generate comfort noise using dithering
196 * for some output formats.
198 typedef void (*yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2],
199 const int16_t *chrUSrc[2],
200 const int16_t *chrVSrc[2],
201 const int16_t *alpSrc[2],
203 int dstW, int yalpha, int uvalpha, int y);
205 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
206 * output by doing multi-point vertical scaling between input pixels.
208 * @param c SWS scaling context
209 * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
210 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
211 * 19-bit for 16bit output (in int32_t)
212 * @param lumFilterSize number of vertical luma/alpha input lines to scale
213 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
214 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
215 * 19-bit for 16bit output (in int32_t)
216 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
217 * 19-bit for 16bit output (in int32_t)
218 * @param chrFilterSize number of vertical chroma input lines to scale
219 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
220 * 19-bit for 16bit output (in int32_t)
221 * @param dest pointer to the output plane. For 16bit output, this is
223 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
224 * to write into dest[]
225 * @param y vertical line number for this output. This does not need
226 * to be used to calculate the offset in the destination,
227 * but can be used to generate comfort noise using dithering
228 * or some output formats.
230 typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter,
231 const int16_t **lumSrc, int lumFilterSize,
232 const int16_t *chrFilter,
233 const int16_t **chrUSrc,
234 const int16_t **chrVSrc, int chrFilterSize,
235 const int16_t **alpSrc, uint8_t *dest,
239 * Write one line of horizontally scaled Y/U/V/A to YUV/RGB
240 * output by doing multi-point vertical scaling between input pixels.
242 * @param c SWS scaling context
243 * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
244 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
245 * 19-bit for 16bit output (in int32_t)
246 * @param lumFilterSize number of vertical luma/alpha input lines to scale
247 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
248 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
249 * 19-bit for 16bit output (in int32_t)
250 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
251 * 19-bit for 16bit output (in int32_t)
252 * @param chrFilterSize number of vertical chroma input lines to scale
253 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
254 * 19-bit for 16bit output (in int32_t)
255 * @param dest pointer to the output planes. For 16bit output, this is
257 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
258 * to write into dest[]
259 * @param y vertical line number for this output. This does not need
260 * to be used to calculate the offset in the destination,
261 * but can be used to generate comfort noise using dithering
262 * or some output formats.
264 typedef void (*yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter,
265 const int16_t **lumSrc, int lumFilterSize,
266 const int16_t *chrFilter,
267 const int16_t **chrUSrc,
268 const int16_t **chrVSrc, int chrFilterSize,
269 const int16_t **alpSrc, uint8_t **dest,
272 /* This struct should be aligned on at least a 32-byte boundary. */
273 typedef struct SwsContext {
275 * info on struct for av_log
277 const AVClass *av_class;
280 * Note that src, dst, srcStride, dstStride will be copied in the
281 * sws_scale() wrapper so they can be freely modified here.
284 int srcW; ///< Width of source luma/alpha planes.
285 int srcH; ///< Height of source luma/alpha planes.
286 int dstH; ///< Height of destination luma/alpha planes.
287 int chrSrcW; ///< Width of source chroma planes.
288 int chrSrcH; ///< Height of source chroma planes.
289 int chrDstW; ///< Width of destination chroma planes.
290 int chrDstH; ///< Height of destination chroma planes.
291 int lumXInc, chrXInc;
292 int lumYInc, chrYInc;
293 enum AVPixelFormat dstFormat; ///< Destination pixel format.
294 enum AVPixelFormat srcFormat; ///< Source pixel format.
295 int dstFormatBpp; ///< Number of bits per pixel of the destination pixel format.
296 int srcFormatBpp; ///< Number of bits per pixel of the source pixel format.
298 int chrSrcHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source image.
299 int chrSrcVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in source image.
300 int chrDstHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination image.
301 int chrDstVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in destination image.
302 int vChrDrop; ///< Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user.
303 int sliceDir; ///< Direction that slices are fed to the scaler (1 = top-to-bottom, -1 = bottom-to-top).
304 double param[2]; ///< Input parameters for scaling algorithms that need them.
306 /* The cascaded_* fields allow spliting a scaler task into multiple
307 * sequential steps, this is for example used to limit the maximum
308 * downscaling factor that needs to be supported in one scaler.
310 struct SwsContext *cascaded_context[3];
311 int cascaded_tmpStride[4];
312 uint8_t *cascaded_tmp[4];
313 int cascaded1_tmpStride[4];
314 uint8_t *cascaded1_tmp[4];
318 int is_internal_gamma;
322 uint32_t pal_yuv[256];
323 uint32_t pal_rgb[256];
326 * @name Scaled horizontal lines ring buffer.
327 * The horizontal scaler keeps just enough scaled lines in a ring buffer
328 * so they may be passed to the vertical scaler. The pointers to the
329 * allocated buffers for each line are duplicated in sequence in the ring
330 * buffer to simplify indexing and avoid wrapping around between lines
331 * inside the vertical scaler code. The wrapping is done before the
332 * vertical scaler is called.
335 int16_t **lumPixBuf; ///< Ring buffer for scaled horizontal luma plane lines to be fed to the vertical scaler.
336 int16_t **chrUPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
337 int16_t **chrVPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
338 int16_t **alpPixBuf; ///< Ring buffer for scaled horizontal alpha plane lines to be fed to the vertical scaler.
339 int vLumBufSize; ///< Number of vertical luma/alpha lines allocated in the ring buffer.
340 int vChrBufSize; ///< Number of vertical chroma lines allocated in the ring buffer.
341 int lastInLumBuf; ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
342 int lastInChrBuf; ///< Last scaled horizontal chroma line from source in the ring buffer.
343 int lumBufIndex; ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
344 int chrBufIndex; ///< Index in ring buffer of the last scaled horizontal chroma line from source.
347 uint8_t *formatConvBuffer;
350 * @name Horizontal and vertical filters.
351 * To better understand the following fields, here is a pseudo-code of
352 * their usage in filtering a horizontal line:
354 * for (i = 0; i < width; i++) {
356 * for (j = 0; j < filterSize; j++)
357 * dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
358 * dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
363 int16_t *hLumFilter; ///< Array of horizontal filter coefficients for luma/alpha planes.
364 int16_t *hChrFilter; ///< Array of horizontal filter coefficients for chroma planes.
365 int16_t *vLumFilter; ///< Array of vertical filter coefficients for luma/alpha planes.
366 int16_t *vChrFilter; ///< Array of vertical filter coefficients for chroma planes.
367 int32_t *hLumFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
368 int32_t *hChrFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for chroma planes.
369 int32_t *vLumFilterPos; ///< Array of vertical filter starting positions for each dst[i] for luma/alpha planes.
370 int32_t *vChrFilterPos; ///< Array of vertical filter starting positions for each dst[i] for chroma planes.
371 int hLumFilterSize; ///< Horizontal filter size for luma/alpha pixels.
372 int hChrFilterSize; ///< Horizontal filter size for chroma pixels.
373 int vLumFilterSize; ///< Vertical filter size for luma/alpha pixels.
374 int vChrFilterSize; ///< Vertical filter size for chroma pixels.
377 int lumMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
378 int chrMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
379 uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
380 uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
384 int dstY; ///< Last destination vertical line output from last slice.
385 int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
386 void *yuvTable; // pointer to the yuv->rgb table start so it can be freed()
387 // alignment ensures the offset can be added in a single
388 // instruction on e.g. ARM
389 DECLARE_ALIGNED(16, int, table_gV)[256 + 2*YUVRGB_TABLE_HEADROOM];
390 uint8_t *table_rV[256 + 2*YUVRGB_TABLE_HEADROOM];
391 uint8_t *table_gU[256 + 2*YUVRGB_TABLE_HEADROOM];
392 uint8_t *table_bU[256 + 2*YUVRGB_TABLE_HEADROOM];
393 DECLARE_ALIGNED(16, int32_t, input_rgb2yuv_table)[16+40*4]; // This table can contain both C and SIMD formatted values, the C vales are always at the XY_IDX points
403 #define RGB2YUV_SHIFT 15
405 int *dither_error[4];
408 int contrast, brightness, saturation; // for sws_getColorspaceDetails
409 int srcColorspaceTable[4];
410 int dstColorspaceTable[4];
411 int srcRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (source image).
412 int dstRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (destination image).
421 int yuv2rgb_y_offset;
423 int yuv2rgb_v2r_coeff;
424 int yuv2rgb_v2g_coeff;
425 int yuv2rgb_u2g_coeff;
426 int yuv2rgb_u2b_coeff;
428 #define RED_DITHER "0*8"
429 #define GREEN_DITHER "1*8"
430 #define BLUE_DITHER "2*8"
431 #define Y_COEFF "3*8"
432 #define VR_COEFF "4*8"
433 #define UB_COEFF "5*8"
434 #define VG_COEFF "6*8"
435 #define UG_COEFF "7*8"
436 #define Y_OFFSET "8*8"
437 #define U_OFFSET "9*8"
438 #define V_OFFSET "10*8"
439 #define LUM_MMX_FILTER_OFFSET "11*8"
440 #define CHR_MMX_FILTER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)
441 #define DSTW_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2"
442 #define ESP_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+8"
443 #define VROUNDER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+16"
444 #define U_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+24"
445 #define V_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+32"
446 #define Y_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+40"
447 #define ALP_MMX_FILTER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+48"
448 #define UV_OFF_PX "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+48"
449 #define UV_OFF_BYTE "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+56"
450 #define DITHER16 "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+64"
451 #define DITHER32 "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+80"
452 #define DITHER32_INT (11*8+4*4*MAX_FILTER_SIZE*3+80) // value equal to above, used for checking that the struct hasn't been changed by mistake
454 DECLARE_ALIGNED(8, uint64_t, redDither);
455 DECLARE_ALIGNED(8, uint64_t, greenDither);
456 DECLARE_ALIGNED(8, uint64_t, blueDither);
458 DECLARE_ALIGNED(8, uint64_t, yCoeff);
459 DECLARE_ALIGNED(8, uint64_t, vrCoeff);
460 DECLARE_ALIGNED(8, uint64_t, ubCoeff);
461 DECLARE_ALIGNED(8, uint64_t, vgCoeff);
462 DECLARE_ALIGNED(8, uint64_t, ugCoeff);
463 DECLARE_ALIGNED(8, uint64_t, yOffset);
464 DECLARE_ALIGNED(8, uint64_t, uOffset);
465 DECLARE_ALIGNED(8, uint64_t, vOffset);
466 int32_t lumMmxFilter[4 * MAX_FILTER_SIZE];
467 int32_t chrMmxFilter[4 * MAX_FILTER_SIZE];
468 int dstW; ///< Width of destination luma/alpha planes.
469 DECLARE_ALIGNED(8, uint64_t, esp);
470 DECLARE_ALIGNED(8, uint64_t, vRounder);
471 DECLARE_ALIGNED(8, uint64_t, u_temp);
472 DECLARE_ALIGNED(8, uint64_t, v_temp);
473 DECLARE_ALIGNED(8, uint64_t, y_temp);
474 int32_t alpMmxFilter[4 * MAX_FILTER_SIZE];
475 // alignment of these values is not necessary, but merely here
476 // to maintain the same offset across x8632 and x86-64. Once we
477 // use proper offset macros in the asm, they can be removed.
478 DECLARE_ALIGNED(8, ptrdiff_t, uv_off); ///< offset (in pixels) between u and v planes
479 DECLARE_ALIGNED(8, ptrdiff_t, uv_offx2); ///< offset (in bytes) between u and v planes
480 DECLARE_ALIGNED(8, uint16_t, dither16)[8];
481 DECLARE_ALIGNED(8, uint32_t, dither32)[8];
483 const uint8_t *chrDither8, *lumDither8;
486 vector signed short CY;
487 vector signed short CRV;
488 vector signed short CBU;
489 vector signed short CGU;
490 vector signed short CGV;
491 vector signed short OY;
492 vector unsigned short CSHIFT;
493 vector signed short *vYCoeffsBank, *vCCoeffsBank;
498 /* pre defined color-spaces gamma */
499 #define XYZ_GAMMA (2.6f)
500 #define RGB_GAMMA (2.2f)
503 int16_t *xyzgammainv;
504 int16_t *rgbgammainv;
505 int16_t xyz2rgb_matrix[3][4];
506 int16_t rgb2xyz_matrix[3][4];
508 /* function pointers for swscale() */
509 yuv2planar1_fn yuv2plane1;
510 yuv2planarX_fn yuv2planeX;
511 yuv2interleavedX_fn yuv2nv12cX;
512 yuv2packed1_fn yuv2packed1;
513 yuv2packed2_fn yuv2packed2;
514 yuv2packedX_fn yuv2packedX;
515 yuv2anyX_fn yuv2anyX;
517 /// Unscaled conversion of luma plane to YV12 for horizontal scaler.
518 void (*lumToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
519 int width, uint32_t *pal);
520 /// Unscaled conversion of alpha plane to YV12 for horizontal scaler.
521 void (*alpToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
522 int width, uint32_t *pal);
523 /// Unscaled conversion of chroma planes to YV12 for horizontal scaler.
524 void (*chrToYV12)(uint8_t *dstU, uint8_t *dstV,
525 const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
526 int width, uint32_t *pal);
529 * Functions to read planar input, such as planar RGB, and convert
530 * internally to Y/UV/A.
533 void (*readLumPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
534 void (*readChrPlanar)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4],
535 int width, int32_t *rgb2yuv);
536 void (*readAlpPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
540 * Scale one horizontal line of input data using a bilinear filter
541 * to produce one line of output data. Compared to SwsContext->hScale(),
542 * please take note of the following caveats when using these:
543 * - Scaling is done using only 7bit instead of 14bit coefficients.
544 * - You can use no more than 5 input pixels to produce 4 output
545 * pixels. Therefore, this filter should not be used for downscaling
546 * by more than ~20% in width (because that equals more than 5/4th
547 * downscaling and thus more than 5 pixels input per 4 pixels output).
548 * - In general, bilinear filters create artifacts during downscaling
549 * (even when <20%), because one output pixel will span more than one
550 * input pixel, and thus some pixels will need edges of both neighbor
551 * pixels to interpolate the output pixel. Since you can use at most
552 * two input pixels per output pixel in bilinear scaling, this is
553 * impossible and thus downscaling by any size will create artifacts.
554 * To enable this type of scaling, set SWS_FLAG_FAST_BILINEAR
555 * in SwsContext->flags.
558 void (*hyscale_fast)(struct SwsContext *c,
559 int16_t *dst, int dstWidth,
560 const uint8_t *src, int srcW, int xInc);
561 void (*hcscale_fast)(struct SwsContext *c,
562 int16_t *dst1, int16_t *dst2, int dstWidth,
563 const uint8_t *src1, const uint8_t *src2,
568 * Scale one horizontal line of input data using a filter over the input
569 * lines, to produce one (differently sized) line of output data.
571 * @param dst pointer to destination buffer for horizontally scaled
572 * data. If the number of bits per component of one
573 * destination pixel (SwsContext->dstBpc) is <= 10, data
574 * will be 15bpc in 16bits (int16_t) width. Else (i.e.
575 * SwsContext->dstBpc == 16), data will be 19bpc in
576 * 32bits (int32_t) width.
577 * @param dstW width of destination image
578 * @param src pointer to source data to be scaled. If the number of
579 * bits per component of a source pixel (SwsContext->srcBpc)
580 * is 8, this is 8bpc in 8bits (uint8_t) width. Else
581 * (i.e. SwsContext->dstBpc > 8), this is native depth
582 * in 16bits (uint16_t) width. In other words, for 9-bit
583 * YUV input, this is 9bpc, for 10-bit YUV input, this is
584 * 10bpc, and for 16-bit RGB or YUV, this is 16bpc.
585 * @param filter filter coefficients to be used per output pixel for
586 * scaling. This contains 14bpp filtering coefficients.
587 * Guaranteed to contain dstW * filterSize entries.
588 * @param filterPos position of the first input pixel to be used for
589 * each output pixel during scaling. Guaranteed to
590 * contain dstW entries.
591 * @param filterSize the number of input coefficients to be used (and
592 * thus the number of input pixels to be used) for
593 * creating a single output pixel. Is aligned to 4
594 * (and input coefficients thus padded with zeroes)
595 * to simplify creating SIMD code.
598 void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
599 const uint8_t *src, const int16_t *filter,
600 const int32_t *filterPos, int filterSize);
601 void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
602 const uint8_t *src, const int16_t *filter,
603 const int32_t *filterPos, int filterSize);
606 /// Color range conversion function for luma plane if needed.
607 void (*lumConvertRange)(int16_t *dst, int width);
608 /// Color range conversion function for chroma planes if needed.
609 void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width);
611 int needs_hcscale; ///< Set if there are chroma planes to be converted.
615 //FIXME check init (where 0)
617 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c);
618 int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
619 int fullRange, int brightness,
620 int contrast, int saturation);
621 void ff_yuv2rgb_init_tables_ppc(SwsContext *c, const int inv_table[4],
622 int brightness, int contrast, int saturation);
624 void ff_updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
625 int lastInLumBuf, int lastInChrBuf);
627 av_cold void ff_sws_init_range_convert(SwsContext *c);
629 SwsFunc ff_yuv2rgb_init_x86(SwsContext *c);
630 SwsFunc ff_yuv2rgb_init_ppc(SwsContext *c);
632 static av_always_inline int is16BPS(enum AVPixelFormat pix_fmt)
634 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
636 return desc->comp[0].depth_minus1 == 15;
639 static av_always_inline int is9_OR_10BPS(enum AVPixelFormat pix_fmt)
641 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
643 return desc->comp[0].depth_minus1 >= 8 && desc->comp[0].depth_minus1 <= 13;
646 #define isNBPS(x) is9_OR_10BPS(x)
648 static av_always_inline int isBE(enum AVPixelFormat pix_fmt)
650 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
652 return desc->flags & AV_PIX_FMT_FLAG_BE;
655 static av_always_inline int isYUV(enum AVPixelFormat pix_fmt)
657 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
659 return !(desc->flags & AV_PIX_FMT_FLAG_RGB) && desc->nb_components >= 2;
662 static av_always_inline int isPlanarYUV(enum AVPixelFormat pix_fmt)
664 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
666 return ((desc->flags & AV_PIX_FMT_FLAG_PLANAR) && isYUV(pix_fmt));
669 static av_always_inline int isRGB(enum AVPixelFormat pix_fmt)
671 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
673 return (desc->flags & AV_PIX_FMT_FLAG_RGB);
678 (!(av_pix_fmt_desc_get(x)->flags & AV_PIX_FMT_FLAG_PAL) && \
679 av_pix_fmt_desc_get(x)->nb_components <= 2)
682 ((x) == AV_PIX_FMT_GRAY8 || \
683 (x) == AV_PIX_FMT_YA8 || \
684 (x) == AV_PIX_FMT_GRAY16BE || \
685 (x) == AV_PIX_FMT_GRAY16LE || \
686 (x) == AV_PIX_FMT_YA16BE || \
687 (x) == AV_PIX_FMT_YA16LE)
690 #define isRGBinInt(x) \
692 (x) == AV_PIX_FMT_RGB48BE || \
693 (x) == AV_PIX_FMT_RGB48LE || \
694 (x) == AV_PIX_FMT_RGB32 || \
695 (x) == AV_PIX_FMT_RGB32_1 || \
696 (x) == AV_PIX_FMT_RGB24 || \
697 (x) == AV_PIX_FMT_RGB565BE || \
698 (x) == AV_PIX_FMT_RGB565LE || \
699 (x) == AV_PIX_FMT_RGB555BE || \
700 (x) == AV_PIX_FMT_RGB555LE || \
701 (x) == AV_PIX_FMT_RGB444BE || \
702 (x) == AV_PIX_FMT_RGB444LE || \
703 (x) == AV_PIX_FMT_RGB8 || \
704 (x) == AV_PIX_FMT_RGB4 || \
705 (x) == AV_PIX_FMT_RGB4_BYTE || \
706 (x) == AV_PIX_FMT_RGBA64BE || \
707 (x) == AV_PIX_FMT_RGBA64LE || \
708 (x) == AV_PIX_FMT_MONOBLACK || \
709 (x) == AV_PIX_FMT_MONOWHITE \
711 #define isBGRinInt(x) \
713 (x) == AV_PIX_FMT_BGR48BE || \
714 (x) == AV_PIX_FMT_BGR48LE || \
715 (x) == AV_PIX_FMT_BGR32 || \
716 (x) == AV_PIX_FMT_BGR32_1 || \
717 (x) == AV_PIX_FMT_BGR24 || \
718 (x) == AV_PIX_FMT_BGR565BE || \
719 (x) == AV_PIX_FMT_BGR565LE || \
720 (x) == AV_PIX_FMT_BGR555BE || \
721 (x) == AV_PIX_FMT_BGR555LE || \
722 (x) == AV_PIX_FMT_BGR444BE || \
723 (x) == AV_PIX_FMT_BGR444LE || \
724 (x) == AV_PIX_FMT_BGR8 || \
725 (x) == AV_PIX_FMT_BGR4 || \
726 (x) == AV_PIX_FMT_BGR4_BYTE || \
727 (x) == AV_PIX_FMT_BGRA64BE || \
728 (x) == AV_PIX_FMT_BGRA64LE || \
729 (x) == AV_PIX_FMT_MONOBLACK || \
730 (x) == AV_PIX_FMT_MONOWHITE \
733 #define isRGBinBytes(x) ( \
734 (x) == AV_PIX_FMT_RGB48BE \
735 || (x) == AV_PIX_FMT_RGB48LE \
736 || (x) == AV_PIX_FMT_RGBA64BE \
737 || (x) == AV_PIX_FMT_RGBA64LE \
738 || (x) == AV_PIX_FMT_RGBA \
739 || (x) == AV_PIX_FMT_ARGB \
740 || (x) == AV_PIX_FMT_RGB24 \
742 #define isBGRinBytes(x) ( \
743 (x) == AV_PIX_FMT_BGR48BE \
744 || (x) == AV_PIX_FMT_BGR48LE \
745 || (x) == AV_PIX_FMT_BGRA64BE \
746 || (x) == AV_PIX_FMT_BGRA64LE \
747 || (x) == AV_PIX_FMT_BGRA \
748 || (x) == AV_PIX_FMT_ABGR \
749 || (x) == AV_PIX_FMT_BGR24 \
752 #define isBayer(x) ( \
753 (x)==AV_PIX_FMT_BAYER_BGGR8 \
754 || (x)==AV_PIX_FMT_BAYER_BGGR16LE \
755 || (x)==AV_PIX_FMT_BAYER_BGGR16BE \
756 || (x)==AV_PIX_FMT_BAYER_RGGB8 \
757 || (x)==AV_PIX_FMT_BAYER_RGGB16LE \
758 || (x)==AV_PIX_FMT_BAYER_RGGB16BE \
759 || (x)==AV_PIX_FMT_BAYER_GBRG8 \
760 || (x)==AV_PIX_FMT_BAYER_GBRG16LE \
761 || (x)==AV_PIX_FMT_BAYER_GBRG16BE \
762 || (x)==AV_PIX_FMT_BAYER_GRBG8 \
763 || (x)==AV_PIX_FMT_BAYER_GRBG16LE \
764 || (x)==AV_PIX_FMT_BAYER_GRBG16BE \
767 #define isAnyRGB(x) \
775 static av_always_inline int isALPHA(enum AVPixelFormat pix_fmt)
777 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
779 if (pix_fmt == AV_PIX_FMT_PAL8)
781 return desc->flags & AV_PIX_FMT_FLAG_ALPHA;
785 #define isPacked(x) ( \
786 (x)==AV_PIX_FMT_PAL8 \
787 || (x)==AV_PIX_FMT_YUYV422 \
788 || (x)==AV_PIX_FMT_YVYU422 \
789 || (x)==AV_PIX_FMT_UYVY422 \
790 || (x)==AV_PIX_FMT_YA8 \
791 || (x)==AV_PIX_FMT_YA16LE \
792 || (x)==AV_PIX_FMT_YA16BE \
797 static av_always_inline int isPacked(enum AVPixelFormat pix_fmt)
799 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
801 return ((desc->nb_components >= 2 && !(desc->flags & AV_PIX_FMT_FLAG_PLANAR)) ||
802 pix_fmt == AV_PIX_FMT_PAL8);
806 static av_always_inline int isPlanar(enum AVPixelFormat pix_fmt)
808 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
810 return (desc->nb_components >= 2 && (desc->flags & AV_PIX_FMT_FLAG_PLANAR));
813 static av_always_inline int isPackedRGB(enum AVPixelFormat pix_fmt)
815 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
817 return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) == AV_PIX_FMT_FLAG_RGB);
820 static av_always_inline int isPlanarRGB(enum AVPixelFormat pix_fmt)
822 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
824 return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) ==
825 (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB));
828 static av_always_inline int usePal(enum AVPixelFormat pix_fmt)
830 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
832 return (desc->flags & AV_PIX_FMT_FLAG_PAL) || (desc->flags & AV_PIX_FMT_FLAG_PSEUDOPAL);
835 extern const uint64_t ff_dither4[2];
836 extern const uint64_t ff_dither8[2];
838 extern const uint8_t ff_dither_2x2_4[3][8];
839 extern const uint8_t ff_dither_2x2_8[3][8];
840 extern const uint8_t ff_dither_4x4_16[5][8];
841 extern const uint8_t ff_dither_8x8_32[9][8];
842 extern const uint8_t ff_dither_8x8_73[9][8];
843 extern const uint8_t ff_dither_8x8_128[9][8];
844 extern const uint8_t ff_dither_8x8_220[9][8];
846 extern const int32_t ff_yuv2rgb_coeffs[8][4];
848 extern const AVClass sws_context_class;
851 * Set c->swscale to an unscaled converter if one exists for the specific
852 * source and destination formats, bit depths, flags, etc.
854 void ff_get_unscaled_swscale(SwsContext *c);
855 void ff_get_unscaled_swscale_ppc(SwsContext *c);
856 void ff_get_unscaled_swscale_arm(SwsContext *c);
859 * Return function pointer to fastest main scaler path function depending
860 * on architecture and available optimizations.
862 SwsFunc ff_getSwsFunc(SwsContext *c);
864 void ff_sws_init_input_funcs(SwsContext *c);
865 void ff_sws_init_output_funcs(SwsContext *c,
866 yuv2planar1_fn *yuv2plane1,
867 yuv2planarX_fn *yuv2planeX,
868 yuv2interleavedX_fn *yuv2nv12cX,
869 yuv2packed1_fn *yuv2packed1,
870 yuv2packed2_fn *yuv2packed2,
871 yuv2packedX_fn *yuv2packedX,
872 yuv2anyX_fn *yuv2anyX);
873 void ff_sws_init_swscale_ppc(SwsContext *c);
874 void ff_sws_init_swscale_x86(SwsContext *c);
876 void ff_hyscale_fast_c(SwsContext *c, int16_t *dst, int dstWidth,
877 const uint8_t *src, int srcW, int xInc);
878 void ff_hcscale_fast_c(SwsContext *c, int16_t *dst1, int16_t *dst2,
879 int dstWidth, const uint8_t *src1,
880 const uint8_t *src2, int srcW, int xInc);
881 int ff_init_hscaler_mmxext(int dstW, int xInc, uint8_t *filterCode,
882 int16_t *filter, int32_t *filterPos,
884 void ff_hyscale_fast_mmxext(SwsContext *c, int16_t *dst,
885 int dstWidth, const uint8_t *src,
887 void ff_hcscale_fast_mmxext(SwsContext *c, int16_t *dst1, int16_t *dst2,
888 int dstWidth, const uint8_t *src1,
889 const uint8_t *src2, int srcW, int xInc);
891 static inline void fillPlane16(uint8_t *plane, int stride, int width, int height, int y,
892 int alpha, int bits, const int big_endian)
895 uint8_t *ptr = plane + stride * y;
896 int v = alpha ? 0xFFFF>>(15-bits) : (1<<bits);
897 for (i = 0; i < height; i++) {
898 #define FILL(wfunc) \
899 for (j = 0; j < width; j++) {\
911 #endif /* SWSCALE_SWSCALE_INTERNAL_H */