FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
swscale_internal.h
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
3  *
4  * This file is part of FFmpeg.
5  *
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.
10  *
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.
15  *
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
19  */
20 
21 #ifndef SWSCALE_SWSCALE_INTERNAL_H
22 #define SWSCALE_SWSCALE_INTERNAL_H
23 
24 #include "config.h"
25 
26 #if HAVE_ALTIVEC_H
27 #include <altivec.h>
28 #endif
29 
30 #include "libavutil/avassert.h"
31 #include "libavutil/avutil.h"
32 #include "libavutil/common.h"
33 #include "libavutil/intreadwrite.h"
34 #include "libavutil/log.h"
35 #include "libavutil/pixfmt.h"
36 #include "libavutil/pixdesc.h"
37 
38 #define STR(s) AV_TOSTRING(s) // AV_STRINGIFY is too long
39 
40 #define YUVRGB_TABLE_HEADROOM 128
41 
42 #define MAX_FILTER_SIZE 256
43 
44 #define DITHER1XBPP
45 
46 #if HAVE_BIGENDIAN
47 #define ALT32_CORR (-1)
48 #else
49 #define ALT32_CORR 1
50 #endif
51 
52 #if ARCH_X86_64
53 # define APCK_PTR2 8
54 # define APCK_COEF 16
55 # define APCK_SIZE 24
56 #else
57 # define APCK_PTR2 4
58 # define APCK_COEF 8
59 # define APCK_SIZE 16
60 #endif
61 
62 struct SwsContext;
63 
64 typedef enum SwsDither {
70 } SwsDither;
71 
72 typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
73  int srcStride[], int srcSliceY, int srcSliceH,
74  uint8_t *dst[], int dstStride[]);
75 
76 /**
77  * Write one line of horizontally scaled data to planar output
78  * without any additional vertical scaling (or point-scaling).
79  *
80  * @param src scaled source data, 15bit for 8-10bit output,
81  * 19-bit for 16bit output (in int32_t)
82  * @param dest pointer to the output plane. For >8bit
83  * output, this is in uint16_t
84  * @param dstW width of destination in pixels
85  * @param dither ordered dither array of type int16_t and size 8
86  * @param offset Dither offset
87  */
88 typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
89  const uint8_t *dither, int offset);
90 
91 /**
92  * Write one line of horizontally scaled data to planar output
93  * with multi-point vertical scaling between input pixels.
94  *
95  * @param filter vertical luma/alpha scaling coefficients, 12bit [0,4096]
96  * @param src scaled luma (Y) or alpha (A) source data, 15bit for 8-10bit output,
97  * 19-bit for 16bit output (in int32_t)
98  * @param filterSize number of vertical input lines to scale
99  * @param dest pointer to output plane. For >8bit
100  * output, this is in uint16_t
101  * @param dstW width of destination pixels
102  * @param offset Dither offset
103  */
104 typedef void (*yuv2planarX_fn)(const int16_t *filter, int filterSize,
105  const int16_t **src, uint8_t *dest, int dstW,
106  const uint8_t *dither, int offset);
107 
108 /**
109  * Write one line of horizontally scaled chroma to interleaved output
110  * with multi-point vertical scaling between input pixels.
111  *
112  * @param c SWS scaling context
113  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
114  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
115  * 19-bit for 16bit output (in int32_t)
116  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
117  * 19-bit for 16bit output (in int32_t)
118  * @param chrFilterSize number of vertical chroma input lines to scale
119  * @param dest pointer to the output plane. For >8bit
120  * output, this is in uint16_t
121  * @param dstW width of chroma planes
122  */
124  const int16_t *chrFilter,
125  int chrFilterSize,
126  const int16_t **chrUSrc,
127  const int16_t **chrVSrc,
128  uint8_t *dest, int dstW);
129 
130 /**
131  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
132  * output without any additional vertical scaling (or point-scaling). Note
133  * that this function may do chroma scaling, see the "uvalpha" argument.
134  *
135  * @param c SWS scaling context
136  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
137  * 19-bit for 16bit output (in int32_t)
138  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
139  * 19-bit for 16bit output (in int32_t)
140  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
141  * 19-bit for 16bit output (in int32_t)
142  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
143  * 19-bit for 16bit output (in int32_t)
144  * @param dest pointer to the output plane. For 16bit output, this is
145  * uint16_t
146  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
147  * to write into dest[]
148  * @param uvalpha chroma scaling coefficient for the second line of chroma
149  * pixels, either 2048 or 0. If 0, one chroma input is used
150  * for 2 output pixels (or if the SWS_FLAG_FULL_CHR_INT flag
151  * is set, it generates 1 output pixel). If 2048, two chroma
152  * input pixels should be averaged for 2 output pixels (this
153  * only happens if SWS_FLAG_FULL_CHR_INT is not set)
154  * @param y vertical line number for this output. This does not need
155  * to be used to calculate the offset in the destination,
156  * but can be used to generate comfort noise using dithering
157  * for some output formats.
158  */
159 typedef void (*yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc,
160  const int16_t *chrUSrc[2],
161  const int16_t *chrVSrc[2],
162  const int16_t *alpSrc, uint8_t *dest,
163  int dstW, int uvalpha, int y);
164 /**
165  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
166  * output by doing bilinear scaling between two input lines.
167  *
168  * @param c SWS scaling context
169  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
170  * 19-bit for 16bit output (in int32_t)
171  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
172  * 19-bit for 16bit output (in int32_t)
173  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
174  * 19-bit for 16bit output (in int32_t)
175  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
176  * 19-bit for 16bit output (in int32_t)
177  * @param dest pointer to the output plane. For 16bit output, this is
178  * uint16_t
179  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
180  * to write into dest[]
181  * @param yalpha luma/alpha scaling coefficients for the second input line.
182  * The first line's coefficients can be calculated by using
183  * 4096 - yalpha
184  * @param uvalpha chroma scaling coefficient for the second input line. The
185  * first line's coefficients can be calculated by using
186  * 4096 - uvalpha
187  * @param y vertical line number for this output. This does not need
188  * to be used to calculate the offset in the destination,
189  * but can be used to generate comfort noise using dithering
190  * for some output formats.
191  */
192 typedef void (*yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2],
193  const int16_t *chrUSrc[2],
194  const int16_t *chrVSrc[2],
195  const int16_t *alpSrc[2],
196  uint8_t *dest,
197  int dstW, int yalpha, int uvalpha, int y);
198 /**
199  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
200  * output by doing multi-point vertical scaling between input pixels.
201  *
202  * @param c SWS scaling context
203  * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
204  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
205  * 19-bit for 16bit output (in int32_t)
206  * @param lumFilterSize number of vertical luma/alpha input lines to scale
207  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
208  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
209  * 19-bit for 16bit output (in int32_t)
210  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
211  * 19-bit for 16bit output (in int32_t)
212  * @param chrFilterSize number of vertical chroma input lines to scale
213  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
214  * 19-bit for 16bit output (in int32_t)
215  * @param dest pointer to the output plane. For 16bit output, this is
216  * uint16_t
217  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
218  * to write into dest[]
219  * @param y vertical line number for this output. This does not need
220  * to be used to calculate the offset in the destination,
221  * but can be used to generate comfort noise using dithering
222  * or some output formats.
223  */
224 typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter,
225  const int16_t **lumSrc, int lumFilterSize,
226  const int16_t *chrFilter,
227  const int16_t **chrUSrc,
228  const int16_t **chrVSrc, int chrFilterSize,
229  const int16_t **alpSrc, uint8_t *dest,
230  int dstW, int y);
231 
232 /**
233  * Write one line of horizontally scaled Y/U/V/A to YUV/RGB
234  * output by doing multi-point vertical scaling between input pixels.
235  *
236  * @param c SWS scaling context
237  * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
238  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
239  * 19-bit for 16bit output (in int32_t)
240  * @param lumFilterSize number of vertical luma/alpha input lines to scale
241  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
242  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
243  * 19-bit for 16bit output (in int32_t)
244  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
245  * 19-bit for 16bit output (in int32_t)
246  * @param chrFilterSize number of vertical chroma input lines to scale
247  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
248  * 19-bit for 16bit output (in int32_t)
249  * @param dest pointer to the output planes. For 16bit output, this is
250  * uint16_t
251  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
252  * to write into dest[]
253  * @param y vertical line number for this output. This does not need
254  * to be used to calculate the offset in the destination,
255  * but can be used to generate comfort noise using dithering
256  * or some output formats.
257  */
258 typedef void (*yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter,
259  const int16_t **lumSrc, int lumFilterSize,
260  const int16_t *chrFilter,
261  const int16_t **chrUSrc,
262  const int16_t **chrVSrc, int chrFilterSize,
263  const int16_t **alpSrc, uint8_t **dest,
264  int dstW, int y);
265 
266 /* This struct should be aligned on at least a 32-byte boundary. */
267 typedef struct SwsContext {
268  /**
269  * info on struct for av_log
270  */
272 
273  /**
274  * Note that src, dst, srcStride, dstStride will be copied in the
275  * sws_scale() wrapper so they can be freely modified here.
276  */
278  int srcW; ///< Width of source luma/alpha planes.
279  int srcH; ///< Height of source luma/alpha planes.
280  int dstH; ///< Height of destination luma/alpha planes.
281  int chrSrcW; ///< Width of source chroma planes.
282  int chrSrcH; ///< Height of source chroma planes.
283  int chrDstW; ///< Width of destination chroma planes.
284  int chrDstH; ///< Height of destination chroma planes.
287  enum AVPixelFormat dstFormat; ///< Destination pixel format.
288  enum AVPixelFormat srcFormat; ///< Source pixel format.
289  int dstFormatBpp; ///< Number of bits per pixel of the destination pixel format.
290  int srcFormatBpp; ///< Number of bits per pixel of the source pixel format.
292  int chrSrcHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source image.
293  int chrSrcVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in source image.
294  int chrDstHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination image.
295  int chrDstVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in destination image.
296  int vChrDrop; ///< Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user.
297  int sliceDir; ///< Direction that slices are fed to the scaler (1 = top-to-bottom, -1 = bottom-to-top).
298  double param[2]; ///< Input parameters for scaling algorithms that need them.
299 
300  uint32_t pal_yuv[256];
301  uint32_t pal_rgb[256];
302 
303  /**
304  * @name Scaled horizontal lines ring buffer.
305  * The horizontal scaler keeps just enough scaled lines in a ring buffer
306  * so they may be passed to the vertical scaler. The pointers to the
307  * allocated buffers for each line are duplicated in sequence in the ring
308  * buffer to simplify indexing and avoid wrapping around between lines
309  * inside the vertical scaler code. The wrapping is done before the
310  * vertical scaler is called.
311  */
312  //@{
313  int16_t **lumPixBuf; ///< Ring buffer for scaled horizontal luma plane lines to be fed to the vertical scaler.
314  int16_t **chrUPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
315  int16_t **chrVPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
316  int16_t **alpPixBuf; ///< Ring buffer for scaled horizontal alpha plane lines to be fed to the vertical scaler.
317  int vLumBufSize; ///< Number of vertical luma/alpha lines allocated in the ring buffer.
318  int vChrBufSize; ///< Number of vertical chroma lines allocated in the ring buffer.
319  int lastInLumBuf; ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
320  int lastInChrBuf; ///< Last scaled horizontal chroma line from source in the ring buffer.
321  int lumBufIndex; ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
322  int chrBufIndex; ///< Index in ring buffer of the last scaled horizontal chroma line from source.
323  //@}
324 
326 
327  /**
328  * @name Horizontal and vertical filters.
329  * To better understand the following fields, here is a pseudo-code of
330  * their usage in filtering a horizontal line:
331  * @code
332  * for (i = 0; i < width; i++) {
333  * dst[i] = 0;
334  * for (j = 0; j < filterSize; j++)
335  * dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
336  * dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
337  * }
338  * @endcode
339  */
340  //@{
341  int16_t *hLumFilter; ///< Array of horizontal filter coefficients for luma/alpha planes.
342  int16_t *hChrFilter; ///< Array of horizontal filter coefficients for chroma planes.
343  int16_t *vLumFilter; ///< Array of vertical filter coefficients for luma/alpha planes.
344  int16_t *vChrFilter; ///< Array of vertical filter coefficients for chroma planes.
345  int32_t *hLumFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
346  int32_t *hChrFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for chroma planes.
347  int32_t *vLumFilterPos; ///< Array of vertical filter starting positions for each dst[i] for luma/alpha planes.
348  int32_t *vChrFilterPos; ///< Array of vertical filter starting positions for each dst[i] for chroma planes.
349  int hLumFilterSize; ///< Horizontal filter size for luma/alpha pixels.
350  int hChrFilterSize; ///< Horizontal filter size for chroma pixels.
351  int vLumFilterSize; ///< Vertical filter size for luma/alpha pixels.
352  int vChrFilterSize; ///< Vertical filter size for chroma pixels.
353  //@}
354 
355  int lumMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
356  int chrMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
357  uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
358  uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
359 
361 
362  int dstY; ///< Last destination vertical line output from last slice.
363  int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
364  void *yuvTable; // pointer to the yuv->rgb table start so it can be freed()
365  // alignment ensures the offset can be added in a single
366  // instruction on e.g. ARM
371  DECLARE_ALIGNED(16, int32_t, input_rgb2yuv_table)[16+40*4]; // This table can contain both C and SIMD formatted values, teh C vales are always at the XY_IDX points
372 #define RY_IDX 0
373 #define GY_IDX 1
374 #define BY_IDX 2
375 #define RU_IDX 3
376 #define GU_IDX 4
377 #define BU_IDX 5
378 #define RV_IDX 6
379 #define GV_IDX 7
380 #define BV_IDX 8
381 #define RGB2YUV_SHIFT 15
382 
383  int *dither_error[4];
384 
385  //Colorspace stuff
386  int contrast, brightness, saturation; // for sws_getColorspaceDetails
389  int srcRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (source image).
390  int dstRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (destination image).
393  int srcXYZ;
394  int dstXYZ;
405 
406 #define RED_DITHER "0*8"
407 #define GREEN_DITHER "1*8"
408 #define BLUE_DITHER "2*8"
409 #define Y_COEFF "3*8"
410 #define VR_COEFF "4*8"
411 #define UB_COEFF "5*8"
412 #define VG_COEFF "6*8"
413 #define UG_COEFF "7*8"
414 #define Y_OFFSET "8*8"
415 #define U_OFFSET "9*8"
416 #define V_OFFSET "10*8"
417 #define LUM_MMX_FILTER_OFFSET "11*8"
418 #define CHR_MMX_FILTER_OFFSET "11*8+4*4*256"
419 #define DSTW_OFFSET "11*8+4*4*256*2" //do not change, it is hardcoded in the ASM
420 #define ESP_OFFSET "11*8+4*4*256*2+8"
421 #define VROUNDER_OFFSET "11*8+4*4*256*2+16"
422 #define U_TEMP "11*8+4*4*256*2+24"
423 #define V_TEMP "11*8+4*4*256*2+32"
424 #define Y_TEMP "11*8+4*4*256*2+40"
425 #define ALP_MMX_FILTER_OFFSET "11*8+4*4*256*2+48"
426 #define UV_OFF_PX "11*8+4*4*256*3+48"
427 #define UV_OFF_BYTE "11*8+4*4*256*3+56"
428 #define DITHER16 "11*8+4*4*256*3+64"
429 #define DITHER32 "11*8+4*4*256*3+80"
430 
431  DECLARE_ALIGNED(8, uint64_t, redDither);
434 
435  DECLARE_ALIGNED(8, uint64_t, yCoeff);
436  DECLARE_ALIGNED(8, uint64_t, vrCoeff);
437  DECLARE_ALIGNED(8, uint64_t, ubCoeff);
438  DECLARE_ALIGNED(8, uint64_t, vgCoeff);
439  DECLARE_ALIGNED(8, uint64_t, ugCoeff);
440  DECLARE_ALIGNED(8, uint64_t, yOffset);
441  DECLARE_ALIGNED(8, uint64_t, uOffset);
442  DECLARE_ALIGNED(8, uint64_t, vOffset);
445  int dstW; ///< Width of destination luma/alpha planes.
446  DECLARE_ALIGNED(8, uint64_t, esp);
447  DECLARE_ALIGNED(8, uint64_t, vRounder);
448  DECLARE_ALIGNED(8, uint64_t, u_temp);
449  DECLARE_ALIGNED(8, uint64_t, v_temp);
450  DECLARE_ALIGNED(8, uint64_t, y_temp);
452  // alignment of these values is not necessary, but merely here
453  // to maintain the same offset across x8632 and x86-64. Once we
454  // use proper offset macros in the asm, they can be removed.
455  DECLARE_ALIGNED(8, ptrdiff_t, uv_off); ///< offset (in pixels) between u and v planes
456  DECLARE_ALIGNED(8, ptrdiff_t, uv_offx2); ///< offset (in bytes) between u and v planes
457  DECLARE_ALIGNED(8, uint16_t, dither16)[8];
458  DECLARE_ALIGNED(8, uint32_t, dither32)[8];
459 
461 
462 #if HAVE_ALTIVEC
463  vector signed short CY;
464  vector signed short CRV;
465  vector signed short CBU;
466  vector signed short CGU;
467  vector signed short CGV;
468  vector signed short OY;
469  vector unsigned short CSHIFT;
470  vector signed short *vYCoeffsBank, *vCCoeffsBank;
471 #endif
472 
473 #if ARCH_BFIN
474  DECLARE_ALIGNED(4, uint32_t, oy);
475  DECLARE_ALIGNED(4, uint32_t, oc);
476  DECLARE_ALIGNED(4, uint32_t, zero);
477  DECLARE_ALIGNED(4, uint32_t, cy);
478  DECLARE_ALIGNED(4, uint32_t, crv);
479  DECLARE_ALIGNED(4, uint32_t, rmask);
480  DECLARE_ALIGNED(4, uint32_t, cbu);
481  DECLARE_ALIGNED(4, uint32_t, bmask);
482  DECLARE_ALIGNED(4, uint32_t, cgu);
483  DECLARE_ALIGNED(4, uint32_t, cgv);
484  DECLARE_ALIGNED(4, uint32_t, gmask);
485 #endif
486 
487 #if HAVE_VIS
488  DECLARE_ALIGNED(8, uint64_t, sparc_coeffs)[10];
489 #endif
491 
492 /* pre defined color-spaces gamma */
493 #define XYZ_GAMMA (2.6f)
494 #define RGB_GAMMA (2.2f)
495  int16_t *xyzgamma;
496  int16_t *rgbgamma;
497  int16_t *xyzgammainv;
498  int16_t *rgbgammainv;
499  int16_t xyz2rgb_matrix[3][4];
500  int16_t rgb2xyz_matrix[3][4];
501 
502  /* function pointers for swscale() */
510 
511  /// Unscaled conversion of luma plane to YV12 for horizontal scaler.
512  void (*lumToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
513  int width, uint32_t *pal);
514  /// Unscaled conversion of alpha plane to YV12 for horizontal scaler.
515  void (*alpToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
516  int width, uint32_t *pal);
517  /// Unscaled conversion of chroma planes to YV12 for horizontal scaler.
518  void (*chrToYV12)(uint8_t *dstU, uint8_t *dstV,
519  const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
520  int width, uint32_t *pal);
521 
522  /**
523  * Functions to read planar input, such as planar RGB, and convert
524  * internally to Y/UV/A.
525  */
526  /** @{ */
527  void (*readLumPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
528  void (*readChrPlanar)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4],
529  int width, int32_t *rgb2yuv);
530  void (*readAlpPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
531  /** @} */
532 
533  /**
534  * Scale one horizontal line of input data using a bilinear filter
535  * to produce one line of output data. Compared to SwsContext->hScale(),
536  * please take note of the following caveats when using these:
537  * - Scaling is done using only 7bit instead of 14bit coefficients.
538  * - You can use no more than 5 input pixels to produce 4 output
539  * pixels. Therefore, this filter should not be used for downscaling
540  * by more than ~20% in width (because that equals more than 5/4th
541  * downscaling and thus more than 5 pixels input per 4 pixels output).
542  * - In general, bilinear filters create artifacts during downscaling
543  * (even when <20%), because one output pixel will span more than one
544  * input pixel, and thus some pixels will need edges of both neighbor
545  * pixels to interpolate the output pixel. Since you can use at most
546  * two input pixels per output pixel in bilinear scaling, this is
547  * impossible and thus downscaling by any size will create artifacts.
548  * To enable this type of scaling, set SWS_FLAG_FAST_BILINEAR
549  * in SwsContext->flags.
550  */
551  /** @{ */
553  int16_t *dst, int dstWidth,
554  const uint8_t *src, int srcW, int xInc);
556  int16_t *dst1, int16_t *dst2, int dstWidth,
557  const uint8_t *src1, const uint8_t *src2,
558  int srcW, int xInc);
559  /** @} */
560 
561  /**
562  * Scale one horizontal line of input data using a filter over the input
563  * lines, to produce one (differently sized) line of output data.
564  *
565  * @param dst pointer to destination buffer for horizontally scaled
566  * data. If the number of bits per component of one
567  * destination pixel (SwsContext->dstBpc) is <= 10, data
568  * will be 15bpc in 16bits (int16_t) width. Else (i.e.
569  * SwsContext->dstBpc == 16), data will be 19bpc in
570  * 32bits (int32_t) width.
571  * @param dstW width of destination image
572  * @param src pointer to source data to be scaled. If the number of
573  * bits per component of a source pixel (SwsContext->srcBpc)
574  * is 8, this is 8bpc in 8bits (uint8_t) width. Else
575  * (i.e. SwsContext->dstBpc > 8), this is native depth
576  * in 16bits (uint16_t) width. In other words, for 9-bit
577  * YUV input, this is 9bpc, for 10-bit YUV input, this is
578  * 10bpc, and for 16-bit RGB or YUV, this is 16bpc.
579  * @param filter filter coefficients to be used per output pixel for
580  * scaling. This contains 14bpp filtering coefficients.
581  * Guaranteed to contain dstW * filterSize entries.
582  * @param filterPos position of the first input pixel to be used for
583  * each output pixel during scaling. Guaranteed to
584  * contain dstW entries.
585  * @param filterSize the number of input coefficients to be used (and
586  * thus the number of input pixels to be used) for
587  * creating a single output pixel. Is aligned to 4
588  * (and input coefficients thus padded with zeroes)
589  * to simplify creating SIMD code.
590  */
591  /** @{ */
592  void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
593  const uint8_t *src, const int16_t *filter,
594  const int32_t *filterPos, int filterSize);
595  void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
596  const uint8_t *src, const int16_t *filter,
597  const int32_t *filterPos, int filterSize);
598  /** @} */
599 
600  /// Color range conversion function for luma plane if needed.
601  void (*lumConvertRange)(int16_t *dst, int width);
602  /// Color range conversion function for chroma planes if needed.
603  void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width);
604 
605  int needs_hcscale; ///< Set if there are chroma planes to be converted.
606 
608 } SwsContext;
609 //FIXME check init (where 0)
610 
612 int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
613  int fullRange, int brightness,
614  int contrast, int saturation);
615 void ff_yuv2rgb_init_tables_ppc(SwsContext *c, const int inv_table[4],
616  int brightness, int contrast, int saturation);
617 
619  int lastInLumBuf, int lastInChrBuf);
620 
625 
626 #if FF_API_SWS_FORMAT_NAME
627 /**
628  * @deprecated Use av_get_pix_fmt_name() instead.
629  */
631 const char *sws_format_name(enum AVPixelFormat format);
632 #endif
633 
635 {
636  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
637  av_assert0(desc);
638  return desc->comp[0].depth_minus1 == 15;
639 }
640 
642 {
643  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
644  av_assert0(desc);
645  return desc->comp[0].depth_minus1 >= 8 && desc->comp[0].depth_minus1 <= 13;
646 }
647 
648 #define isNBPS(x) is9_OR_10BPS(x)
649 
651 {
652  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
653  av_assert0(desc);
654  return desc->flags & AV_PIX_FMT_FLAG_BE;
655 }
656 
658 {
659  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
660  av_assert0(desc);
661  return !(desc->flags & AV_PIX_FMT_FLAG_RGB) && desc->nb_components >= 2;
662 }
663 
665 {
666  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
667  av_assert0(desc);
668  return ((desc->flags & AV_PIX_FMT_FLAG_PLANAR) && isYUV(pix_fmt));
669 }
670 
672 {
673  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
674  av_assert0(desc);
675  return (desc->flags & AV_PIX_FMT_FLAG_RGB);
676 }
677 
678 #if 0 // FIXME
679 #define isGray(x) \
680  (!(av_pix_fmt_desc_get(x)->flags & AV_PIX_FMT_FLAG_PAL) && \
681  av_pix_fmt_desc_get(x)->nb_components <= 2)
682 #else
683 #define isGray(x) \
684  ((x) == AV_PIX_FMT_GRAY8 || \
685  (x) == AV_PIX_FMT_Y400A || \
686  (x) == AV_PIX_FMT_GRAY16BE || \
687  (x) == AV_PIX_FMT_GRAY16LE)
688 #endif
689 
690 #define isRGBinInt(x) \
691  ( \
692  (x) == AV_PIX_FMT_RGB48BE || \
693  (x) == AV_PIX_FMT_RGB48LE || \
694  (x) == AV_PIX_FMT_RGBA64BE || \
695  (x) == AV_PIX_FMT_RGBA64LE || \
696  (x) == AV_PIX_FMT_RGB32 || \
697  (x) == AV_PIX_FMT_RGB32_1 || \
698  (x) == AV_PIX_FMT_RGB24 || \
699  (x) == AV_PIX_FMT_RGB565BE || \
700  (x) == AV_PIX_FMT_RGB565LE || \
701  (x) == AV_PIX_FMT_RGB555BE || \
702  (x) == AV_PIX_FMT_RGB555LE || \
703  (x) == AV_PIX_FMT_RGB444BE || \
704  (x) == AV_PIX_FMT_RGB444LE || \
705  (x) == AV_PIX_FMT_RGB8 || \
706  (x) == AV_PIX_FMT_RGB4 || \
707  (x) == AV_PIX_FMT_RGB4_BYTE || \
708  (x) == AV_PIX_FMT_MONOBLACK || \
709  (x) == AV_PIX_FMT_MONOWHITE \
710  )
711 #define isBGRinInt(x) \
712  ( \
713  (x) == AV_PIX_FMT_BGR48BE || \
714  (x) == AV_PIX_FMT_BGR48LE || \
715  (x) == AV_PIX_FMT_BGRA64BE || \
716  (x) == AV_PIX_FMT_BGRA64LE || \
717  (x) == AV_PIX_FMT_BGR32 || \
718  (x) == AV_PIX_FMT_BGR32_1 || \
719  (x) == AV_PIX_FMT_BGR24 || \
720  (x) == AV_PIX_FMT_BGR565BE || \
721  (x) == AV_PIX_FMT_BGR565LE || \
722  (x) == AV_PIX_FMT_BGR555BE || \
723  (x) == AV_PIX_FMT_BGR555LE || \
724  (x) == AV_PIX_FMT_BGR444BE || \
725  (x) == AV_PIX_FMT_BGR444LE || \
726  (x) == AV_PIX_FMT_BGR8 || \
727  (x) == AV_PIX_FMT_BGR4 || \
728  (x) == AV_PIX_FMT_BGR4_BYTE || \
729  (x) == AV_PIX_FMT_MONOBLACK || \
730  (x) == AV_PIX_FMT_MONOWHITE \
731  )
732 
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 \
741  )
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 \
750  )
751 
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 \
765  )
766 
767 #define isAnyRGB(x) \
768  ( \
769  isBayer(x) || \
770  isRGBinInt(x) || \
771  isBGRinInt(x) || \
772  isRGB(x) \
773  )
774 
776 {
777  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
778  av_assert0(desc);
779  if (pix_fmt == AV_PIX_FMT_PAL8)
780  return 1;
781  return desc->flags & AV_PIX_FMT_FLAG_ALPHA;
782 }
783 
784 #if 1
785 #define isPacked(x) ( \
786  (x)==AV_PIX_FMT_PAL8 \
787  || (x)==AV_PIX_FMT_YUYV422 \
788  || (x)==AV_PIX_FMT_UYVY422 \
789  || (x)==AV_PIX_FMT_Y400A \
790  || isRGBinInt(x) \
791  || isBGRinInt(x) \
792  )
793 #else
795 {
796  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
797  av_assert0(desc);
798  return ((desc->nb_components >= 2 && !(desc->flags & AV_PIX_FMT_FLAG_PLANAR)) ||
799  pix_fmt == AV_PIX_FMT_PAL8);
800 }
801 
802 #endif
804 {
805  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
806  av_assert0(desc);
807  return (desc->nb_components >= 2 && (desc->flags & AV_PIX_FMT_FLAG_PLANAR));
808 }
809 
811 {
812  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
813  av_assert0(desc);
815 }
816 
818 {
819  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
820  av_assert0(desc);
821  return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) ==
823 }
824 
826 {
827  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
828  av_assert0(desc);
829  return (desc->flags & AV_PIX_FMT_FLAG_PAL) || (desc->flags & AV_PIX_FMT_FLAG_PSEUDOPAL);
830 }
831 
832 extern const uint64_t ff_dither4[2];
833 extern const uint64_t ff_dither8[2];
834 
835 extern const uint8_t ff_dither_2x2_4[3][8];
836 extern const uint8_t ff_dither_2x2_8[3][8];
837 extern const uint8_t ff_dither_4x4_16[5][8];
838 extern const uint8_t ff_dither_8x8_32[9][8];
839 extern const uint8_t ff_dither_8x8_73[9][8];
840 extern const uint8_t ff_dither_8x8_128[9][8];
841 extern const uint8_t ff_dither_8x8_220[9][8];
842 
843 extern const int32_t ff_yuv2rgb_coeffs[8][4];
844 
845 extern const AVClass sws_context_class;
846 
847 /**
848  * Set c->swscale to an unscaled converter if one exists for the specific
849  * source and destination formats, bit depths, flags, etc.
850  */
855 
856 /**
857  * Return function pointer to fastest main scaler path function depending
858  * on architecture and available optimizations.
859  */
861 
873 
874 static inline void fillPlane16(uint8_t *plane, int stride, int width, int height, int y,
875  int alpha, int bits, const int big_endian)
876 {
877  int i, j;
878  uint8_t *ptr = plane + stride * y;
879  int v = alpha ? 0xFFFF>>(15-bits) : (1<<bits);
880  for (i = 0; i < height; i++) {
881 #define FILL(wfunc) \
882  for (j = 0; j < width; j++) {\
883  wfunc(ptr+2*j, v);\
884  }
885  if (big_endian) {
886  FILL(AV_WB16);
887  } else {
888  FILL(AV_WL16);
889  }
890  ptr += stride;
891  }
892 }
893 
894 #endif /* SWSCALE_SWSCALE_INTERNAL_H */