2 * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 #ifndef SWSCALE_SWSCALE_INTERNAL_H
22 #define SWSCALE_SWSCALE_INTERNAL_H
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"
38 #define STR(s) AV_TOSTRING(s) // AV_STRINGIFY is too long
40 #define YUVRGB_TABLE_HEADROOM 128
42 #define MAX_FILTER_SIZE 256
47 #define ALT32_CORR (-1)
64 typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
65 int srcStride[], int srcSliceY, int srcSliceH,
66 uint8_t *dst[], int dstStride[]);
69 * Write one line of horizontally scaled data to planar output
70 * without any additional vertical scaling (or point-scaling).
72 * @param src scaled source data, 15bit for 8-10bit output,
73 * 19-bit for 16bit output (in int32_t)
74 * @param dest pointer to the output plane. For >8bit
75 * output, this is in uint16_t
76 * @param dstW width of destination in pixels
77 * @param dither ordered dither array of type int16_t and size 8
78 * @param offset Dither offset
80 typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
81 const uint8_t *dither, int offset);
84 * Write one line of horizontally scaled data to planar output
85 * with multi-point vertical scaling between input pixels.
87 * @param filter vertical luma/alpha scaling coefficients, 12bit [0,4096]
88 * @param src scaled luma (Y) or alpha (A) source data, 15bit for 8-10bit output,
89 * 19-bit for 16bit output (in int32_t)
90 * @param filterSize number of vertical input lines to scale
91 * @param dest pointer to output plane. For >8bit
92 * output, this is in uint16_t
93 * @param dstW width of destination pixels
94 * @param offset Dither offset
96 typedef void (*yuv2planarX_fn)(const int16_t *filter, int filterSize,
97 const int16_t **src, uint8_t *dest, int dstW,
98 const uint8_t *dither, int offset);
101 * Write one line of horizontally scaled chroma to interleaved output
102 * with multi-point vertical scaling between input pixels.
104 * @param c SWS scaling context
105 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
106 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
107 * 19-bit for 16bit output (in int32_t)
108 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
109 * 19-bit for 16bit output (in int32_t)
110 * @param chrFilterSize number of vertical chroma input lines to scale
111 * @param dest pointer to the output plane. For >8bit
112 * output, this is in uint16_t
113 * @param dstW width of chroma planes
115 typedef void (*yuv2interleavedX_fn)(struct SwsContext *c,
116 const int16_t *chrFilter,
118 const int16_t **chrUSrc,
119 const int16_t **chrVSrc,
120 uint8_t *dest, int dstW);
123 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
124 * output without any additional vertical scaling (or point-scaling). Note
125 * that this function may do chroma scaling, see the "uvalpha" argument.
127 * @param c SWS scaling context
128 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
129 * 19-bit for 16bit output (in int32_t)
130 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
131 * 19-bit for 16bit output (in int32_t)
132 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
133 * 19-bit for 16bit output (in int32_t)
134 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
135 * 19-bit for 16bit output (in int32_t)
136 * @param dest pointer to the output plane. For 16bit output, this is
138 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
139 * to write into dest[]
140 * @param uvalpha chroma scaling coefficient for the second line of chroma
141 * pixels, either 2048 or 0. If 0, one chroma input is used
142 * for 2 output pixels (or if the SWS_FLAG_FULL_CHR_INT flag
143 * is set, it generates 1 output pixel). If 2048, two chroma
144 * input pixels should be averaged for 2 output pixels (this
145 * only happens if SWS_FLAG_FULL_CHR_INT is not set)
146 * @param y vertical line number for this output. This does not need
147 * to be used to calculate the offset in the destination,
148 * but can be used to generate comfort noise using dithering
149 * for some output formats.
151 typedef void (*yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc,
152 const int16_t *chrUSrc[2],
153 const int16_t *chrVSrc[2],
154 const int16_t *alpSrc, uint8_t *dest,
155 int dstW, int uvalpha, int y);
157 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
158 * output by doing bilinear scaling between two input lines.
160 * @param c SWS scaling context
161 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
162 * 19-bit for 16bit output (in int32_t)
163 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
164 * 19-bit for 16bit output (in int32_t)
165 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
166 * 19-bit for 16bit output (in int32_t)
167 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
168 * 19-bit for 16bit output (in int32_t)
169 * @param dest pointer to the output plane. For 16bit output, this is
171 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
172 * to write into dest[]
173 * @param yalpha luma/alpha scaling coefficients for the second input line.
174 * The first line's coefficients can be calculated by using
176 * @param uvalpha chroma scaling coefficient for the second input line. The
177 * first line's coefficients can be calculated by using
179 * @param y vertical line number for this output. This does not need
180 * to be used to calculate the offset in the destination,
181 * but can be used to generate comfort noise using dithering
182 * for some output formats.
184 typedef void (*yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2],
185 const int16_t *chrUSrc[2],
186 const int16_t *chrVSrc[2],
187 const int16_t *alpSrc[2],
189 int dstW, int yalpha, int uvalpha, int y);
191 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
192 * output by doing multi-point vertical scaling between input pixels.
194 * @param c SWS scaling context
195 * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
196 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
197 * 19-bit for 16bit output (in int32_t)
198 * @param lumFilterSize number of vertical luma/alpha input lines to scale
199 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
200 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
201 * 19-bit for 16bit output (in int32_t)
202 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
203 * 19-bit for 16bit output (in int32_t)
204 * @param chrFilterSize number of vertical chroma input lines to scale
205 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
206 * 19-bit for 16bit output (in int32_t)
207 * @param dest pointer to the output plane. For 16bit output, this is
209 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
210 * to write into dest[]
211 * @param y vertical line number for this output. This does not need
212 * to be used to calculate the offset in the destination,
213 * but can be used to generate comfort noise using dithering
214 * or some output formats.
216 typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter,
217 const int16_t **lumSrc, int lumFilterSize,
218 const int16_t *chrFilter,
219 const int16_t **chrUSrc,
220 const int16_t **chrVSrc, int chrFilterSize,
221 const int16_t **alpSrc, uint8_t *dest,
225 * Write one line of horizontally scaled Y/U/V/A to YUV/RGB
226 * output by doing multi-point vertical scaling between input pixels.
228 * @param c SWS scaling context
229 * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
230 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
231 * 19-bit for 16bit output (in int32_t)
232 * @param lumFilterSize number of vertical luma/alpha input lines to scale
233 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
234 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
235 * 19-bit for 16bit output (in int32_t)
236 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
237 * 19-bit for 16bit output (in int32_t)
238 * @param chrFilterSize number of vertical chroma input lines to scale
239 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
240 * 19-bit for 16bit output (in int32_t)
241 * @param dest pointer to the output planes. For 16bit output, this is
243 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
244 * to write into dest[]
245 * @param y vertical line number for this output. This does not need
246 * to be used to calculate the offset in the destination,
247 * but can be used to generate comfort noise using dithering
248 * or some output formats.
250 typedef void (*yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter,
251 const int16_t **lumSrc, int lumFilterSize,
252 const int16_t *chrFilter,
253 const int16_t **chrUSrc,
254 const int16_t **chrVSrc, int chrFilterSize,
255 const int16_t **alpSrc, uint8_t **dest,
258 /* This struct should be aligned on at least a 32-byte boundary. */
259 typedef struct SwsContext {
261 * info on struct for av_log
263 const AVClass *av_class;
266 * Note that src, dst, srcStride, dstStride will be copied in the
267 * sws_scale() wrapper so they can be freely modified here.
270 int srcW; ///< Width of source luma/alpha planes.
271 int srcH; ///< Height of source luma/alpha planes.
272 int dstH; ///< Height of destination luma/alpha planes.
273 int chrSrcW; ///< Width of source chroma planes.
274 int chrSrcH; ///< Height of source chroma planes.
275 int chrDstW; ///< Width of destination chroma planes.
276 int chrDstH; ///< Height of destination chroma planes.
277 int lumXInc, chrXInc;
278 int lumYInc, chrYInc;
279 enum AVPixelFormat dstFormat; ///< Destination pixel format.
280 enum AVPixelFormat srcFormat; ///< Source pixel format.
281 int dstFormatBpp; ///< Number of bits per pixel of the destination pixel format.
282 int srcFormatBpp; ///< Number of bits per pixel of the source pixel format.
284 int chrSrcHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source image.
285 int chrSrcVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in source image.
286 int chrDstHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination image.
287 int chrDstVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in destination image.
288 int vChrDrop; ///< Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user.
289 int sliceDir; ///< Direction that slices are fed to the scaler (1 = top-to-bottom, -1 = bottom-to-top).
290 double param[2]; ///< Input parameters for scaling algorithms that need them.
292 uint32_t pal_yuv[256];
293 uint32_t pal_rgb[256];
296 * @name Scaled horizontal lines ring buffer.
297 * The horizontal scaler keeps just enough scaled lines in a ring buffer
298 * so they may be passed to the vertical scaler. The pointers to the
299 * allocated buffers for each line are duplicated in sequence in the ring
300 * buffer to simplify indexing and avoid wrapping around between lines
301 * inside the vertical scaler code. The wrapping is done before the
302 * vertical scaler is called.
305 int16_t **lumPixBuf; ///< Ring buffer for scaled horizontal luma plane lines to be fed to the vertical scaler.
306 int16_t **chrUPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
307 int16_t **chrVPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
308 int16_t **alpPixBuf; ///< Ring buffer for scaled horizontal alpha plane lines to be fed to the vertical scaler.
309 int vLumBufSize; ///< Number of vertical luma/alpha lines allocated in the ring buffer.
310 int vChrBufSize; ///< Number of vertical chroma lines allocated in the ring buffer.
311 int lastInLumBuf; ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
312 int lastInChrBuf; ///< Last scaled horizontal chroma line from source in the ring buffer.
313 int lumBufIndex; ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
314 int chrBufIndex; ///< Index in ring buffer of the last scaled horizontal chroma line from source.
317 uint8_t *formatConvBuffer;
320 * @name Horizontal and vertical filters.
321 * To better understand the following fields, here is a pseudo-code of
322 * their usage in filtering a horizontal line:
324 * for (i = 0; i < width; i++) {
326 * for (j = 0; j < filterSize; j++)
327 * dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
328 * dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
333 int16_t *hLumFilter; ///< Array of horizontal filter coefficients for luma/alpha planes.
334 int16_t *hChrFilter; ///< Array of horizontal filter coefficients for chroma planes.
335 int16_t *vLumFilter; ///< Array of vertical filter coefficients for luma/alpha planes.
336 int16_t *vChrFilter; ///< Array of vertical filter coefficients for chroma planes.
337 int32_t *hLumFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
338 int32_t *hChrFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for chroma planes.
339 int32_t *vLumFilterPos; ///< Array of vertical filter starting positions for each dst[i] for luma/alpha planes.
340 int32_t *vChrFilterPos; ///< Array of vertical filter starting positions for each dst[i] for chroma planes.
341 int hLumFilterSize; ///< Horizontal filter size for luma/alpha pixels.
342 int hChrFilterSize; ///< Horizontal filter size for chroma pixels.
343 int vLumFilterSize; ///< Vertical filter size for luma/alpha pixels.
344 int vChrFilterSize; ///< Vertical filter size for chroma pixels.
347 int lumMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
348 int chrMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
349 uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
350 uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
354 int dstY; ///< Last destination vertical line output from last slice.
355 int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
356 void *yuvTable; // pointer to the yuv->rgb table start so it can be freed()
357 uint8_t *table_rV[256 + 2*YUVRGB_TABLE_HEADROOM];
358 uint8_t *table_gU[256 + 2*YUVRGB_TABLE_HEADROOM];
359 int table_gV[256 + 2*YUVRGB_TABLE_HEADROOM];
360 uint8_t *table_bU[256 + 2*YUVRGB_TABLE_HEADROOM];
361 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
371 #define RGB2YUV_SHIFT 15
373 int *dither_error[4];
376 int contrast, brightness, saturation; // for sws_getColorspaceDetails
377 int srcColorspaceTable[4];
378 int dstColorspaceTable[4];
379 int srcRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (source image).
380 int dstRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (destination image).
389 int yuv2rgb_y_offset;
391 int yuv2rgb_v2r_coeff;
392 int yuv2rgb_v2g_coeff;
393 int yuv2rgb_u2g_coeff;
394 int yuv2rgb_u2b_coeff;
396 #define RED_DITHER "0*8"
397 #define GREEN_DITHER "1*8"
398 #define BLUE_DITHER "2*8"
399 #define Y_COEFF "3*8"
400 #define VR_COEFF "4*8"
401 #define UB_COEFF "5*8"
402 #define VG_COEFF "6*8"
403 #define UG_COEFF "7*8"
404 #define Y_OFFSET "8*8"
405 #define U_OFFSET "9*8"
406 #define V_OFFSET "10*8"
407 #define LUM_MMX_FILTER_OFFSET "11*8"
408 #define CHR_MMX_FILTER_OFFSET "11*8+4*4*256"
409 #define DSTW_OFFSET "11*8+4*4*256*2" //do not change, it is hardcoded in the ASM
410 #define ESP_OFFSET "11*8+4*4*256*2+8"
411 #define VROUNDER_OFFSET "11*8+4*4*256*2+16"
412 #define U_TEMP "11*8+4*4*256*2+24"
413 #define V_TEMP "11*8+4*4*256*2+32"
414 #define Y_TEMP "11*8+4*4*256*2+40"
415 #define ALP_MMX_FILTER_OFFSET "11*8+4*4*256*2+48"
416 #define UV_OFF_PX "11*8+4*4*256*3+48"
417 #define UV_OFF_BYTE "11*8+4*4*256*3+56"
418 #define DITHER16 "11*8+4*4*256*3+64"
419 #define DITHER32 "11*8+4*4*256*3+80"
421 DECLARE_ALIGNED(8, uint64_t, redDither);
422 DECLARE_ALIGNED(8, uint64_t, greenDither);
423 DECLARE_ALIGNED(8, uint64_t, blueDither);
425 DECLARE_ALIGNED(8, uint64_t, yCoeff);
426 DECLARE_ALIGNED(8, uint64_t, vrCoeff);
427 DECLARE_ALIGNED(8, uint64_t, ubCoeff);
428 DECLARE_ALIGNED(8, uint64_t, vgCoeff);
429 DECLARE_ALIGNED(8, uint64_t, ugCoeff);
430 DECLARE_ALIGNED(8, uint64_t, yOffset);
431 DECLARE_ALIGNED(8, uint64_t, uOffset);
432 DECLARE_ALIGNED(8, uint64_t, vOffset);
433 int32_t lumMmxFilter[4 * MAX_FILTER_SIZE];
434 int32_t chrMmxFilter[4 * MAX_FILTER_SIZE];
435 int dstW; ///< Width of destination luma/alpha planes.
436 DECLARE_ALIGNED(8, uint64_t, esp);
437 DECLARE_ALIGNED(8, uint64_t, vRounder);
438 DECLARE_ALIGNED(8, uint64_t, u_temp);
439 DECLARE_ALIGNED(8, uint64_t, v_temp);
440 DECLARE_ALIGNED(8, uint64_t, y_temp);
441 int32_t alpMmxFilter[4 * MAX_FILTER_SIZE];
442 // alignment of these values is not necessary, but merely here
443 // to maintain the same offset across x8632 and x86-64. Once we
444 // use proper offset macros in the asm, they can be removed.
445 DECLARE_ALIGNED(8, ptrdiff_t, uv_off); ///< offset (in pixels) between u and v planes
446 DECLARE_ALIGNED(8, ptrdiff_t, uv_offx2); ///< offset (in bytes) between u and v planes
447 DECLARE_ALIGNED(8, uint16_t, dither16)[8];
448 DECLARE_ALIGNED(8, uint32_t, dither32)[8];
450 const uint8_t *chrDither8, *lumDither8;
453 vector signed short CY;
454 vector signed short CRV;
455 vector signed short CBU;
456 vector signed short CGU;
457 vector signed short CGV;
458 vector signed short OY;
459 vector unsigned short CSHIFT;
460 vector signed short *vYCoeffsBank, *vCCoeffsBank;
464 DECLARE_ALIGNED(4, uint32_t, oy);
465 DECLARE_ALIGNED(4, uint32_t, oc);
466 DECLARE_ALIGNED(4, uint32_t, zero);
467 DECLARE_ALIGNED(4, uint32_t, cy);
468 DECLARE_ALIGNED(4, uint32_t, crv);
469 DECLARE_ALIGNED(4, uint32_t, rmask);
470 DECLARE_ALIGNED(4, uint32_t, cbu);
471 DECLARE_ALIGNED(4, uint32_t, bmask);
472 DECLARE_ALIGNED(4, uint32_t, cgu);
473 DECLARE_ALIGNED(4, uint32_t, cgv);
474 DECLARE_ALIGNED(4, uint32_t, gmask);
478 DECLARE_ALIGNED(8, uint64_t, sparc_coeffs)[10];
482 /* pre defined color-spaces gamma */
483 #define XYZ_GAMMA (2.6f)
484 #define RGB_GAMMA (2.2f)
487 int16_t xyz2rgb_matrix[3][4];
489 /* function pointers for swScale() */
490 yuv2planar1_fn yuv2plane1;
491 yuv2planarX_fn yuv2planeX;
492 yuv2interleavedX_fn yuv2nv12cX;
493 yuv2packed1_fn yuv2packed1;
494 yuv2packed2_fn yuv2packed2;
495 yuv2packedX_fn yuv2packedX;
496 yuv2anyX_fn yuv2anyX;
498 /// Unscaled conversion of luma plane to YV12 for horizontal scaler.
499 void (*lumToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
500 int width, uint32_t *pal);
501 /// Unscaled conversion of alpha plane to YV12 for horizontal scaler.
502 void (*alpToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
503 int width, uint32_t *pal);
504 /// Unscaled conversion of chroma planes to YV12 for horizontal scaler.
505 void (*chrToYV12)(uint8_t *dstU, uint8_t *dstV,
506 const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
507 int width, uint32_t *pal);
510 * Functions to read planar input, such as planar RGB, and convert
511 * internally to Y/UV/A.
514 void (*readLumPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
515 void (*readChrPlanar)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4],
516 int width, int32_t *rgb2yuv);
517 void (*readAlpPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
521 * Scale one horizontal line of input data using a bilinear filter
522 * to produce one line of output data. Compared to SwsContext->hScale(),
523 * please take note of the following caveats when using these:
524 * - Scaling is done using only 7bit instead of 14bit coefficients.
525 * - You can use no more than 5 input pixels to produce 4 output
526 * pixels. Therefore, this filter should not be used for downscaling
527 * by more than ~20% in width (because that equals more than 5/4th
528 * downscaling and thus more than 5 pixels input per 4 pixels output).
529 * - In general, bilinear filters create artifacts during downscaling
530 * (even when <20%), because one output pixel will span more than one
531 * input pixel, and thus some pixels will need edges of both neighbor
532 * pixels to interpolate the output pixel. Since you can use at most
533 * two input pixels per output pixel in bilinear scaling, this is
534 * impossible and thus downscaling by any size will create artifacts.
535 * To enable this type of scaling, set SWS_FLAG_FAST_BILINEAR
536 * in SwsContext->flags.
539 void (*hyscale_fast)(struct SwsContext *c,
540 int16_t *dst, int dstWidth,
541 const uint8_t *src, int srcW, int xInc);
542 void (*hcscale_fast)(struct SwsContext *c,
543 int16_t *dst1, int16_t *dst2, int dstWidth,
544 const uint8_t *src1, const uint8_t *src2,
549 * Scale one horizontal line of input data using a filter over the input
550 * lines, to produce one (differently sized) line of output data.
552 * @param dst pointer to destination buffer for horizontally scaled
553 * data. If the number of bits per component of one
554 * destination pixel (SwsContext->dstBpc) is <= 10, data
555 * will be 15bpc in 16bits (int16_t) width. Else (i.e.
556 * SwsContext->dstBpc == 16), data will be 19bpc in
557 * 32bits (int32_t) width.
558 * @param dstW width of destination image
559 * @param src pointer to source data to be scaled. If the number of
560 * bits per component of a source pixel (SwsContext->srcBpc)
561 * is 8, this is 8bpc in 8bits (uint8_t) width. Else
562 * (i.e. SwsContext->dstBpc > 8), this is native depth
563 * in 16bits (uint16_t) width. In other words, for 9-bit
564 * YUV input, this is 9bpc, for 10-bit YUV input, this is
565 * 10bpc, and for 16-bit RGB or YUV, this is 16bpc.
566 * @param filter filter coefficients to be used per output pixel for
567 * scaling. This contains 14bpp filtering coefficients.
568 * Guaranteed to contain dstW * filterSize entries.
569 * @param filterPos position of the first input pixel to be used for
570 * each output pixel during scaling. Guaranteed to
571 * contain dstW entries.
572 * @param filterSize the number of input coefficients to be used (and
573 * thus the number of input pixels to be used) for
574 * creating a single output pixel. Is aligned to 4
575 * (and input coefficients thus padded with zeroes)
576 * to simplify creating SIMD code.
579 void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
580 const uint8_t *src, const int16_t *filter,
581 const int32_t *filterPos, int filterSize);
582 void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
583 const uint8_t *src, const int16_t *filter,
584 const int32_t *filterPos, int filterSize);
587 /// Color range conversion function for luma plane if needed.
588 void (*lumConvertRange)(int16_t *dst, int width);
589 /// Color range conversion function for chroma planes if needed.
590 void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width);
592 int needs_hcscale; ///< Set if there are chroma planes to be converted.
594 //FIXME check init (where 0)
596 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c);
597 int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
598 int fullRange, int brightness,
599 int contrast, int saturation);
601 void ff_yuv2rgb_init_tables_altivec(SwsContext *c, const int inv_table[4],
602 int brightness, int contrast, int saturation);
603 void updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
604 int lastInLumBuf, int lastInChrBuf);
606 SwsFunc ff_yuv2rgb_init_mmx(SwsContext *c);
607 SwsFunc ff_yuv2rgb_init_vis(SwsContext *c);
608 SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c);
609 SwsFunc ff_yuv2rgb_get_func_ptr_bfin(SwsContext *c);
610 void ff_bfin_get_unscaled_swscale(SwsContext *c);
612 #if FF_API_SWS_FORMAT_NAME
614 * @deprecated Use av_get_pix_fmt_name() instead.
617 const char *sws_format_name(enum AVPixelFormat format);
620 static av_always_inline int is16BPS(enum AVPixelFormat pix_fmt)
622 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
624 return desc->comp[0].depth_minus1 == 15;
627 static av_always_inline int is9_OR_10BPS(enum AVPixelFormat pix_fmt)
629 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
631 return desc->comp[0].depth_minus1 >= 8 && desc->comp[0].depth_minus1 <= 13;
634 #define isNBPS(x) is9_OR_10BPS(x)
636 static av_always_inline int isBE(enum AVPixelFormat pix_fmt)
638 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
640 return desc->flags & AV_PIX_FMT_FLAG_BE;
643 static av_always_inline int isYUV(enum AVPixelFormat pix_fmt)
645 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
647 return !(desc->flags & AV_PIX_FMT_FLAG_RGB) && desc->nb_components >= 2;
650 static av_always_inline int isPlanarYUV(enum AVPixelFormat pix_fmt)
652 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
654 return ((desc->flags & AV_PIX_FMT_FLAG_PLANAR) && isYUV(pix_fmt));
657 static av_always_inline int isRGB(enum AVPixelFormat pix_fmt)
659 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
661 return (desc->flags & AV_PIX_FMT_FLAG_RGB);
666 (!(av_pix_fmt_desc_get(x)->flags & AV_PIX_FMT_FLAG_PAL) && \
667 av_pix_fmt_desc_get(x)->nb_components <= 2)
670 ((x) == AV_PIX_FMT_GRAY8 || \
671 (x) == AV_PIX_FMT_Y400A || \
672 (x) == AV_PIX_FMT_GRAY16BE || \
673 (x) == AV_PIX_FMT_GRAY16LE)
676 #define isRGBinInt(x) \
678 (x) == AV_PIX_FMT_RGB48BE || \
679 (x) == AV_PIX_FMT_RGB48LE || \
680 (x) == AV_PIX_FMT_RGBA64BE || \
681 (x) == AV_PIX_FMT_RGBA64LE || \
682 (x) == AV_PIX_FMT_RGB32 || \
683 (x) == AV_PIX_FMT_RGB32_1 || \
684 (x) == AV_PIX_FMT_RGB24 || \
685 (x) == AV_PIX_FMT_RGB565BE || \
686 (x) == AV_PIX_FMT_RGB565LE || \
687 (x) == AV_PIX_FMT_RGB555BE || \
688 (x) == AV_PIX_FMT_RGB555LE || \
689 (x) == AV_PIX_FMT_RGB444BE || \
690 (x) == AV_PIX_FMT_RGB444LE || \
691 (x) == AV_PIX_FMT_RGB8 || \
692 (x) == AV_PIX_FMT_RGB4 || \
693 (x) == AV_PIX_FMT_RGB4_BYTE || \
694 (x) == AV_PIX_FMT_MONOBLACK || \
695 (x) == AV_PIX_FMT_MONOWHITE \
697 #define isBGRinInt(x) \
699 (x) == AV_PIX_FMT_BGR48BE || \
700 (x) == AV_PIX_FMT_BGR48LE || \
701 (x) == AV_PIX_FMT_BGRA64BE || \
702 (x) == AV_PIX_FMT_BGRA64LE || \
703 (x) == AV_PIX_FMT_BGR32 || \
704 (x) == AV_PIX_FMT_BGR32_1 || \
705 (x) == AV_PIX_FMT_BGR24 || \
706 (x) == AV_PIX_FMT_BGR565BE || \
707 (x) == AV_PIX_FMT_BGR565LE || \
708 (x) == AV_PIX_FMT_BGR555BE || \
709 (x) == AV_PIX_FMT_BGR555LE || \
710 (x) == AV_PIX_FMT_BGR444BE || \
711 (x) == AV_PIX_FMT_BGR444LE || \
712 (x) == AV_PIX_FMT_BGR8 || \
713 (x) == AV_PIX_FMT_BGR4 || \
714 (x) == AV_PIX_FMT_BGR4_BYTE || \
715 (x) == AV_PIX_FMT_MONOBLACK || \
716 (x) == AV_PIX_FMT_MONOWHITE \
719 #define isRGBinBytes(x) ( \
720 (x) == AV_PIX_FMT_RGB48BE \
721 || (x) == AV_PIX_FMT_RGB48LE \
722 || (x) == AV_PIX_FMT_RGBA64BE \
723 || (x) == AV_PIX_FMT_RGBA64LE \
724 || (x) == AV_PIX_FMT_RGBA \
725 || (x) == AV_PIX_FMT_ARGB \
726 || (x) == AV_PIX_FMT_RGB24 \
728 #define isBGRinBytes(x) ( \
729 (x) == AV_PIX_FMT_BGR48BE \
730 || (x) == AV_PIX_FMT_BGR48LE \
731 || (x) == AV_PIX_FMT_BGRA64BE \
732 || (x) == AV_PIX_FMT_BGRA64LE \
733 || (x) == AV_PIX_FMT_BGRA \
734 || (x) == AV_PIX_FMT_ABGR \
735 || (x) == AV_PIX_FMT_BGR24 \
738 #define isAnyRGB(x) \
745 static av_always_inline int isALPHA(enum AVPixelFormat pix_fmt)
747 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
749 if (pix_fmt == AV_PIX_FMT_PAL8)
751 return desc->flags & AV_PIX_FMT_FLAG_ALPHA;
755 #define isPacked(x) ( \
756 (x)==AV_PIX_FMT_PAL8 \
757 || (x)==AV_PIX_FMT_YUYV422 \
758 || (x)==AV_PIX_FMT_UYVY422 \
759 || (x)==AV_PIX_FMT_Y400A \
764 static av_always_inline int isPacked(enum AVPixelFormat pix_fmt)
766 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
768 return ((desc->nb_components >= 2 && !(desc->flags & AV_PIX_FMT_FLAG_PLANAR)) ||
769 pix_fmt == AV_PIX_FMT_PAL8);
773 static av_always_inline int isPlanar(enum AVPixelFormat pix_fmt)
775 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
777 return (desc->nb_components >= 2 && (desc->flags & AV_PIX_FMT_FLAG_PLANAR));
780 static av_always_inline int isPackedRGB(enum AVPixelFormat pix_fmt)
782 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
784 return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) == AV_PIX_FMT_FLAG_RGB);
787 static av_always_inline int isPlanarRGB(enum AVPixelFormat pix_fmt)
789 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
791 return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) ==
792 (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB));
795 static av_always_inline int usePal(enum AVPixelFormat pix_fmt)
797 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
799 return (desc->flags & AV_PIX_FMT_FLAG_PAL) || (desc->flags & AV_PIX_FMT_FLAG_PSEUDOPAL);
802 extern const uint64_t ff_dither4[2];
803 extern const uint64_t ff_dither8[2];
804 extern const uint8_t dithers[8][8][8];
805 extern const uint16_t dither_scale[15][16];
808 extern const AVClass sws_context_class;
811 * Set c->swScale to an unscaled converter if one exists for the specific
812 * source and destination formats, bit depths, flags, etc.
814 void ff_get_unscaled_swscale(SwsContext *c);
816 void ff_swscale_get_unscaled_altivec(SwsContext *c);
819 * Return function pointer to fastest main scaler path function depending
820 * on architecture and available optimizations.
822 SwsFunc ff_getSwsFunc(SwsContext *c);
824 void ff_sws_init_input_funcs(SwsContext *c);
825 void ff_sws_init_output_funcs(SwsContext *c,
826 yuv2planar1_fn *yuv2plane1,
827 yuv2planarX_fn *yuv2planeX,
828 yuv2interleavedX_fn *yuv2nv12cX,
829 yuv2packed1_fn *yuv2packed1,
830 yuv2packed2_fn *yuv2packed2,
831 yuv2packedX_fn *yuv2packedX,
832 yuv2anyX_fn *yuv2anyX);
833 void ff_sws_init_swScale_altivec(SwsContext *c);
834 void ff_sws_init_swScale_mmx(SwsContext *c);
836 static inline void fillPlane16(uint8_t *plane, int stride, int width, int height, int y,
837 int alpha, int bits, const int big_endian)
840 uint8_t *ptr = plane + stride * y;
841 int v = alpha ? 0xFFFF>>(15-bits) : (1<<bits);
842 for (i = 0; i < height; i++) {
843 #define FILL(wfunc) \
844 for (j = 0; j < width; j++) {\
856 #endif /* SWSCALE_SWSCALE_INTERNAL_H */