2 * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 #ifndef SWSCALE_SWSCALE_INTERNAL_H
22 #define SWSCALE_SWSCALE_INTERNAL_H
32 #include "libavutil/avassert.h"
33 #include "libavutil/avutil.h"
34 #include "libavutil/common.h"
35 #include "libavutil/intreadwrite.h"
36 #include "libavutil/log.h"
37 #include "libavutil/pixfmt.h"
38 #include "libavutil/pixdesc.h"
40 #define STR(s) AV_TOSTRING(s) // AV_STRINGIFY is too long
42 #define YUVRGB_TABLE_HEADROOM 256
43 #define YUVRGB_TABLE_LUMA_HEADROOM 0
45 #define MAX_FILTER_SIZE SWS_MAX_FILTER_SIZE
50 #define ALT32_CORR (-1)
65 #define RETCODE_USE_CASCADE -12345
69 typedef enum SwsDither {
79 typedef enum SwsAlphaBlend {
80 SWS_ALPHA_BLEND_NONE = 0,
81 SWS_ALPHA_BLEND_UNIFORM,
82 SWS_ALPHA_BLEND_CHECKERBOARD,
86 typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
87 int srcStride[], int srcSliceY, int srcSliceH,
88 uint8_t *dst[], int dstStride[]);
91 * Write one line of horizontally scaled data to planar output
92 * without any additional vertical scaling (or point-scaling).
94 * @param src scaled source data, 15bit for 8-10bit output,
95 * 19-bit for 16bit output (in int32_t)
96 * @param dest pointer to the output plane. For >8bit
97 * output, this is in uint16_t
98 * @param dstW width of destination in pixels
99 * @param dither ordered dither array of type int16_t and size 8
100 * @param offset Dither offset
102 typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
103 const uint8_t *dither, int offset);
106 * Write one line of horizontally scaled data to planar output
107 * with multi-point vertical scaling between input pixels.
109 * @param filter vertical luma/alpha scaling coefficients, 12bit [0,4096]
110 * @param src scaled luma (Y) or alpha (A) source data, 15bit for 8-10bit output,
111 * 19-bit for 16bit output (in int32_t)
112 * @param filterSize number of vertical input lines to scale
113 * @param dest pointer to output plane. For >8bit
114 * output, this is in uint16_t
115 * @param dstW width of destination pixels
116 * @param offset Dither offset
118 typedef void (*yuv2planarX_fn)(const int16_t *filter, int filterSize,
119 const int16_t **src, uint8_t *dest, int dstW,
120 const uint8_t *dither, int offset);
123 * Write one line of horizontally scaled chroma to interleaved output
124 * with multi-point vertical scaling between input pixels.
126 * @param c SWS scaling context
127 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
128 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
129 * 19-bit for 16bit output (in int32_t)
130 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
131 * 19-bit for 16bit output (in int32_t)
132 * @param chrFilterSize number of vertical chroma input lines to scale
133 * @param dest pointer to the output plane. For >8bit
134 * output, this is in uint16_t
135 * @param dstW width of chroma planes
137 typedef void (*yuv2interleavedX_fn)(struct SwsContext *c,
138 const int16_t *chrFilter,
140 const int16_t **chrUSrc,
141 const int16_t **chrVSrc,
142 uint8_t *dest, int dstW);
145 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
146 * output without any additional vertical scaling (or point-scaling). Note
147 * that this function may do chroma scaling, see the "uvalpha" argument.
149 * @param c SWS scaling context
150 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
151 * 19-bit for 16bit output (in int32_t)
152 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
153 * 19-bit for 16bit output (in int32_t)
154 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
155 * 19-bit for 16bit output (in int32_t)
156 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
157 * 19-bit for 16bit output (in int32_t)
158 * @param dest pointer to the output plane. For 16bit output, this is
160 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
161 * to write into dest[]
162 * @param uvalpha chroma scaling coefficient for the second line of chroma
163 * pixels, either 2048 or 0. If 0, one chroma input is used
164 * for 2 output pixels (or if the SWS_FLAG_FULL_CHR_INT flag
165 * is set, it generates 1 output pixel). If 2048, two chroma
166 * input pixels should be averaged for 2 output pixels (this
167 * only happens if SWS_FLAG_FULL_CHR_INT is not set)
168 * @param y vertical line number for this output. This does not need
169 * to be used to calculate the offset in the destination,
170 * but can be used to generate comfort noise using dithering
171 * for some output formats.
173 typedef void (*yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc,
174 const int16_t *chrUSrc[2],
175 const int16_t *chrVSrc[2],
176 const int16_t *alpSrc, uint8_t *dest,
177 int dstW, int uvalpha, int y);
179 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
180 * output by doing bilinear scaling between two input lines.
182 * @param c SWS scaling context
183 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
184 * 19-bit for 16bit output (in int32_t)
185 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
186 * 19-bit for 16bit output (in int32_t)
187 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
188 * 19-bit for 16bit output (in int32_t)
189 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
190 * 19-bit for 16bit output (in int32_t)
191 * @param dest pointer to the output plane. For 16bit output, this is
193 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
194 * to write into dest[]
195 * @param yalpha luma/alpha scaling coefficients for the second input line.
196 * The first line's coefficients can be calculated by using
198 * @param uvalpha chroma scaling coefficient for the second input line. The
199 * first line's coefficients can be calculated by using
201 * @param y vertical line number for this output. This does not need
202 * to be used to calculate the offset in the destination,
203 * but can be used to generate comfort noise using dithering
204 * for some output formats.
206 typedef void (*yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2],
207 const int16_t *chrUSrc[2],
208 const int16_t *chrVSrc[2],
209 const int16_t *alpSrc[2],
211 int dstW, int yalpha, int uvalpha, int y);
213 * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
214 * output by doing multi-point vertical scaling between input pixels.
216 * @param c SWS scaling context
217 * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
218 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
219 * 19-bit for 16bit output (in int32_t)
220 * @param lumFilterSize number of vertical luma/alpha input lines to scale
221 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
222 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
223 * 19-bit for 16bit output (in int32_t)
224 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
225 * 19-bit for 16bit output (in int32_t)
226 * @param chrFilterSize number of vertical chroma input lines to scale
227 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
228 * 19-bit for 16bit output (in int32_t)
229 * @param dest pointer to the output plane. For 16bit output, this is
231 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
232 * to write into dest[]
233 * @param y vertical line number for this output. This does not need
234 * to be used to calculate the offset in the destination,
235 * but can be used to generate comfort noise using dithering
236 * or some output formats.
238 typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter,
239 const int16_t **lumSrc, int lumFilterSize,
240 const int16_t *chrFilter,
241 const int16_t **chrUSrc,
242 const int16_t **chrVSrc, int chrFilterSize,
243 const int16_t **alpSrc, uint8_t *dest,
247 * Write one line of horizontally scaled Y/U/V/A to YUV/RGB
248 * output by doing multi-point vertical scaling between input pixels.
250 * @param c SWS scaling context
251 * @param lumFilter vertical luma/alpha scaling coefficients, 12bit [0,4096]
252 * @param lumSrc scaled luma (Y) source data, 15bit for 8-10bit output,
253 * 19-bit for 16bit output (in int32_t)
254 * @param lumFilterSize number of vertical luma/alpha input lines to scale
255 * @param chrFilter vertical chroma scaling coefficients, 12bit [0,4096]
256 * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
257 * 19-bit for 16bit output (in int32_t)
258 * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
259 * 19-bit for 16bit output (in int32_t)
260 * @param chrFilterSize number of vertical chroma input lines to scale
261 * @param alpSrc scaled alpha (A) source data, 15bit for 8-10bit output,
262 * 19-bit for 16bit output (in int32_t)
263 * @param dest pointer to the output planes. For 16bit output, this is
265 * @param dstW width of lumSrc and alpSrc in pixels, number of pixels
266 * to write into dest[]
267 * @param y vertical line number for this output. This does not need
268 * to be used to calculate the offset in the destination,
269 * but can be used to generate comfort noise using dithering
270 * or some output formats.
272 typedef void (*yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter,
273 const int16_t **lumSrc, int lumFilterSize,
274 const int16_t *chrFilter,
275 const int16_t **chrUSrc,
276 const int16_t **chrVSrc, int chrFilterSize,
277 const int16_t **alpSrc, uint8_t **dest,
281 struct SwsFilterDescriptor;
283 /* This struct should be aligned on at least a 32-byte boundary. */
284 typedef struct SwsContext {
286 * info on struct for av_log
288 const AVClass *av_class;
291 * Note that src, dst, srcStride, dstStride will be copied in the
292 * sws_scale() wrapper so they can be freely modified here.
295 int srcW; ///< Width of source luma/alpha planes.
296 int srcH; ///< Height of source luma/alpha planes.
297 int dstH; ///< Height of destination luma/alpha planes.
298 int chrSrcW; ///< Width of source chroma planes.
299 int chrSrcH; ///< Height of source chroma planes.
300 int chrDstW; ///< Width of destination chroma planes.
301 int chrDstH; ///< Height of destination chroma planes.
302 int lumXInc, chrXInc;
303 int lumYInc, chrYInc;
304 enum AVPixelFormat dstFormat; ///< Destination pixel format.
305 enum AVPixelFormat srcFormat; ///< Source pixel format.
306 int dstFormatBpp; ///< Number of bits per pixel of the destination pixel format.
307 int srcFormatBpp; ///< Number of bits per pixel of the source pixel format.
309 int chrSrcHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source image.
310 int chrSrcVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in source image.
311 int chrDstHSubSample; ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination image.
312 int chrDstVSubSample; ///< Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in destination image.
313 int vChrDrop; ///< Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user.
314 int sliceDir; ///< Direction that slices are fed to the scaler (1 = top-to-bottom, -1 = bottom-to-top).
315 double param[2]; ///< Input parameters for scaling algorithms that need them.
317 /* The cascaded_* fields allow spliting a scaler task into multiple
318 * sequential steps, this is for example used to limit the maximum
319 * downscaling factor that needs to be supported in one scaler.
321 struct SwsContext *cascaded_context[3];
322 int cascaded_tmpStride[4];
323 uint8_t *cascaded_tmp[4];
324 int cascaded1_tmpStride[4];
325 uint8_t *cascaded1_tmp[4];
326 int cascaded_mainindex;
330 int is_internal_gamma;
337 struct SwsSlice *slice;
338 struct SwsFilterDescriptor *desc;
340 uint32_t pal_yuv[256];
341 uint32_t pal_rgb[256];
344 * @name Scaled horizontal lines ring buffer.
345 * The horizontal scaler keeps just enough scaled lines in a ring buffer
346 * so they may be passed to the vertical scaler. The pointers to the
347 * allocated buffers for each line are duplicated in sequence in the ring
348 * buffer to simplify indexing and avoid wrapping around between lines
349 * inside the vertical scaler code. The wrapping is done before the
350 * vertical scaler is called.
353 int16_t **lumPixBuf; ///< Ring buffer for scaled horizontal luma plane lines to be fed to the vertical scaler.
354 int16_t **chrUPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
355 int16_t **chrVPixBuf; ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
356 int16_t **alpPixBuf; ///< Ring buffer for scaled horizontal alpha plane lines to be fed to the vertical scaler.
357 int vLumBufSize; ///< Number of vertical luma/alpha lines allocated in the ring buffer.
358 int vChrBufSize; ///< Number of vertical chroma lines allocated in the ring buffer.
359 int lastInLumBuf; ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
360 int lastInChrBuf; ///< Last scaled horizontal chroma line from source in the ring buffer.
361 int lumBufIndex; ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
362 int chrBufIndex; ///< Index in ring buffer of the last scaled horizontal chroma line from source.
365 uint8_t *formatConvBuffer;
368 * @name Horizontal and vertical filters.
369 * To better understand the following fields, here is a pseudo-code of
370 * their usage in filtering a horizontal line:
372 * for (i = 0; i < width; i++) {
374 * for (j = 0; j < filterSize; j++)
375 * dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
376 * dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
381 int16_t *hLumFilter; ///< Array of horizontal filter coefficients for luma/alpha planes.
382 int16_t *hChrFilter; ///< Array of horizontal filter coefficients for chroma planes.
383 int16_t *vLumFilter; ///< Array of vertical filter coefficients for luma/alpha planes.
384 int16_t *vChrFilter; ///< Array of vertical filter coefficients for chroma planes.
385 int32_t *hLumFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
386 int32_t *hChrFilterPos; ///< Array of horizontal filter starting positions for each dst[i] for chroma planes.
387 int32_t *vLumFilterPos; ///< Array of vertical filter starting positions for each dst[i] for luma/alpha planes.
388 int32_t *vChrFilterPos; ///< Array of vertical filter starting positions for each dst[i] for chroma planes.
389 int hLumFilterSize; ///< Horizontal filter size for luma/alpha pixels.
390 int hChrFilterSize; ///< Horizontal filter size for chroma pixels.
391 int vLumFilterSize; ///< Vertical filter size for luma/alpha pixels.
392 int vChrFilterSize; ///< Vertical filter size for chroma pixels.
395 int lumMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
396 int chrMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
397 uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
398 uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
402 int dstY; ///< Last destination vertical line output from last slice.
403 int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
404 void *yuvTable; // pointer to the yuv->rgb table start so it can be freed()
405 // alignment ensures the offset can be added in a single
406 // instruction on e.g. ARM
407 DECLARE_ALIGNED(16, int, table_gV)[256 + 2*YUVRGB_TABLE_HEADROOM];
408 uint8_t *table_rV[256 + 2*YUVRGB_TABLE_HEADROOM];
409 uint8_t *table_gU[256 + 2*YUVRGB_TABLE_HEADROOM];
410 uint8_t *table_bU[256 + 2*YUVRGB_TABLE_HEADROOM];
411 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
421 #define RGB2YUV_SHIFT 15
423 int *dither_error[4];
426 int contrast, brightness, saturation; // for sws_getColorspaceDetails
427 int srcColorspaceTable[4];
428 int dstColorspaceTable[4];
429 int srcRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (source image).
430 int dstRange; ///< 0 = MPG YUV range, 1 = JPG YUV range (destination image).
439 int yuv2rgb_y_offset;
441 int yuv2rgb_v2r_coeff;
442 int yuv2rgb_v2g_coeff;
443 int yuv2rgb_u2g_coeff;
444 int yuv2rgb_u2b_coeff;
446 #define RED_DITHER "0*8"
447 #define GREEN_DITHER "1*8"
448 #define BLUE_DITHER "2*8"
449 #define Y_COEFF "3*8"
450 #define VR_COEFF "4*8"
451 #define UB_COEFF "5*8"
452 #define VG_COEFF "6*8"
453 #define UG_COEFF "7*8"
454 #define Y_OFFSET "8*8"
455 #define U_OFFSET "9*8"
456 #define V_OFFSET "10*8"
457 #define LUM_MMX_FILTER_OFFSET "11*8"
458 #define CHR_MMX_FILTER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)
459 #define DSTW_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2"
460 #define ESP_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+8"
461 #define VROUNDER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+16"
462 #define U_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+24"
463 #define V_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+32"
464 #define Y_TEMP "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+40"
465 #define ALP_MMX_FILTER_OFFSET "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*2+48"
466 #define UV_OFF_PX "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+48"
467 #define UV_OFF_BYTE "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+56"
468 #define DITHER16 "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+64"
469 #define DITHER32 "11*8+4*4*"AV_STRINGIFY(MAX_FILTER_SIZE)"*3+80"
470 #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
472 DECLARE_ALIGNED(8, uint64_t, redDither);
473 DECLARE_ALIGNED(8, uint64_t, greenDither);
474 DECLARE_ALIGNED(8, uint64_t, blueDither);
476 DECLARE_ALIGNED(8, uint64_t, yCoeff);
477 DECLARE_ALIGNED(8, uint64_t, vrCoeff);
478 DECLARE_ALIGNED(8, uint64_t, ubCoeff);
479 DECLARE_ALIGNED(8, uint64_t, vgCoeff);
480 DECLARE_ALIGNED(8, uint64_t, ugCoeff);
481 DECLARE_ALIGNED(8, uint64_t, yOffset);
482 DECLARE_ALIGNED(8, uint64_t, uOffset);
483 DECLARE_ALIGNED(8, uint64_t, vOffset);
484 int32_t lumMmxFilter[4 * MAX_FILTER_SIZE];
485 int32_t chrMmxFilter[4 * MAX_FILTER_SIZE];
486 int dstW; ///< Width of destination luma/alpha planes.
487 DECLARE_ALIGNED(8, uint64_t, esp);
488 DECLARE_ALIGNED(8, uint64_t, vRounder);
489 DECLARE_ALIGNED(8, uint64_t, u_temp);
490 DECLARE_ALIGNED(8, uint64_t, v_temp);
491 DECLARE_ALIGNED(8, uint64_t, y_temp);
492 int32_t alpMmxFilter[4 * MAX_FILTER_SIZE];
493 // alignment of these values is not necessary, but merely here
494 // to maintain the same offset across x8632 and x86-64. Once we
495 // use proper offset macros in the asm, they can be removed.
496 DECLARE_ALIGNED(8, ptrdiff_t, uv_off); ///< offset (in pixels) between u and v planes
497 DECLARE_ALIGNED(8, ptrdiff_t, uv_offx2); ///< offset (in bytes) between u and v planes
498 DECLARE_ALIGNED(8, uint16_t, dither16)[8];
499 DECLARE_ALIGNED(8, uint32_t, dither32)[8];
501 const uint8_t *chrDither8, *lumDither8;
504 vector signed short CY;
505 vector signed short CRV;
506 vector signed short CBU;
507 vector signed short CGU;
508 vector signed short CGV;
509 vector signed short OY;
510 vector unsigned short CSHIFT;
511 vector signed short *vYCoeffsBank, *vCCoeffsBank;
516 /* pre defined color-spaces gamma */
517 #define XYZ_GAMMA (2.6f)
518 #define RGB_GAMMA (2.2f)
521 int16_t *xyzgammainv;
522 int16_t *rgbgammainv;
523 int16_t xyz2rgb_matrix[3][4];
524 int16_t rgb2xyz_matrix[3][4];
526 /* function pointers for swscale() */
527 yuv2planar1_fn yuv2plane1;
528 yuv2planarX_fn yuv2planeX;
529 yuv2interleavedX_fn yuv2nv12cX;
530 yuv2packed1_fn yuv2packed1;
531 yuv2packed2_fn yuv2packed2;
532 yuv2packedX_fn yuv2packedX;
533 yuv2anyX_fn yuv2anyX;
535 /// Unscaled conversion of luma plane to YV12 for horizontal scaler.
536 void (*lumToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
537 int width, uint32_t *pal);
538 /// Unscaled conversion of alpha plane to YV12 for horizontal scaler.
539 void (*alpToYV12)(uint8_t *dst, const uint8_t *src, const uint8_t *src2, const uint8_t *src3,
540 int width, uint32_t *pal);
541 /// Unscaled conversion of chroma planes to YV12 for horizontal scaler.
542 void (*chrToYV12)(uint8_t *dstU, uint8_t *dstV,
543 const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
544 int width, uint32_t *pal);
547 * Functions to read planar input, such as planar RGB, and convert
548 * internally to Y/UV/A.
551 void (*readLumPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
552 void (*readChrPlanar)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4],
553 int width, int32_t *rgb2yuv);
554 void (*readAlpPlanar)(uint8_t *dst, const uint8_t *src[4], int width, int32_t *rgb2yuv);
558 * Scale one horizontal line of input data using a bilinear filter
559 * to produce one line of output data. Compared to SwsContext->hScale(),
560 * please take note of the following caveats when using these:
561 * - Scaling is done using only 7bit instead of 14bit coefficients.
562 * - You can use no more than 5 input pixels to produce 4 output
563 * pixels. Therefore, this filter should not be used for downscaling
564 * by more than ~20% in width (because that equals more than 5/4th
565 * downscaling and thus more than 5 pixels input per 4 pixels output).
566 * - In general, bilinear filters create artifacts during downscaling
567 * (even when <20%), because one output pixel will span more than one
568 * input pixel, and thus some pixels will need edges of both neighbor
569 * pixels to interpolate the output pixel. Since you can use at most
570 * two input pixels per output pixel in bilinear scaling, this is
571 * impossible and thus downscaling by any size will create artifacts.
572 * To enable this type of scaling, set SWS_FLAG_FAST_BILINEAR
573 * in SwsContext->flags.
576 void (*hyscale_fast)(struct SwsContext *c,
577 int16_t *dst, int dstWidth,
578 const uint8_t *src, int srcW, int xInc);
579 void (*hcscale_fast)(struct SwsContext *c,
580 int16_t *dst1, int16_t *dst2, int dstWidth,
581 const uint8_t *src1, const uint8_t *src2,
586 * Scale one horizontal line of input data using a filter over the input
587 * lines, to produce one (differently sized) line of output data.
589 * @param dst pointer to destination buffer for horizontally scaled
590 * data. If the number of bits per component of one
591 * destination pixel (SwsContext->dstBpc) is <= 10, data
592 * will be 15bpc in 16bits (int16_t) width. Else (i.e.
593 * SwsContext->dstBpc == 16), data will be 19bpc in
594 * 32bits (int32_t) width.
595 * @param dstW width of destination image
596 * @param src pointer to source data to be scaled. If the number of
597 * bits per component of a source pixel (SwsContext->srcBpc)
598 * is 8, this is 8bpc in 8bits (uint8_t) width. Else
599 * (i.e. SwsContext->dstBpc > 8), this is native depth
600 * in 16bits (uint16_t) width. In other words, for 9-bit
601 * YUV input, this is 9bpc, for 10-bit YUV input, this is
602 * 10bpc, and for 16-bit RGB or YUV, this is 16bpc.
603 * @param filter filter coefficients to be used per output pixel for
604 * scaling. This contains 14bpp filtering coefficients.
605 * Guaranteed to contain dstW * filterSize entries.
606 * @param filterPos position of the first input pixel to be used for
607 * each output pixel during scaling. Guaranteed to
608 * contain dstW entries.
609 * @param filterSize the number of input coefficients to be used (and
610 * thus the number of input pixels to be used) for
611 * creating a single output pixel. Is aligned to 4
612 * (and input coefficients thus padded with zeroes)
613 * to simplify creating SIMD code.
616 void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
617 const uint8_t *src, const int16_t *filter,
618 const int32_t *filterPos, int filterSize);
619 void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
620 const uint8_t *src, const int16_t *filter,
621 const int32_t *filterPos, int filterSize);
624 /// Color range conversion function for luma plane if needed.
625 void (*lumConvertRange)(int16_t *dst, int width);
626 /// Color range conversion function for chroma planes if needed.
627 void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width);
629 int needs_hcscale; ///< Set if there are chroma planes to be converted.
633 SwsAlphaBlend alphablend;
635 //FIXME check init (where 0)
637 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c);
638 int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
639 int fullRange, int brightness,
640 int contrast, int saturation);
641 void ff_yuv2rgb_init_tables_ppc(SwsContext *c, const int inv_table[4],
642 int brightness, int contrast, int saturation);
644 void ff_updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
645 int lastInLumBuf, int lastInChrBuf);
647 av_cold void ff_sws_init_range_convert(SwsContext *c);
649 SwsFunc ff_yuv2rgb_init_x86(SwsContext *c);
650 SwsFunc ff_yuv2rgb_init_ppc(SwsContext *c);
652 static av_always_inline int is16BPS(enum AVPixelFormat pix_fmt)
654 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
656 return desc->comp[0].depth == 16;
659 static av_always_inline int is9_OR_10BPS(enum AVPixelFormat pix_fmt)
661 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
663 return desc->comp[0].depth >= 9 && desc->comp[0].depth <= 14;
666 #define isNBPS(x) is9_OR_10BPS(x)
668 static av_always_inline int isBE(enum AVPixelFormat pix_fmt)
670 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
672 return desc->flags & AV_PIX_FMT_FLAG_BE;
675 static av_always_inline int isYUV(enum AVPixelFormat pix_fmt)
677 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
679 return !(desc->flags & AV_PIX_FMT_FLAG_RGB) && desc->nb_components >= 2;
682 static av_always_inline int isPlanarYUV(enum AVPixelFormat pix_fmt)
684 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
686 return ((desc->flags & AV_PIX_FMT_FLAG_PLANAR) && isYUV(pix_fmt));
689 static av_always_inline int isRGB(enum AVPixelFormat pix_fmt)
691 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
693 return (desc->flags & AV_PIX_FMT_FLAG_RGB);
698 (!(av_pix_fmt_desc_get(x)->flags & AV_PIX_FMT_FLAG_PAL) && \
699 av_pix_fmt_desc_get(x)->nb_components <= 2)
702 ((x) == AV_PIX_FMT_GRAY8 || \
703 (x) == AV_PIX_FMT_YA8 || \
704 (x) == AV_PIX_FMT_GRAY16BE || \
705 (x) == AV_PIX_FMT_GRAY16LE || \
706 (x) == AV_PIX_FMT_YA16BE || \
707 (x) == AV_PIX_FMT_YA16LE)
710 #define isRGBinInt(x) \
712 (x) == AV_PIX_FMT_RGB48BE || \
713 (x) == AV_PIX_FMT_RGB48LE || \
714 (x) == AV_PIX_FMT_RGB32 || \
715 (x) == AV_PIX_FMT_RGB32_1 || \
716 (x) == AV_PIX_FMT_RGB24 || \
717 (x) == AV_PIX_FMT_RGB565BE || \
718 (x) == AV_PIX_FMT_RGB565LE || \
719 (x) == AV_PIX_FMT_RGB555BE || \
720 (x) == AV_PIX_FMT_RGB555LE || \
721 (x) == AV_PIX_FMT_RGB444BE || \
722 (x) == AV_PIX_FMT_RGB444LE || \
723 (x) == AV_PIX_FMT_RGB8 || \
724 (x) == AV_PIX_FMT_RGB4 || \
725 (x) == AV_PIX_FMT_RGB4_BYTE || \
726 (x) == AV_PIX_FMT_RGBA64BE || \
727 (x) == AV_PIX_FMT_RGBA64LE || \
728 (x) == AV_PIX_FMT_MONOBLACK || \
729 (x) == AV_PIX_FMT_MONOWHITE \
731 #define isBGRinInt(x) \
733 (x) == AV_PIX_FMT_BGR48BE || \
734 (x) == AV_PIX_FMT_BGR48LE || \
735 (x) == AV_PIX_FMT_BGR32 || \
736 (x) == AV_PIX_FMT_BGR32_1 || \
737 (x) == AV_PIX_FMT_BGR24 || \
738 (x) == AV_PIX_FMT_BGR565BE || \
739 (x) == AV_PIX_FMT_BGR565LE || \
740 (x) == AV_PIX_FMT_BGR555BE || \
741 (x) == AV_PIX_FMT_BGR555LE || \
742 (x) == AV_PIX_FMT_BGR444BE || \
743 (x) == AV_PIX_FMT_BGR444LE || \
744 (x) == AV_PIX_FMT_BGR8 || \
745 (x) == AV_PIX_FMT_BGR4 || \
746 (x) == AV_PIX_FMT_BGR4_BYTE || \
747 (x) == AV_PIX_FMT_BGRA64BE || \
748 (x) == AV_PIX_FMT_BGRA64LE || \
749 (x) == AV_PIX_FMT_MONOBLACK || \
750 (x) == AV_PIX_FMT_MONOWHITE \
753 #define isRGBinBytes(x) ( \
754 (x) == AV_PIX_FMT_RGB48BE \
755 || (x) == AV_PIX_FMT_RGB48LE \
756 || (x) == AV_PIX_FMT_RGBA64BE \
757 || (x) == AV_PIX_FMT_RGBA64LE \
758 || (x) == AV_PIX_FMT_RGBA \
759 || (x) == AV_PIX_FMT_ARGB \
760 || (x) == AV_PIX_FMT_RGB24 \
762 #define isBGRinBytes(x) ( \
763 (x) == AV_PIX_FMT_BGR48BE \
764 || (x) == AV_PIX_FMT_BGR48LE \
765 || (x) == AV_PIX_FMT_BGRA64BE \
766 || (x) == AV_PIX_FMT_BGRA64LE \
767 || (x) == AV_PIX_FMT_BGRA \
768 || (x) == AV_PIX_FMT_ABGR \
769 || (x) == AV_PIX_FMT_BGR24 \
772 #define isBayer(x) ( \
773 (x)==AV_PIX_FMT_BAYER_BGGR8 \
774 || (x)==AV_PIX_FMT_BAYER_BGGR16LE \
775 || (x)==AV_PIX_FMT_BAYER_BGGR16BE \
776 || (x)==AV_PIX_FMT_BAYER_RGGB8 \
777 || (x)==AV_PIX_FMT_BAYER_RGGB16LE \
778 || (x)==AV_PIX_FMT_BAYER_RGGB16BE \
779 || (x)==AV_PIX_FMT_BAYER_GBRG8 \
780 || (x)==AV_PIX_FMT_BAYER_GBRG16LE \
781 || (x)==AV_PIX_FMT_BAYER_GBRG16BE \
782 || (x)==AV_PIX_FMT_BAYER_GRBG8 \
783 || (x)==AV_PIX_FMT_BAYER_GRBG16LE \
784 || (x)==AV_PIX_FMT_BAYER_GRBG16BE \
787 #define isAnyRGB(x) \
795 static av_always_inline int isALPHA(enum AVPixelFormat pix_fmt)
797 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
799 if (pix_fmt == AV_PIX_FMT_PAL8)
801 return desc->flags & AV_PIX_FMT_FLAG_ALPHA;
805 #define isPacked(x) ( \
806 (x)==AV_PIX_FMT_PAL8 \
807 || (x)==AV_PIX_FMT_YUYV422 \
808 || (x)==AV_PIX_FMT_YVYU422 \
809 || (x)==AV_PIX_FMT_UYVY422 \
810 || (x)==AV_PIX_FMT_YA8 \
811 || (x)==AV_PIX_FMT_YA16LE \
812 || (x)==AV_PIX_FMT_YA16BE \
813 || (x)==AV_PIX_FMT_AYUV64LE \
814 || (x)==AV_PIX_FMT_AYUV64BE \
819 static av_always_inline int isPacked(enum AVPixelFormat pix_fmt)
821 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
823 return ((desc->nb_components >= 2 && !(desc->flags & AV_PIX_FMT_FLAG_PLANAR)) ||
824 pix_fmt == AV_PIX_FMT_PAL8);
828 static av_always_inline int isPlanar(enum AVPixelFormat pix_fmt)
830 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
832 return (desc->nb_components >= 2 && (desc->flags & AV_PIX_FMT_FLAG_PLANAR));
835 static av_always_inline int isPackedRGB(enum AVPixelFormat pix_fmt)
837 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
839 return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) == AV_PIX_FMT_FLAG_RGB);
842 static av_always_inline int isPlanarRGB(enum AVPixelFormat pix_fmt)
844 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
846 return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) ==
847 (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB));
850 static av_always_inline int usePal(enum AVPixelFormat pix_fmt)
852 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
854 return (desc->flags & AV_PIX_FMT_FLAG_PAL) || (desc->flags & AV_PIX_FMT_FLAG_PSEUDOPAL);
857 extern const uint64_t ff_dither4[2];
858 extern const uint64_t ff_dither8[2];
860 extern const uint8_t ff_dither_2x2_4[3][8];
861 extern const uint8_t ff_dither_2x2_8[3][8];
862 extern const uint8_t ff_dither_4x4_16[5][8];
863 extern const uint8_t ff_dither_8x8_32[9][8];
864 extern const uint8_t ff_dither_8x8_73[9][8];
865 extern const uint8_t ff_dither_8x8_128[9][8];
866 extern const uint8_t ff_dither_8x8_220[9][8];
868 extern const int32_t ff_yuv2rgb_coeffs[8][4];
870 extern const AVClass ff_sws_context_class;
873 * Set c->swscale to an unscaled converter if one exists for the specific
874 * source and destination formats, bit depths, flags, etc.
876 void ff_get_unscaled_swscale(SwsContext *c);
877 void ff_get_unscaled_swscale_ppc(SwsContext *c);
878 void ff_get_unscaled_swscale_arm(SwsContext *c);
881 * Return function pointer to fastest main scaler path function depending
882 * on architecture and available optimizations.
884 SwsFunc ff_getSwsFunc(SwsContext *c);
886 void ff_sws_init_input_funcs(SwsContext *c);
887 void ff_sws_init_output_funcs(SwsContext *c,
888 yuv2planar1_fn *yuv2plane1,
889 yuv2planarX_fn *yuv2planeX,
890 yuv2interleavedX_fn *yuv2nv12cX,
891 yuv2packed1_fn *yuv2packed1,
892 yuv2packed2_fn *yuv2packed2,
893 yuv2packedX_fn *yuv2packedX,
894 yuv2anyX_fn *yuv2anyX);
895 void ff_sws_init_swscale_ppc(SwsContext *c);
896 void ff_sws_init_swscale_x86(SwsContext *c);
898 void ff_hyscale_fast_c(SwsContext *c, int16_t *dst, int dstWidth,
899 const uint8_t *src, int srcW, int xInc);
900 void ff_hcscale_fast_c(SwsContext *c, int16_t *dst1, int16_t *dst2,
901 int dstWidth, const uint8_t *src1,
902 const uint8_t *src2, int srcW, int xInc);
903 int ff_init_hscaler_mmxext(int dstW, int xInc, uint8_t *filterCode,
904 int16_t *filter, int32_t *filterPos,
906 void ff_hyscale_fast_mmxext(SwsContext *c, int16_t *dst,
907 int dstWidth, const uint8_t *src,
909 void ff_hcscale_fast_mmxext(SwsContext *c, int16_t *dst1, int16_t *dst2,
910 int dstWidth, const uint8_t *src1,
911 const uint8_t *src2, int srcW, int xInc);
914 * Allocate and return an SwsContext.
915 * This is like sws_getContext() but does not perform the init step, allowing
916 * the user to set additional AVOptions.
918 * @see sws_getContext()
920 struct SwsContext *sws_alloc_set_opts(int srcW, int srcH, enum AVPixelFormat srcFormat,
921 int dstW, int dstH, enum AVPixelFormat dstFormat,
922 int flags, const double *param);
924 int ff_sws_alphablendaway(SwsContext *c, const uint8_t *src[],
925 int srcStride[], int srcSliceY, int srcSliceH,
926 uint8_t *dst[], int dstStride[]);
928 static inline void fillPlane16(uint8_t *plane, int stride, int width, int height, int y,
929 int alpha, int bits, const int big_endian)
932 uint8_t *ptr = plane + stride * y;
933 int v = alpha ? 0xFFFF>>(16-bits) : (1<<(bits-1));
934 for (i = 0; i < height; i++) {
935 #define FILL(wfunc) \
936 for (j = 0; j < width; j++) {\
948 #define MAX_SLICE_PLANES 4
951 typedef struct SwsPlane
953 int available_lines; ///< max number of lines that can be hold by this plane
954 int sliceY; ///< index of first line
955 int sliceH; ///< number of lines
956 uint8_t **line; ///< line buffer
957 uint8_t **tmp; ///< Tmp line buffer used by mmx code
961 * Struct which defines a slice of an image to be scaled or a output for
963 * A slice can also be used as intermediate ring buffer for scaling steps.
965 typedef struct SwsSlice
967 int width; ///< Slice line width
968 int h_chr_sub_sample; ///< horizontal chroma subsampling factor
969 int v_chr_sub_sample; ///< vertical chroma subsampling factor
970 int is_ring; ///< flag to identify if this slice is a ring buffer
971 int should_free_lines; ///< flag to identify if there are dynamic allocated lines
972 enum AVPixelFormat fmt; ///< planes pixel format
973 SwsPlane plane[MAX_SLICE_PLANES]; ///< color planes
977 * Struct which holds all necessary data for processing a slice.
978 * A processing step can be a color conversion or horizontal/vertical scaling.
980 typedef struct SwsFilterDescriptor
982 SwsSlice *src; ///< Source slice
983 SwsSlice *dst; ///< Output slice
985 int alpha; ///< Flag for processing alpha channel
986 void *instance; ///< Filter instance data
988 /// Function for processing input slice sliceH lines starting from line sliceY
989 int (*process)(SwsContext *c, struct SwsFilterDescriptor *desc, int sliceY, int sliceH);
990 } SwsFilterDescriptor;
992 /// Color conversion instance data
993 typedef struct ColorContext
998 /// Scaler instance data
999 typedef struct FilterContext
1007 typedef struct VScalerContext
1009 uint16_t *filter[2];
1010 int32_t *filter_pos;
1016 // warp input lines in the form (src + width*i + j) to slice format (line[i][j])
1017 // relative=true means first line src[x][0] otherwise first line is src[x][lum/crh Y]
1018 int ff_init_slice_from_src(SwsSlice * s, uint8_t *src[4], int stride[4], int srcW, int lumY, int lumH, int chrY, int chrH, int relative);
1020 // Initialize scaler filter descriptor chain
1021 int ff_init_filters(SwsContext *c);
1023 // Free all filter data
1024 int ff_free_filters(SwsContext *c);
1027 function for applying ring buffer logic into slice s
1028 It checks if the slice can hold more @lum lines, if yes
1029 do nothing otherwise remove @lum least used lines.
1030 It applies the same procedure for @chr lines.
1032 int ff_rotate_slice(SwsSlice *s, int lum, int chr);
1034 /// initializes gamma conversion descriptor
1035 int ff_init_gamma_convert(SwsFilterDescriptor *desc, SwsSlice * src, uint16_t *table);
1037 /// initializes lum pixel format conversion descriptor
1038 int ff_init_desc_fmt_convert(SwsFilterDescriptor *desc, SwsSlice * src, SwsSlice *dst, uint32_t *pal);
1040 /// initializes lum horizontal scaling descriptor
1041 int ff_init_desc_hscale(SwsFilterDescriptor *desc, SwsSlice *src, SwsSlice *dst, uint16_t *filter, int * filter_pos, int filter_size, int xInc);
1043 /// initializes chr pixel format conversion descriptor
1044 int ff_init_desc_cfmt_convert(SwsFilterDescriptor *desc, SwsSlice * src, SwsSlice *dst, uint32_t *pal);
1046 /// initializes chr horizontal scaling descriptor
1047 int ff_init_desc_chscale(SwsFilterDescriptor *desc, SwsSlice *src, SwsSlice *dst, uint16_t *filter, int * filter_pos, int filter_size, int xInc);
1049 int ff_init_desc_no_chr(SwsFilterDescriptor *desc, SwsSlice * src, SwsSlice *dst);
1051 /// initializes vertical scaling descriptors
1052 int ff_init_vscale(SwsContext *c, SwsFilterDescriptor *desc, SwsSlice *src, SwsSlice *dst);
1054 /// setup vertical scaler functions
1055 void ff_init_vscale_pfn(SwsContext *c, yuv2planar1_fn yuv2plane1, yuv2planarX_fn yuv2planeX,
1056 yuv2interleavedX_fn yuv2nv12cX, yuv2packed1_fn yuv2packed1, yuv2packed2_fn yuv2packed2,
1057 yuv2packedX_fn yuv2packedX, yuv2anyX_fn yuv2anyX, int use_mmx);
1059 //number of extra lines to process
1060 #define MAX_LINES_AHEAD 4
1062 // enable use of refactored scaler code
1065 #endif /* SWSCALE_SWSCALE_INTERNAL_H */