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 "version.h"
31 
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"
39 
40 #define STR(s) AV_TOSTRING(s) // AV_STRINGIFY is too long
41 
42 #define YUVRGB_TABLE_HEADROOM 128
43 
44 #define MAX_FILTER_SIZE SWS_MAX_FILTER_SIZE
45 
46 #define DITHER1XBPP
47 
48 #if HAVE_BIGENDIAN
49 #define ALT32_CORR (-1)
50 #else
51 #define ALT32_CORR 1
52 #endif
53 
54 #if ARCH_X86_64
55 # define APCK_PTR2 8
56 # define APCK_COEF 16
57 # define APCK_SIZE 24
58 #else
59 # define APCK_PTR2 4
60 # define APCK_COEF 8
61 # define APCK_SIZE 16
62 #endif
63 
64 struct SwsContext;
65 
66 typedef enum SwsDither {
74 } SwsDither;
75 
76 typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
77  int srcStride[], int srcSliceY, int srcSliceH,
78  uint8_t *dst[], int dstStride[]);
79 
80 /**
81  * Write one line of horizontally scaled data to planar output
82  * without any additional vertical scaling (or point-scaling).
83  *
84  * @param src scaled source data, 15bit for 8-10bit output,
85  * 19-bit for 16bit output (in int32_t)
86  * @param dest pointer to the output plane. For >8bit
87  * output, this is in uint16_t
88  * @param dstW width of destination in pixels
89  * @param dither ordered dither array of type int16_t and size 8
90  * @param offset Dither offset
91  */
92 typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
93  const uint8_t *dither, int offset);
94 
95 /**
96  * Write one line of horizontally scaled data to planar output
97  * with multi-point vertical scaling between input pixels.
98  *
99  * @param filter vertical luma/alpha scaling coefficients, 12bit [0,4096]
100  * @param src scaled luma (Y) or alpha (A) source data, 15bit for 8-10bit output,
101  * 19-bit for 16bit output (in int32_t)
102  * @param filterSize number of vertical input lines to scale
103  * @param dest pointer to output plane. For >8bit
104  * output, this is in uint16_t
105  * @param dstW width of destination pixels
106  * @param offset Dither offset
107  */
108 typedef void (*yuv2planarX_fn)(const int16_t *filter, int filterSize,
109  const int16_t **src, uint8_t *dest, int dstW,
110  const uint8_t *dither, int offset);
111 
112 /**
113  * Write one line of horizontally scaled chroma to interleaved output
114  * with multi-point vertical scaling between input pixels.
115  *
116  * @param c SWS scaling context
117  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
118  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
119  * 19-bit for 16bit output (in int32_t)
120  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
121  * 19-bit for 16bit output (in int32_t)
122  * @param chrFilterSize number of vertical chroma input lines to scale
123  * @param dest pointer to the output plane. For >8bit
124  * output, this is in uint16_t
125  * @param dstW width of chroma planes
126  */
128  const int16_t *chrFilter,
129  int chrFilterSize,
130  const int16_t **chrUSrc,
131  const int16_t **chrVSrc,
132  uint8_t *dest, int dstW);
133 
134 /**
135  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
136  * output without any additional vertical scaling (or point-scaling). Note
137  * that this function may do chroma scaling, see the "uvalpha" argument.
138  *
139  * @param c SWS scaling context
140  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
141  * 19-bit for 16bit output (in int32_t)
142  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
143  * 19-bit for 16bit output (in int32_t)
144  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
145  * 19-bit for 16bit output (in int32_t)
146  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
147  * 19-bit for 16bit output (in int32_t)
148  * @param dest pointer to the output plane. For 16bit output, this is
149  * uint16_t
150  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
151  * to write into dest[]
152  * @param uvalpha chroma scaling coefficient for the second line of chroma
153  * pixels, either 2048 or 0. If 0, one chroma input is used
154  * for 2 output pixels (or if the SWS_FLAG_FULL_CHR_INT flag
155  * is set, it generates 1 output pixel). If 2048, two chroma
156  * input pixels should be averaged for 2 output pixels (this
157  * only happens if SWS_FLAG_FULL_CHR_INT is not set)
158  * @param y vertical line number for this output. This does not need
159  * to be used to calculate the offset in the destination,
160  * but can be used to generate comfort noise using dithering
161  * for some output formats.
162  */
163 typedef void (*yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc,
164  const int16_t *chrUSrc[2],
165  const int16_t *chrVSrc[2],
166  const int16_t *alpSrc, uint8_t *dest,
167  int dstW, int uvalpha, int y);
168 /**
169  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
170  * output by doing bilinear scaling between two input lines.
171  *
172  * @param c SWS scaling context
173  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
174  * 19-bit for 16bit output (in int32_t)
175  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
176  * 19-bit for 16bit output (in int32_t)
177  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
178  * 19-bit for 16bit output (in int32_t)
179  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
180  * 19-bit for 16bit output (in int32_t)
181  * @param dest pointer to the output plane. For 16bit output, this is
182  * uint16_t
183  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
184  * to write into dest[]
185  * @param yalpha luma/alpha scaling coefficients for the second input line.
186  * The first line's coefficients can be calculated by using
187  * 4096 - yalpha
188  * @param uvalpha chroma scaling coefficient for the second input line. The
189  * first line's coefficients can be calculated by using
190  * 4096 - uvalpha
191  * @param y vertical line number for this output. This does not need
192  * to be used to calculate the offset in the destination,
193  * but can be used to generate comfort noise using dithering
194  * for some output formats.
195  */
196 typedef void (*yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2],
197  const int16_t *chrUSrc[2],
198  const int16_t *chrVSrc[2],
199  const int16_t *alpSrc[2],
200  uint8_t *dest,
201  int dstW, int yalpha, int uvalpha, int y);
202 /**
203  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
204  * output by doing multi-point vertical scaling between input pixels.
205  *
206  * @param c SWS scaling context
207  * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
208  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
209  * 19-bit for 16bit output (in int32_t)
210  * @param lumFilterSize number of vertical luma/alpha input lines to scale
211  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
212  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
213  * 19-bit for 16bit output (in int32_t)
214  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
215  * 19-bit for 16bit output (in int32_t)
216  * @param chrFilterSize number of vertical chroma input lines to scale
217  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
218  * 19-bit for 16bit output (in int32_t)
219  * @param dest pointer to the output plane. For 16bit output, this is
220  * uint16_t
221  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
222  * to write into dest[]
223  * @param y vertical line number for this output. This does not need
224  * to be used to calculate the offset in the destination,
225  * but can be used to generate comfort noise using dithering
226  * or some output formats.
227  */
228 typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter,
229  const int16_t **lumSrc, int lumFilterSize,
230  const int16_t *chrFilter,
231  const int16_t **chrUSrc,
232  const int16_t **chrVSrc, int chrFilterSize,
233  const int16_t **alpSrc, uint8_t *dest,
234  int dstW, int y);
235 
236 /**
237  * Write one line of horizontally scaled Y/U/V/A to YUV/RGB
238  * output by doing multi-point vertical scaling between input pixels.
239  *
240  * @param c SWS scaling context
241  * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
242  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
243  * 19-bit for 16bit output (in int32_t)
244  * @param lumFilterSize number of vertical luma/alpha input lines to scale
245  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
246  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
247  * 19-bit for 16bit output (in int32_t)
248  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
249  * 19-bit for 16bit output (in int32_t)
250  * @param chrFilterSize number of vertical chroma input lines to scale
251  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
252  * 19-bit for 16bit output (in int32_t)
253  * @param dest pointer to the output planes. For 16bit output, this is
254  * uint16_t
255  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
256  * to write into dest[]
257  * @param y vertical line number for this output. This does not need
258  * to be used to calculate the offset in the destination,
259  * but can be used to generate comfort noise using dithering
260  * or some output formats.
261  */
262 typedef void (*yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter,
263  const int16_t **lumSrc, int lumFilterSize,
264  const int16_t *chrFilter,
265  const int16_t **chrUSrc,
266  const int16_t **chrVSrc, int chrFilterSize,
267  const int16_t **alpSrc, uint8_t **dest,
268  int dstW, int y);
269 
270 /* This struct should be aligned on at least a 32-byte boundary. */
271 typedef struct SwsContext {
272  /**
273  * info on struct for av_log
274  */
276 
277  /**
278  * Note that src, dst, srcStride, dstStride will be copied in the
279  * sws_scale() wrapper so they can be freely modified here.
280  */
282  int srcW; ///< Width of source luma/alpha planes.
283  int srcH; ///< Height of source luma/alpha planes.
284  int dstH; ///< Height of destination luma/alpha planes.
285  int chrSrcW; ///< Width of source chroma planes.
286  int chrSrcH; ///< Height of source chroma planes.
287  int chrDstW; ///< Width of destination chroma planes.
288  int chrDstH; ///< Height of destination chroma planes.
291  enum AVPixelFormat dstFormat; ///< Destination pixel format.
292  enum AVPixelFormat srcFormat; ///< Source pixel format.
293  int dstFormatBpp; ///< Number of bits per pixel of the destination pixel format.
294  int srcFormatBpp; ///< Number of bits per pixel of the source pixel format.
296  int chrSrcHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source image.
297  int chrSrcVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in source image.
298  int chrDstHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination image.
299  int chrDstVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in destination image.
300  int vChrDrop; ///< Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user.
301  int sliceDir; ///< Direction that slices are fed to the scaler (1 = top-to-bottom, -1 = bottom-to-top).
302  double param[2]; ///< Input parameters for scaling algorithms that need them.
303 
304  uint32_t pal_yuv[256];
305  uint32_t pal_rgb[256];
306 
307  /**
308  * @name Scaled horizontal lines ring buffer.
309  * The horizontal scaler keeps just enough scaled lines in a ring buffer
310  * so they may be passed to the vertical scaler. The pointers to the
311  * allocated buffers for each line are duplicated in sequence in the ring
312  * buffer to simplify indexing and avoid wrapping around between lines
313  * inside the vertical scaler code. The wrapping is done before the
314  * vertical scaler is called.
315  */
316  //@{
317  int16_t **lumPixBuf; ///< Ring buffer for scaled horizontal luma plane lines to be fed to the vertical scaler.
318  int16_t **chrUPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
319  int16_t **chrVPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
320  int16_t **alpPixBuf; ///< Ring buffer for scaled horizontal alpha plane lines to be fed to the vertical scaler.
321  int vLumBufSize; ///< Number of vertical luma/alpha lines allocated in the ring buffer.
322  int vChrBufSize; ///< Number of vertical chroma lines allocated in the ring buffer.
323  int lastInLumBuf; ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
324  int lastInChrBuf; ///< Last scaled horizontal chroma line from source in the ring buffer.
325  int lumBufIndex; ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
326  int chrBufIndex; ///< Index in ring buffer of the last scaled horizontal chroma line from source.
327  //@}
328 
330 
331  /**
332  * @name Horizontal and vertical filters.
333  * To better understand the following fields, here is a pseudo-code of
334  * their usage in filtering a horizontal line:
335  * @code
336  * for (i = 0; i < width; i++) {
337  * dst[i] = 0;
338  * for (j = 0; j < filterSize; j++)
339  * dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
340  * dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
341  * }
342  * @endcode
343  */
344  //@{
345  int16_t *hLumFilter; ///< Array of horizontal filter coefficients for luma/alpha planes.
346  int16_t *hChrFilter; ///< Array of horizontal filter coefficients for chroma planes.
347  int16_t *vLumFilter; ///< Array of vertical filter coefficients for luma/alpha planes.
348  int16_t *vChrFilter; ///< Array of vertical filter coefficients for chroma planes.
349  int32_t *hLumFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
350  int32_t *hChrFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for chroma planes.
351  int32_t *vLumFilterPos; ///< Array of vertical filter starting positions for each dst[i] for luma/alpha planes.
352  int32_t *vChrFilterPos; ///< Array of vertical filter starting positions for each dst[i] for chroma planes.
353  int hLumFilterSize; ///< Horizontal filter size for luma/alpha pixels.
354  int hChrFilterSize; ///< Horizontal filter size for chroma pixels.
355  int vLumFilterSize; ///< Vertical filter size for luma/alpha pixels.
356  int vChrFilterSize; ///< Vertical filter size for chroma pixels.
357  //@}
358 
359  int lumMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
360  int chrMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
361  uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
362  uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
363 
365 
366  int dstY; ///< Last destination vertical line output from last slice.
367  int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
368  void *yuvTable; // pointer to the yuv->rgb table start so it can be freed()
369  // alignment ensures the offset can be added in a single
370  // instruction on e.g. ARM
375  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
376 #define RY_IDX 0
377 #define GY_IDX 1
378 #define BY_IDX 2
379 #define RU_IDX 3
380 #define GU_IDX 4
381 #define BU_IDX 5
382 #define RV_IDX 6
383 #define GV_IDX 7
384 #define BV_IDX 8
385 #define RGB2YUV_SHIFT 15
386 
387  int *dither_error[4];
388 
389  //Colorspace stuff
390  int contrast, brightness, saturation; // for sws_getColorspaceDetails
393  int srcRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (source image).
394  int dstRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (destination image).
397  int srcXYZ;
398  int dstXYZ;
409 
410 #define RED_DITHER "0*8"
411 #define GREEN_DITHER "1*8"
412 #define BLUE_DITHER "2*8"
413 #define Y_COEFF "3*8"
414 #define VR_COEFF "4*8"
415 #define UB_COEFF "5*8"
416 #define VG_COEFF "6*8"
417 #define UG_COEFF "7*8"
418 #define Y_OFFSET "8*8"
419 #define U_OFFSET "9*8"
420 #define V_OFFSET "10*8"
421 #define LUM_MMX_FILTER_OFFSET "11*8"
422 #define CHR_MMX_FILTER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)
423 #define DSTW_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2"
424 #define ESP_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+8"
425 #define VROUNDER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+16"
426 #define U_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+24"
427 #define V_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+32"
428 #define Y_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+40"
429 #define ALP_MMX_FILTER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+48"
430 #define UV_OFF_PX "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+48"
431 #define UV_OFF_BYTE "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+56"
432 #define DITHER16 "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+64"
433 #define DITHER32 "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+80"
434 #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
435 
436  DECLARE_ALIGNED(8, uint64_t, redDither);
439 
440  DECLARE_ALIGNED(8, uint64_t, yCoeff);
441  DECLARE_ALIGNED(8, uint64_t, vrCoeff);
442  DECLARE_ALIGNED(8, uint64_t, ubCoeff);
443  DECLARE_ALIGNED(8, uint64_t, vgCoeff);
444  DECLARE_ALIGNED(8, uint64_t, ugCoeff);
445  DECLARE_ALIGNED(8, uint64_t, yOffset);
446  DECLARE_ALIGNED(8, uint64_t, uOffset);
447  DECLARE_ALIGNED(8, uint64_t, vOffset);
450  int dstW; ///< Width of destination luma/alpha planes.
451  DECLARE_ALIGNED(8, uint64_t, esp);
452  DECLARE_ALIGNED(8, uint64_t, vRounder);
453  DECLARE_ALIGNED(8, uint64_t, u_temp);
454  DECLARE_ALIGNED(8, uint64_t, v_temp);
455  DECLARE_ALIGNED(8, uint64_t, y_temp);
457  // alignment of these values is not necessary, but merely here
458  // to maintain the same offset across x8632 and x86-64. Once we
459  // use proper offset macros in the asm, they can be removed.
460  DECLARE_ALIGNED(8, ptrdiff_t, uv_off); ///< offset (in pixels) between u and v planes
461  DECLARE_ALIGNED(8, ptrdiff_t, uv_offx2); ///< offset (in bytes) between u and v planes
462  DECLARE_ALIGNED(8, uint16_t, dither16)[8];
463  DECLARE_ALIGNED(8, uint32_t, dither32)[8];
464 
466 
467 #if HAVE_ALTIVEC
468  vector signed short CY;
469  vector signed short CRV;
470  vector signed short CBU;
471  vector signed short CGU;
472  vector signed short CGV;
473  vector signed short OY;
474  vector unsigned short CSHIFT;
475  vector signed short *vYCoeffsBank, *vCCoeffsBank;
476 #endif
477 
479 
480 /* pre defined color-spaces gamma */
481 #define XYZ_GAMMA (2.6f)
482 #define RGB_GAMMA (2.2f)
483  int16_t *xyzgamma;
484  int16_t *rgbgamma;
485  int16_t *xyzgammainv;
486  int16_t *rgbgammainv;
487  int16_t xyz2rgb_matrix[3][4];
488  int16_t rgb2xyz_matrix[3][4];
489 
490  /* function pointers for swscale() */
498 
499  /// Unscaled conversion of luma plane to YV12 for horizontal scaler.
500  void (*lumToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
501  int width, uint32_t *pal);
502  /// Unscaled conversion of alpha plane to YV12 for horizontal scaler.
503  void (*alpToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
504  int width, uint32_t *pal);
505  /// Unscaled conversion of chroma planes to YV12 for horizontal scaler.
506  void (*chrToYV12)(uint8_t *dstU, uint8_t *dstV,
507  const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
508  int width, uint32_t *pal);
509 
510  /**
511  * Functions to read planar input, such as planar RGB, and convert
512  * internally to Y/UV/A.
513  */
514  /** @{ */
515  void (*readLumPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
516  void (*readChrPlanar)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4],
517  int width, int32_t *rgb2yuv);
518  void (*readAlpPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
519  /** @} */
520 
521  /**
522  * Scale one horizontal line of input data using a bilinear filter
523  * to produce one line of output data. Compared to SwsContext->hScale(),
524  * please take note of the following caveats when using these:
525  * - Scaling is done using only 7bit instead of 14bit coefficients.
526  * - You can use no more than 5 input pixels to produce 4 output
527  * pixels. Therefore, this filter should not be used for downscaling
528  * by more than ~20% in width (because that equals more than 5/4th
529  * downscaling and thus more than 5 pixels input per 4 pixels output).
530  * - In general, bilinear filters create artifacts during downscaling
531  * (even when <20%), because one output pixel will span more than one
532  * input pixel, and thus some pixels will need edges of both neighbor
533  * pixels to interpolate the output pixel. Since you can use at most
534  * two input pixels per output pixel in bilinear scaling, this is
535  * impossible and thus downscaling by any size will create artifacts.
536  * To enable this type of scaling, set SWS_FLAG_FAST_BILINEAR
537  * in SwsContext->flags.
538  */
539  /** @{ */
541  int16_t *dst, int dstWidth,
542  const uint8_t *src, int srcW, int xInc);
544  int16_t *dst1, int16_t *dst2, int dstWidth,
545  const uint8_t *src1, const uint8_t *src2,
546  int srcW, int xInc);
547  /** @} */
548 
549  /**
550  * Scale one horizontal line of input data using a filter over the input
551  * lines, to produce one (differently sized) line of output data.
552  *
553  * @param dst pointer to destination buffer for horizontally scaled
554  * data. If the number of bits per component of one
555  * destination pixel (SwsContext->dstBpc) is <= 10, data
556  * will be 15bpc in 16bits (int16_t) width. Else (i.e.
557  * SwsContext->dstBpc == 16), data will be 19bpc in
558  * 32bits (int32_t) width.
559  * @param dstW width of destination image
560  * @param src pointer to source data to be scaled. If the number of
561  * bits per component of a source pixel (SwsContext->srcBpc)
562  * is 8, this is 8bpc in 8bits (uint8_t) width. Else
563  * (i.e. SwsContext->dstBpc > 8), this is native depth
564  * in 16bits (uint16_t) width. In other words, for 9-bit
565  * YUV input, this is 9bpc, for 10-bit YUV input, this is
566  * 10bpc, and for 16-bit RGB or YUV, this is 16bpc.
567  * @param filter filter coefficients to be used per output pixel for
568  * scaling. This contains 14bpp filtering coefficients.
569  * Guaranteed to contain dstW * filterSize entries.
570  * @param filterPos position of the first input pixel to be used for
571  * each output pixel during scaling. Guaranteed to
572  * contain dstW entries.
573  * @param filterSize the number of input coefficients to be used (and
574  * thus the number of input pixels to be used) for
575  * creating a single output pixel. Is aligned to 4
576  * (and input coefficients thus padded with zeroes)
577  * to simplify creating SIMD code.
578  */
579  /** @{ */
580  void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
581  const uint8_t *src, const int16_t *filter,
582  const int32_t *filterPos, int filterSize);
583  void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
584  const uint8_t *src, const int16_t *filter,
585  const int32_t *filterPos, int filterSize);
586  /** @} */
587 
588  /// Color range conversion function for luma plane if needed.
589  void (*lumConvertRange)(int16_t *dst, int width);
590  /// Color range conversion function for chroma planes if needed.
591  void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width);
592 
593  int needs_hcscale; ///< Set if there are chroma planes to be converted.
594 
596 } SwsContext;
597 //FIXME check init (where 0)
598 
600 int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
601  int fullRange, int brightness,
602  int contrast, int saturation);
603 void ff_yuv2rgb_init_tables_ppc(SwsContext *c, const int inv_table[4],
604  int brightness, int contrast, int saturation);
605 
607  int lastInLumBuf, int lastInChrBuf);
608 
610 
613 
614 #if FF_API_SWS_FORMAT_NAME
615 /**
616  * @deprecated Use av_get_pix_fmt_name() instead.
617  */
619 const char *sws_format_name(enum AVPixelFormat format);
620 #endif
621 
623 {
624  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
625  av_assert0(desc);
626  return desc->comp[0].depth_minus1 == 15;
627 }
628 
630 {
631  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
632  av_assert0(desc);
633  return desc->comp[0].depth_minus1 >= 8 && desc->comp[0].depth_minus1 <= 13;
634 }
635 
636 #define isNBPS(x) is9_OR_10BPS(x)
637 
639 {
640  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
641  av_assert0(desc);
642  return desc->flags & AV_PIX_FMT_FLAG_BE;
643 }
644 
646 {
647  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
648  av_assert0(desc);
649  return !(desc->flags & AV_PIX_FMT_FLAG_RGB) && desc->nb_components >= 2;
650 }
651 
653 {
654  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
655  av_assert0(desc);
656  return ((desc->flags & AV_PIX_FMT_FLAG_PLANAR) && isYUV(pix_fmt));
657 }
658 
660 {
661  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
662  av_assert0(desc);
663  return (desc->flags & AV_PIX_FMT_FLAG_RGB);
664 }
665 
666 #if 0 // FIXME
667 #define isGray(x) \
668  (!(av_pix_fmt_desc_get(x)->flags & AV_PIX_FMT_FLAG_PAL) && \
669  av_pix_fmt_desc_get(x)->nb_components <= 2)
670 #else
671 #define isGray(x) \
672  ((x) == AV_PIX_FMT_GRAY8 || \
673  (x) == AV_PIX_FMT_YA8 || \
674  (x) == AV_PIX_FMT_GRAY16BE || \
675  (x) == AV_PIX_FMT_GRAY16LE || \
676  (x) == AV_PIX_FMT_YA16BE || \
677  (x) == AV_PIX_FMT_YA16LE)
678 #endif
679 
680 #define isRGBinInt(x) \
681  ( \
682  (x) == AV_PIX_FMT_RGB48BE || \
683  (x) == AV_PIX_FMT_RGB48LE || \
684  (x) == AV_PIX_FMT_RGB32 || \
685  (x) == AV_PIX_FMT_RGB32_1 || \
686  (x) == AV_PIX_FMT_RGB24 || \
687  (x) == AV_PIX_FMT_RGB565BE || \
688  (x) == AV_PIX_FMT_RGB565LE || \
689  (x) == AV_PIX_FMT_RGB555BE || \
690  (x) == AV_PIX_FMT_RGB555LE || \
691  (x) == AV_PIX_FMT_RGB444BE || \
692  (x) == AV_PIX_FMT_RGB444LE || \
693  (x) == AV_PIX_FMT_RGB8 || \
694  (x) == AV_PIX_FMT_RGB4 || \
695  (x) == AV_PIX_FMT_RGB4_BYTE || \
696  (x) == AV_PIX_FMT_RGBA64BE || \
697  (x) == AV_PIX_FMT_RGBA64LE || \
698  (x) == AV_PIX_FMT_MONOBLACK || \
699  (x) == AV_PIX_FMT_MONOWHITE \
700  )
701 #define isBGRinInt(x) \
702  ( \
703  (x) == AV_PIX_FMT_BGR48BE || \
704  (x) == AV_PIX_FMT_BGR48LE || \
705  (x) == AV_PIX_FMT_BGR32 || \
706  (x) == AV_PIX_FMT_BGR32_1 || \
707  (x) == AV_PIX_FMT_BGR24 || \
708  (x) == AV_PIX_FMT_BGR565BE || \
709  (x) == AV_PIX_FMT_BGR565LE || \
710  (x) == AV_PIX_FMT_BGR555BE || \
711  (x) == AV_PIX_FMT_BGR555LE || \
712  (x) == AV_PIX_FMT_BGR444BE || \
713  (x) == AV_PIX_FMT_BGR444LE || \
714  (x) == AV_PIX_FMT_BGR8 || \
715  (x) == AV_PIX_FMT_BGR4 || \
716  (x) == AV_PIX_FMT_BGR4_BYTE || \
717  (x) == AV_PIX_FMT_BGRA64BE || \
718  (x) == AV_PIX_FMT_BGRA64LE || \
719  (x) == AV_PIX_FMT_MONOBLACK || \
720  (x) == AV_PIX_FMT_MONOWHITE \
721  )
722 
723 #define isRGBinBytes(x) ( \
724  (x) == AV_PIX_FMT_RGB48BE \
725  || (x) == AV_PIX_FMT_RGB48LE \
726  || (x) == AV_PIX_FMT_RGBA64BE \
727  || (x) == AV_PIX_FMT_RGBA64LE \
728  || (x) == AV_PIX_FMT_RGBA \
729  || (x) == AV_PIX_FMT_ARGB \
730  || (x) == AV_PIX_FMT_RGB24 \
731  )
732 #define isBGRinBytes(x) ( \
733  (x) == AV_PIX_FMT_BGR48BE \
734  || (x) == AV_PIX_FMT_BGR48LE \
735  || (x) == AV_PIX_FMT_BGRA64BE \
736  || (x) == AV_PIX_FMT_BGRA64LE \
737  || (x) == AV_PIX_FMT_BGRA \
738  || (x) == AV_PIX_FMT_ABGR \
739  || (x) == AV_PIX_FMT_BGR24 \
740  )
741 
742 #define isBayer(x) ( \
743  (x)==AV_PIX_FMT_BAYER_BGGR8 \
744  || (x)==AV_PIX_FMT_BAYER_BGGR16LE \
745  || (x)==AV_PIX_FMT_BAYER_BGGR16BE \
746  || (x)==AV_PIX_FMT_BAYER_RGGB8 \
747  || (x)==AV_PIX_FMT_BAYER_RGGB16LE \
748  || (x)==AV_PIX_FMT_BAYER_RGGB16BE \
749  || (x)==AV_PIX_FMT_BAYER_GBRG8 \
750  || (x)==AV_PIX_FMT_BAYER_GBRG16LE \
751  || (x)==AV_PIX_FMT_BAYER_GBRG16BE \
752  || (x)==AV_PIX_FMT_BAYER_GRBG8 \
753  || (x)==AV_PIX_FMT_BAYER_GRBG16LE \
754  || (x)==AV_PIX_FMT_BAYER_GRBG16BE \
755  )
756 
757 #define isAnyRGB(x) \
758  ( \
759  isBayer(x) || \
760  isRGBinInt(x) || \
761  isBGRinInt(x) || \
762  isRGB(x) \
763  )
764 
766 {
767  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
768  av_assert0(desc);
769  if (pix_fmt == AV_PIX_FMT_PAL8)
770  return 1;
771  return desc->flags & AV_PIX_FMT_FLAG_ALPHA;
772 }
773 
774 #if 1
775 #define isPacked(x) ( \
776  (x)==AV_PIX_FMT_PAL8 \
777  || (x)==AV_PIX_FMT_YUYV422 \
778  || (x)==AV_PIX_FMT_YVYU422 \
779  || (x)==AV_PIX_FMT_UYVY422 \
780  || (x)==AV_PIX_FMT_YA8 \
781  || (x)==AV_PIX_FMT_YA16LE \
782  || (x)==AV_PIX_FMT_YA16BE \
783  || isRGBinInt(x) \
784  || isBGRinInt(x) \
785  )
786 #else
788 {
789  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
790  av_assert0(desc);
791  return ((desc->nb_components >= 2 && !(desc->flags & AV_PIX_FMT_FLAG_PLANAR)) ||
792  pix_fmt == AV_PIX_FMT_PAL8);
793 }
794 
795 #endif
797 {
798  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
799  av_assert0(desc);
800  return (desc->nb_components >= 2 && (desc->flags & AV_PIX_FMT_FLAG_PLANAR));
801 }
802 
804 {
805  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
806  av_assert0(desc);
808 }
809 
811 {
812  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
813  av_assert0(desc);
814  return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) ==
816 }
817 
819 {
820  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
821  av_assert0(desc);
822  return (desc->flags & AV_PIX_FMT_FLAG_PAL) || (desc->flags & AV_PIX_FMT_FLAG_PSEUDOPAL);
823 }
824 
825 extern const uint64_t ff_dither4[2];
826 extern const uint64_t ff_dither8[2];
827 
828 extern const uint8_t ff_dither_2x2_4[3][8];
829 extern const uint8_t ff_dither_2x2_8[3][8];
830 extern const uint8_t ff_dither_4x4_16[5][8];
831 extern const uint8_t ff_dither_8x8_32[9][8];
832 extern const uint8_t ff_dither_8x8_73[9][8];
833 extern const uint8_t ff_dither_8x8_128[9][8];
834 extern const uint8_t ff_dither_8x8_220[9][8];
835 
836 extern const int32_t ff_yuv2rgb_coeffs[8][4];
837 
838 extern const AVClass sws_context_class;
839 
840 /**
841  * Set c->swscale to an unscaled converter if one exists for the specific
842  * source and destination formats, bit depths, flags, etc.
843  */
847 
848 /**
849  * Return function pointer to fastest main scaler path function depending
850  * on architecture and available optimizations.
851  */
853 
865 
866 void ff_hyscale_fast_c(SwsContext *c, int16_t *dst, int dstWidth,
867  const uint8_t *src, int srcW, int xInc);
868 void ff_hcscale_fast_c(SwsContext *c, int16_t *dst1, int16_t *dst2,
869  int dstWidth, const uint8_t *src1,
870  const uint8_t *src2, int srcW, int xInc);
871 int ff_init_hscaler_mmxext(int dstW, int xInc, uint8_t *filterCode,
872  int16_t *filter, int32_t *filterPos,
873  int numSplits);
874 void ff_hyscale_fast_mmxext(SwsContext *c, int16_t *dst,
875  int dstWidth, const uint8_t *src,
876  int srcW, int xInc);
877 void ff_hcscale_fast_mmxext(SwsContext *c, int16_t *dst1, int16_t *dst2,
878  int dstWidth, const uint8_t *src1,
879  const uint8_t *src2, int srcW, int xInc);
880 
881 static inline void fillPlane16(uint8_t *plane, int stride, int width, int height, int y,
882  int alpha, int bits, const int big_endian)
883 {
884  int i, j;
885  uint8_t *ptr = plane + stride * y;
886  int v = alpha ? 0xFFFF>>(15-bits) : (1<<bits);
887  for (i = 0; i < height; i++) {
888 #define FILL(wfunc) \
889  for (j = 0; j < width; j++) {\
890  wfunc(ptr+2*j, v);\
891  }
892  if (big_endian) {
893  FILL(AV_WB16);
894  } else {
895  FILL(AV_WL16);
896  }
897  ptr += stride;
898  }
899 }
900 
901 #endif /* SWSCALE_SWSCALE_INTERNAL_H */