FFmpeg
 All Data Structures 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 FAST_BGR2YV12 // use 7-bit instead of 15-bit coefficients
43 
44 #define MAX_FILTER_SIZE 256
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 int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
67  int srcStride[], int srcSliceY, int srcSliceH,
68  uint8_t *dst[], int dstStride[]);
69 
70 /**
71  * Write one line of horizontally scaled data to planar output
72  * without any additional vertical scaling (or point-scaling).
73  *
74  * @param src scaled source data, 15bit for 8-10bit output,
75  * 19-bit for 16bit output (in int32_t)
76  * @param dest pointer to the output plane. For >8bit
77  * output, this is in uint16_t
78  * @param dstW width of destination in pixels
79  * @param dither ordered dither array of type int16_t and size 8
80  * @param offset Dither offset
81  */
82 typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
83  const uint8_t *dither, int offset);
84 
85 /**
86  * Write one line of horizontally scaled data to planar output
87  * with multi-point vertical scaling between input pixels.
88  *
89  * @param filter vertical luma/alpha scaling coefficients, 12bit [0,4096]
90  * @param src scaled luma (Y) or alpha (A) source data, 15bit for 8-10bit output,
91  * 19-bit for 16bit output (in int32_t)
92  * @param filterSize number of vertical input lines to scale
93  * @param dest pointer to output plane. For >8bit
94  * output, this is in uint16_t
95  * @param dstW width of destination pixels
96  * @param offset Dither offset
97  */
98 typedef void (*yuv2planarX_fn)(const int16_t *filter, int filterSize,
99  const int16_t **src, uint8_t *dest, int dstW,
100  const uint8_t *dither, int offset);
101 
102 /**
103  * Write one line of horizontally scaled chroma to interleaved output
104  * with multi-point vertical scaling between input pixels.
105  *
106  * @param c SWS scaling context
107  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
108  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
109  * 19-bit for 16bit output (in int32_t)
110  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
111  * 19-bit for 16bit output (in int32_t)
112  * @param chrFilterSize number of vertical chroma input lines to scale
113  * @param dest pointer to the output plane. For >8bit
114  * output, this is in uint16_t
115  * @param dstW width of chroma planes
116  */
117 typedef void (*yuv2interleavedX_fn)(struct SwsContext *c,
118  const int16_t *chrFilter,
119  int chrFilterSize,
120  const int16_t **chrUSrc,
121  const int16_t **chrVSrc,
122  uint8_t *dest, int dstW);
123 
124 /**
125  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
126  * output without any additional vertical scaling (or point-scaling). Note
127  * that this function may do chroma scaling, see the "uvalpha" argument.
128  *
129  * @param c SWS scaling context
130  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
131  * 19-bit for 16bit output (in int32_t)
132  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
133  * 19-bit for 16bit output (in int32_t)
134  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
135  * 19-bit for 16bit output (in int32_t)
136  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
137  * 19-bit for 16bit output (in int32_t)
138  * @param dest pointer to the output plane. For 16bit output, this is
139  * uint16_t
140  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
141  * to write into dest[]
142  * @param uvalpha chroma scaling coefficient for the second line of chroma
143  * pixels, either 2048 or 0. If 0, one chroma input is used
144  * for 2 output pixels (or if the SWS_FLAG_FULL_CHR_INT flag
145  * is set, it generates 1 output pixel). If 2048, two chroma
146  * input pixels should be averaged for 2 output pixels (this
147  * only happens if SWS_FLAG_FULL_CHR_INT is not set)
148  * @param y vertical line number for this output. This does not need
149  * to be used to calculate the offset in the destination,
150  * but can be used to generate comfort noise using dithering
151  * for some output formats.
152  */
153 typedef void (*yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc,
154  const int16_t *chrUSrc[2],
155  const int16_t *chrVSrc[2],
156  const int16_t *alpSrc, uint8_t *dest,
157  int dstW, int uvalpha, int y);
158 /**
159  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
160  * output by doing bilinear scaling between two input lines.
161  *
162  * @param c SWS scaling context
163  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
164  * 19-bit for 16bit output (in int32_t)
165  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
166  * 19-bit for 16bit output (in int32_t)
167  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
168  * 19-bit for 16bit output (in int32_t)
169  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
170  * 19-bit for 16bit output (in int32_t)
171  * @param dest pointer to the output plane. For 16bit output, this is
172  * uint16_t
173  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
174  * to write into dest[]
175  * @param yalpha luma/alpha scaling coefficients for the second input line.
176  * The first line's coefficients can be calculated by using
177  * 4096 - yalpha
178  * @param uvalpha chroma scaling coefficient for the second input line. The
179  * first line's coefficients can be calculated by using
180  * 4096 - uvalpha
181  * @param y vertical line number for this output. This does not need
182  * to be used to calculate the offset in the destination,
183  * but can be used to generate comfort noise using dithering
184  * for some output formats.
185  */
186 typedef void (*yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2],
187  const int16_t *chrUSrc[2],
188  const int16_t *chrVSrc[2],
189  const int16_t *alpSrc[2],
190  uint8_t *dest,
191  int dstW, int yalpha, int uvalpha, int y);
192 /**
193  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
194  * output by doing multi-point vertical scaling between input pixels.
195  *
196  * @param c SWS scaling context
197  * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
198  * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
199  * 19-bit for 16bit output (in int32_t)
200  * @param lumFilterSize number of vertical luma/alpha input lines to scale
201  * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
202  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
203  * 19-bit for 16bit output (in int32_t)
204  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
205  * 19-bit for 16bit output (in int32_t)
206  * @param chrFilterSize number of vertical chroma input lines to scale
207  * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
208  * 19-bit for 16bit output (in int32_t)
209  * @param dest pointer to the output plane. For 16bit output, this is
210  * uint16_t
211  * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
212  * to write into dest[]
213  * @param y vertical line number for this output. This does not need
214  * to be used to calculate the offset in the destination,
215  * but can be used to generate comfort noise using dithering
216  * or some output formats.
217  */
218 typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter,
219  const int16_t **lumSrc, int lumFilterSize,
220  const int16_t *chrFilter,
221  const int16_t **chrUSrc,
222  const int16_t **chrVSrc, int chrFilterSize,
223  const int16_t **alpSrc, uint8_t *dest,
224  int dstW, int y);
225 
226 /* This struct should be aligned on at least a 32-byte boundary. */
227 typedef struct SwsContext {
228  /**
229  * info on struct for av_log
230  */
231  const AVClass *av_class;
232 
233  /**
234  * Note that src, dst, srcStride, dstStride will be copied in the
235  * sws_scale() wrapper so they can be freely modified here.
236  */
237  SwsFunc swScale;
238  int srcW; ///< Width of source luma/alpha planes.
239  int srcH; ///< Height of source luma/alpha planes.
240  int dstH; ///< Height of destination luma/alpha planes.
241  int chrSrcW; ///< Width of source chroma planes.
242  int chrSrcH; ///< Height of source chroma planes.
243  int chrDstW; ///< Width of destination chroma planes.
244  int chrDstH; ///< Height of destination chroma planes.
245  int lumXInc, chrXInc;
246  int lumYInc, chrYInc;
247  enum AVPixelFormat dstFormat; ///< Destination pixel format.
248  enum AVPixelFormat srcFormat; ///< Source pixel format.
249  int dstFormatBpp; ///< Number of bits per pixel of the destination pixel format.
250  int srcFormatBpp; ///< Number of bits per pixel of the source pixel format.
251  int dstBpc, srcBpc;
252  int chrSrcHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source image.
253  int chrSrcVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in source image.
254  int chrDstHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination image.
255  int chrDstVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in destination image.
256  int vChrDrop; ///< Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user.
257  int sliceDir; ///< Direction that slices are fed to the scaler (1 = top-to-bottom, -1 = bottom-to-top).
258  double param[2]; ///< Input parameters for scaling algorithms that need them.
259 
260  uint32_t pal_yuv[256];
261  uint32_t pal_rgb[256];
262 
263  /**
264  * @name Scaled horizontal lines ring buffer.
265  * The horizontal scaler keeps just enough scaled lines in a ring buffer
266  * so they may be passed to the vertical scaler. The pointers to the
267  * allocated buffers for each line are duplicated in sequence in the ring
268  * buffer to simplify indexing and avoid wrapping around between lines
269  * inside the vertical scaler code. The wrapping is done before the
270  * vertical scaler is called.
271  */
272  //@{
273  int16_t **lumPixBuf; ///< Ring buffer for scaled horizontal luma plane lines to be fed to the vertical scaler.
274  int16_t **chrUPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
275  int16_t **chrVPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
276  int16_t **alpPixBuf; ///< Ring buffer for scaled horizontal alpha plane lines to be fed to the vertical scaler.
277  int vLumBufSize; ///< Number of vertical luma/alpha lines allocated in the ring buffer.
278  int vChrBufSize; ///< Number of vertical chroma lines allocated in the ring buffer.
279  int lastInLumBuf; ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
280  int lastInChrBuf; ///< Last scaled horizontal chroma line from source in the ring buffer.
281  int lumBufIndex; ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
282  int chrBufIndex; ///< Index in ring buffer of the last scaled horizontal chroma line from source.
283  //@}
284 
285  uint8_t *formatConvBuffer;
286 
287  /**
288  * @name Horizontal and vertical filters.
289  * To better understand the following fields, here is a pseudo-code of
290  * their usage in filtering a horizontal line:
291  * @code
292  * for (i = 0; i < width; i++) {
293  * dst[i] = 0;
294  * for (j = 0; j < filterSize; j++)
295  * dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
296  * dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
297  * }
298  * @endcode
299  */
300  //@{
301  int16_t *hLumFilter; ///< Array of horizontal filter coefficients for luma/alpha planes.
302  int16_t *hChrFilter; ///< Array of horizontal filter coefficients for chroma planes.
303  int16_t *vLumFilter; ///< Array of vertical filter coefficients for luma/alpha planes.
304  int16_t *vChrFilter; ///< Array of vertical filter coefficients for chroma planes.
305  int32_t *hLumFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
306  int32_t *hChrFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for chroma planes.
307  int32_t *vLumFilterPos; ///< Array of vertical filter starting positions for each dst[i] for luma/alpha planes.
308  int32_t *vChrFilterPos; ///< Array of vertical filter starting positions for each dst[i] for chroma planes.
309  int hLumFilterSize; ///< Horizontal filter size for luma/alpha pixels.
310  int hChrFilterSize; ///< Horizontal filter size for chroma pixels.
311  int vLumFilterSize; ///< Vertical filter size for luma/alpha pixels.
312  int vChrFilterSize; ///< Vertical filter size for chroma pixels.
313  //@}
314 
315  int lumMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
316  int chrMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
317  uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
318  uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
319 
320  int canMMXEXTBeUsed;
321 
322  int dstY; ///< Last destination vertical line output from last slice.
323  int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
324  void *yuvTable; // pointer to the yuv->rgb table start so it can be freed()
325  uint8_t *table_rV[256 + 2*YUVRGB_TABLE_HEADROOM];
326  uint8_t *table_gU[256 + 2*YUVRGB_TABLE_HEADROOM];
327  int table_gV[256 + 2*YUVRGB_TABLE_HEADROOM];
328  uint8_t *table_bU[256 + 2*YUVRGB_TABLE_HEADROOM];
329 
330  //Colorspace stuff
331  int contrast, brightness, saturation; // for sws_getColorspaceDetails
332  int srcColorspaceTable[4];
333  int dstColorspaceTable[4];
334  int srcRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (source image).
335  int dstRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (destination image).
336  int src0Alpha;
337  int dst0Alpha;
338  int yuv2rgb_y_offset;
339  int yuv2rgb_y_coeff;
340  int yuv2rgb_v2r_coeff;
341  int yuv2rgb_v2g_coeff;
342  int yuv2rgb_u2g_coeff;
343  int yuv2rgb_u2b_coeff;
344 
345 #define RED_DITHER "0*8"
346 #define GREEN_DITHER "1*8"
347 #define BLUE_DITHER "2*8"
348 #define Y_COEFF "3*8"
349 #define VR_COEFF "4*8"
350 #define UB_COEFF "5*8"
351 #define VG_COEFF "6*8"
352 #define UG_COEFF "7*8"
353 #define Y_OFFSET "8*8"
354 #define U_OFFSET "9*8"
355 #define V_OFFSET "10*8"
356 #define LUM_MMX_FILTER_OFFSET "11*8"
357 #define CHR_MMX_FILTER_OFFSET "11*8+4*4*256"
358 #define DSTW_OFFSET "11*8+4*4*256*2" //do not change, it is hardcoded in the ASM
359 #define ESP_OFFSET "11*8+4*4*256*2+8"
360 #define VROUNDER_OFFSET "11*8+4*4*256*2+16"
361 #define U_TEMP "11*8+4*4*256*2+24"
362 #define V_TEMP "11*8+4*4*256*2+32"
363 #define Y_TEMP "11*8+4*4*256*2+40"
364 #define ALP_MMX_FILTER_OFFSET "11*8+4*4*256*2+48"
365 #define UV_OFF_PX "11*8+4*4*256*3+48"
366 #define UV_OFF_BYTE "11*8+4*4*256*3+56"
367 #define DITHER16 "11*8+4*4*256*3+64"
368 #define DITHER32 "11*8+4*4*256*3+80"
369 
370  DECLARE_ALIGNED(8, uint64_t, redDither);
371  DECLARE_ALIGNED(8, uint64_t, greenDither);
372  DECLARE_ALIGNED(8, uint64_t, blueDither);
373 
374  DECLARE_ALIGNED(8, uint64_t, yCoeff);
375  DECLARE_ALIGNED(8, uint64_t, vrCoeff);
376  DECLARE_ALIGNED(8, uint64_t, ubCoeff);
377  DECLARE_ALIGNED(8, uint64_t, vgCoeff);
378  DECLARE_ALIGNED(8, uint64_t, ugCoeff);
379  DECLARE_ALIGNED(8, uint64_t, yOffset);
380  DECLARE_ALIGNED(8, uint64_t, uOffset);
381  DECLARE_ALIGNED(8, uint64_t, vOffset);
382  int32_t lumMmxFilter[4 * MAX_FILTER_SIZE];
383  int32_t chrMmxFilter[4 * MAX_FILTER_SIZE];
384  int dstW; ///< Width of destination luma/alpha planes.
385  DECLARE_ALIGNED(8, uint64_t, esp);
386  DECLARE_ALIGNED(8, uint64_t, vRounder);
387  DECLARE_ALIGNED(8, uint64_t, u_temp);
388  DECLARE_ALIGNED(8, uint64_t, v_temp);
389  DECLARE_ALIGNED(8, uint64_t, y_temp);
390  int32_t alpMmxFilter[4 * MAX_FILTER_SIZE];
391  // alignment of these values is not necessary, but merely here
392  // to maintain the same offset across x8632 and x86-64. Once we
393  // use proper offset macros in the asm, they can be removed.
394  DECLARE_ALIGNED(8, ptrdiff_t, uv_off); ///< offset (in pixels) between u and v planes
395  DECLARE_ALIGNED(8, ptrdiff_t, uv_offx2); ///< offset (in bytes) between u and v planes
396  DECLARE_ALIGNED(8, uint16_t, dither16)[8];
397  DECLARE_ALIGNED(8, uint32_t, dither32)[8];
398 
399  const uint8_t *chrDither8, *lumDither8;
400 
401 #if HAVE_ALTIVEC
402  vector signed short CY;
403  vector signed short CRV;
404  vector signed short CBU;
405  vector signed short CGU;
406  vector signed short CGV;
407  vector signed short OY;
408  vector unsigned short CSHIFT;
409  vector signed short *vYCoeffsBank, *vCCoeffsBank;
410 #endif
411 
412 #if ARCH_BFIN
413  DECLARE_ALIGNED(4, uint32_t, oy);
414  DECLARE_ALIGNED(4, uint32_t, oc);
415  DECLARE_ALIGNED(4, uint32_t, zero);
416  DECLARE_ALIGNED(4, uint32_t, cy);
417  DECLARE_ALIGNED(4, uint32_t, crv);
418  DECLARE_ALIGNED(4, uint32_t, rmask);
419  DECLARE_ALIGNED(4, uint32_t, cbu);
420  DECLARE_ALIGNED(4, uint32_t, bmask);
421  DECLARE_ALIGNED(4, uint32_t, cgu);
422  DECLARE_ALIGNED(4, uint32_t, cgv);
423  DECLARE_ALIGNED(4, uint32_t, gmask);
424 #endif
425 
426 #if HAVE_VIS
427  DECLARE_ALIGNED(8, uint64_t, sparc_coeffs)[10];
428 #endif
429  int use_mmx_vfilter;
430 
431  /* function pointers for swScale() */
432  yuv2planar1_fn yuv2plane1;
433  yuv2planarX_fn yuv2planeX;
434  yuv2interleavedX_fn yuv2nv12cX;
435  yuv2packed1_fn yuv2packed1;
436  yuv2packed2_fn yuv2packed2;
437  yuv2packedX_fn yuv2packedX;
438 
439  /// Unscaled conversion of luma plane to YV12 for horizontal scaler.
440  void (*lumToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
441  int width, uint32_t *pal);
442  /// Unscaled conversion of alpha plane to YV12 for horizontal scaler.
443  void (*alpToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
444  int width, uint32_t *pal);
445  /// Unscaled conversion of chroma planes to YV12 for horizontal scaler.
446  void (*chrToYV12)(uint8_t *dstU, uint8_t *dstV,
447  const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
448  int width, uint32_t *pal);
449 
450  /**
451  * Functions to read planar input, such as planar RGB, and convert
452  * internally to Y/UV.
453  */
454  /** @{ */
455  void (*readLumPlanar)(uint8_t *dst, const uint8_t *src[4], int width);
456  void (*readChrPlanar)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4],
457  int width);
458  /** @} */
459 
460  /**
461  * Scale one horizontal line of input data using a bilinear filter
462  * to produce one line of output data. Compared to SwsContext->hScale(),
463  * please take note of the following caveats when using these:
464  * - Scaling is done using only 7bit instead of 14bit coefficients.
465  * - You can use no more than 5 input pixels to produce 4 output
466  * pixels. Therefore, this filter should not be used for downscaling
467  * by more than ~20% in width (because that equals more than 5/4th
468  * downscaling and thus more than 5 pixels input per 4 pixels output).
469  * - In general, bilinear filters create artifacts during downscaling
470  * (even when <20%), because one output pixel will span more than one
471  * input pixel, and thus some pixels will need edges of both neighbor
472  * pixels to interpolate the output pixel. Since you can use at most
473  * two input pixels per output pixel in bilinear scaling, this is
474  * impossible and thus downscaling by any size will create artifacts.
475  * To enable this type of scaling, set SWS_FLAG_FAST_BILINEAR
476  * in SwsContext->flags.
477  */
478  /** @{ */
479  void (*hyscale_fast)(struct SwsContext *c,
480  int16_t *dst, int dstWidth,
481  const uint8_t *src, int srcW, int xInc);
482  void (*hcscale_fast)(struct SwsContext *c,
483  int16_t *dst1, int16_t *dst2, int dstWidth,
484  const uint8_t *src1, const uint8_t *src2,
485  int srcW, int xInc);
486  /** @} */
487 
488  /**
489  * Scale one horizontal line of input data using a filter over the input
490  * lines, to produce one (differently sized) line of output data.
491  *
492  * @param dst pointer to destination buffer for horizontally scaled
493  * data. If the number of bits per component of one
494  * destination pixel (SwsContext->dstBpc) is <= 10, data
495  * will be 15bpc in 16bits (int16_t) width. Else (i.e.
496  * SwsContext->dstBpc == 16), data will be 19bpc in
497  * 32bits (int32_t) width.
498  * @param dstW width of destination image
499  * @param src pointer to source data to be scaled. If the number of
500  * bits per component of a source pixel (SwsContext->srcBpc)
501  * is 8, this is 8bpc in 8bits (uint8_t) width. Else
502  * (i.e. SwsContext->dstBpc > 8), this is native depth
503  * in 16bits (uint16_t) width. In other words, for 9-bit
504  * YUV input, this is 9bpc, for 10-bit YUV input, this is
505  * 10bpc, and for 16-bit RGB or YUV, this is 16bpc.
506  * @param filter filter coefficients to be used per output pixel for
507  * scaling. This contains 14bpp filtering coefficients.
508  * Guaranteed to contain dstW * filterSize entries.
509  * @param filterPos position of the first input pixel to be used for
510  * each output pixel during scaling. Guaranteed to
511  * contain dstW entries.
512  * @param filterSize the number of input coefficients to be used (and
513  * thus the number of input pixels to be used) for
514  * creating a single output pixel. Is aligned to 4
515  * (and input coefficients thus padded with zeroes)
516  * to simplify creating SIMD code.
517  */
518  /** @{ */
519  void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
520  const uint8_t *src, const int16_t *filter,
521  const int32_t *filterPos, int filterSize);
522  void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
523  const uint8_t *src, const int16_t *filter,
524  const int32_t *filterPos, int filterSize);
525  /** @} */
526 
527  /// Color range conversion function for luma plane if needed.
528  void (*lumConvertRange)(int16_t *dst, int width);
529  /// Color range conversion function for chroma planes if needed.
530  void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width);
531 
532  int needs_hcscale; ///< Set if there are chroma planes to be converted.
533 } SwsContext;
534 //FIXME check init (where 0)
535 
536 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c);
537 int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
538  int fullRange, int brightness,
539  int contrast, int saturation);
540 
541 void ff_yuv2rgb_init_tables_altivec(SwsContext *c, const int inv_table[4],
542  int brightness, int contrast, int saturation);
543 void updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
544  int lastInLumBuf, int lastInChrBuf);
545 
546 SwsFunc ff_yuv2rgb_init_mmx(SwsContext *c);
547 SwsFunc ff_yuv2rgb_init_vis(SwsContext *c);
548 SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c);
549 SwsFunc ff_yuv2rgb_get_func_ptr_bfin(SwsContext *c);
550 void ff_bfin_get_unscaled_swscale(SwsContext *c);
551 
552 #if FF_API_SWS_FORMAT_NAME
553 /**
554  * @deprecated Use av_get_pix_fmt_name() instead.
555  */
556 attribute_deprecated
557 const char *sws_format_name(enum AVPixelFormat format);
558 #endif
559 
560 static av_always_inline int is16BPS(enum AVPixelFormat pix_fmt)
561 {
562  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
563  av_assert0(desc);
564  return desc->comp[0].depth_minus1 == 15;
565 }
566 
567 static av_always_inline int is9_OR_10BPS(enum AVPixelFormat pix_fmt)
568 {
569  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
570  av_assert0(desc);
571  return desc->comp[0].depth_minus1 >= 8 && desc->comp[0].depth_minus1 <= 13;
572 }
573 
574 #define isNBPS(x) is9_OR_10BPS(x)
575 
576 static av_always_inline int isBE(enum AVPixelFormat pix_fmt)
577 {
578  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
579  av_assert0(desc);
580  return desc->flags & PIX_FMT_BE;
581 }
582 
583 static av_always_inline int isYUV(enum AVPixelFormat pix_fmt)
584 {
585  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
586  av_assert0(desc);
587  return !(desc->flags & PIX_FMT_RGB) && desc->nb_components >= 2;
588 }
589 
590 static av_always_inline int isPlanarYUV(enum AVPixelFormat pix_fmt)
591 {
592  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
593  av_assert0(desc);
594  return ((desc->flags & PIX_FMT_PLANAR) && isYUV(pix_fmt));
595 }
596 
597 static av_always_inline int isRGB(enum AVPixelFormat pix_fmt)
598 {
599  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
600  av_assert0(desc);
601  return (desc->flags & PIX_FMT_RGB);
602 }
603 
604 #if 0 // FIXME
605 #define isGray(x) \
606  (!(av_pix_fmt_desc_get(x)->flags & PIX_FMT_PAL) && \
607  av_pix_fmt_desc_get(x)->nb_components <= 2)
608 #else
609 #define isGray(x) \
610  ((x) == AV_PIX_FMT_GRAY8 || \
611  (x) == AV_PIX_FMT_Y400A || \
612  (x) == AV_PIX_FMT_GRAY16BE || \
613  (x) == AV_PIX_FMT_GRAY16LE)
614 #endif
615 
616 #define isRGBinInt(x) \
617  ( \
618  (x) == AV_PIX_FMT_RGB48BE || \
619  (x) == AV_PIX_FMT_RGB48LE || \
620  (x) == AV_PIX_FMT_RGBA64BE || \
621  (x) == AV_PIX_FMT_RGBA64LE || \
622  (x) == AV_PIX_FMT_RGB32 || \
623  (x) == AV_PIX_FMT_RGB32_1 || \
624  (x) == AV_PIX_FMT_RGB24 || \
625  (x) == AV_PIX_FMT_RGB565BE || \
626  (x) == AV_PIX_FMT_RGB565LE || \
627  (x) == AV_PIX_FMT_RGB555BE || \
628  (x) == AV_PIX_FMT_RGB555LE || \
629  (x) == AV_PIX_FMT_RGB444BE || \
630  (x) == AV_PIX_FMT_RGB444LE || \
631  (x) == AV_PIX_FMT_RGB8 || \
632  (x) == AV_PIX_FMT_RGB4 || \
633  (x) == AV_PIX_FMT_RGB4_BYTE || \
634  (x) == AV_PIX_FMT_MONOBLACK || \
635  (x) == AV_PIX_FMT_MONOWHITE \
636  )
637 #define isBGRinInt(x) \
638  ( \
639  (x) == AV_PIX_FMT_BGR48BE || \
640  (x) == AV_PIX_FMT_BGR48LE || \
641  (x) == AV_PIX_FMT_BGRA64BE || \
642  (x) == AV_PIX_FMT_BGRA64LE || \
643  (x) == AV_PIX_FMT_BGR32 || \
644  (x) == AV_PIX_FMT_BGR32_1 || \
645  (x) == AV_PIX_FMT_BGR24 || \
646  (x) == AV_PIX_FMT_BGR565BE || \
647  (x) == AV_PIX_FMT_BGR565LE || \
648  (x) == AV_PIX_FMT_BGR555BE || \
649  (x) == AV_PIX_FMT_BGR555LE || \
650  (x) == AV_PIX_FMT_BGR444BE || \
651  (x) == AV_PIX_FMT_BGR444LE || \
652  (x) == AV_PIX_FMT_BGR8 || \
653  (x) == AV_PIX_FMT_BGR4 || \
654  (x) == AV_PIX_FMT_BGR4_BYTE || \
655  (x) == AV_PIX_FMT_MONOBLACK || \
656  (x) == AV_PIX_FMT_MONOWHITE \
657  )
658 
659 #define isRGBinBytes(x) ( \
660  (x) == AV_PIX_FMT_RGB48BE \
661  || (x) == AV_PIX_FMT_RGB48LE \
662  || (x) == AV_PIX_FMT_RGBA64BE \
663  || (x) == AV_PIX_FMT_RGBA64LE \
664  || (x) == AV_PIX_FMT_RGBA \
665  || (x) == AV_PIX_FMT_ARGB \
666  || (x) == AV_PIX_FMT_RGB24 \
667  )
668 #define isBGRinBytes(x) ( \
669  (x) == AV_PIX_FMT_BGR48BE \
670  || (x) == AV_PIX_FMT_BGR48LE \
671  || (x) == AV_PIX_FMT_BGRA64BE \
672  || (x) == AV_PIX_FMT_BGRA64LE \
673  || (x) == AV_PIX_FMT_BGRA \
674  || (x) == AV_PIX_FMT_ABGR \
675  || (x) == AV_PIX_FMT_BGR24 \
676  )
677 
678 #define isAnyRGB(x) \
679  ( \
680  isRGBinInt(x) || \
681  isBGRinInt(x) || \
682  (x)==AV_PIX_FMT_GBR24P \
683  )
684 
685 static av_always_inline int isALPHA(enum AVPixelFormat pix_fmt)
686 {
687  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
688  av_assert0(desc);
689  return desc->nb_components == 2 || desc->nb_components == 4;
690 }
691 
692 #if 1
693 #define isPacked(x) ( \
694  (x)==AV_PIX_FMT_PAL8 \
695  || (x)==AV_PIX_FMT_YUYV422 \
696  || (x)==AV_PIX_FMT_UYVY422 \
697  || (x)==AV_PIX_FMT_Y400A \
698  || isRGBinInt(x) \
699  || isBGRinInt(x) \
700  )
701 #else
702 static av_always_inline int isPacked(enum AVPixelFormat pix_fmt)
703 {
704  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
705  av_assert0(desc);
706  return ((desc->nb_components >= 2 && !(desc->flags & PIX_FMT_PLANAR)) ||
707  pix_fmt == AV_PIX_FMT_PAL8);
708 }
709 
710 #endif
711 static av_always_inline int isPlanar(enum AVPixelFormat pix_fmt)
712 {
713  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
714  av_assert0(desc);
715  return (desc->nb_components >= 2 && (desc->flags & PIX_FMT_PLANAR));
716 }
717 
718 static av_always_inline int isPackedRGB(enum AVPixelFormat pix_fmt)
719 {
720  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
721  av_assert0(desc);
722  return ((desc->flags & (PIX_FMT_PLANAR | PIX_FMT_RGB)) == PIX_FMT_RGB);
723 }
724 
725 static av_always_inline int isPlanarRGB(enum AVPixelFormat pix_fmt)
726 {
727  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
728  av_assert0(desc);
729  return ((desc->flags & (PIX_FMT_PLANAR | PIX_FMT_RGB)) ==
730  (PIX_FMT_PLANAR | PIX_FMT_RGB));
731 }
732 
733 static av_always_inline int usePal(enum AVPixelFormat pix_fmt)
734 {
735  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
736  av_assert0(desc);
737  return (desc->flags & PIX_FMT_PAL) || (desc->flags & PIX_FMT_PSEUDOPAL);
738 }
739 
740 extern const uint64_t ff_dither4[2];
741 extern const uint64_t ff_dither8[2];
742 extern const uint8_t dithers[8][8][8];
743 extern const uint16_t dither_scale[15][16];
744 
745 
746 extern const AVClass sws_context_class;
747 
748 /**
749  * Set c->swScale to an unscaled converter if one exists for the specific
750  * source and destination formats, bit depths, flags, etc.
751  */
752 void ff_get_unscaled_swscale(SwsContext *c);
753 
754 void ff_swscale_get_unscaled_altivec(SwsContext *c);
755 
756 /**
757  * Return function pointer to fastest main scaler path function depending
758  * on architecture and available optimizations.
759  */
760 SwsFunc ff_getSwsFunc(SwsContext *c);
761 
762 void ff_sws_init_input_funcs(SwsContext *c);
763 void ff_sws_init_output_funcs(SwsContext *c,
764  yuv2planar1_fn *yuv2plane1,
765  yuv2planarX_fn *yuv2planeX,
766  yuv2interleavedX_fn *yuv2nv12cX,
767  yuv2packed1_fn *yuv2packed1,
768  yuv2packed2_fn *yuv2packed2,
769  yuv2packedX_fn *yuv2packedX);
770 void ff_sws_init_swScale_altivec(SwsContext *c);
771 void ff_sws_init_swScale_mmx(SwsContext *c);
772 
773 static inline void fillPlane16(uint8_t *plane, int stride, int width, int height, int y,
774  int alpha, int bits, const int big_endian)
775 {
776  int i, j;
777  uint8_t *ptr = plane + stride * y;
778  int v = alpha ? 0xFFFF>>(15-bits) : (1<<bits);
779  for (i = 0; i < height; i++) {
780 #define FILL(wfunc) \
781  for (j = 0; j < width; j++) {\
782  wfunc(ptr+2*j, v);\
783  }
784  if (big_endian) {
785  FILL(AV_WB16);
786  } else {
787  FILL(AV_WL16);
788  }
789  ptr += stride;
790  }
791 }
792 
793 #endif /* SWSCALE_SWSCALE_INTERNAL_H */
Generated on Sat May 25 2013 03:58:51 for FFmpeg by   doxygen 1.8.2