libswscale/swscale_internal.h

   1 /*
   2  * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
   3  *
   4  * This file is part of Libav.
   5  *
   6  * Libav 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  * Libav 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 Libav; 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/log.h"
  34 #include "libavutil/pixfmt.h"
  35 #include "libavutil/pixdesc.h"
  36
  37 #define STR(s) AV_TOSTRING(s) // AV_STRINGIFY is too long
  38
  39 #define FAST_BGR2YV12 // use 7-bit instead of 15-bit coefficients
  40
  41 #define MAX_FILTER_SIZE 256
  42
  43 #if HAVE_BIGENDIAN
  44 #define ALT32_CORR (-1)
  45 #else
  46 #define ALT32_CORR   1
  47 #endif
  48
  49 #if ARCH_X86_64
  50 #   define APCK_PTR2  8
  51 #   define APCK_COEF 16
  52 #   define APCK_SIZE 24
  53 #else
  54 #   define APCK_PTR2  4
  55 #   define APCK_COEF  8
  56 #   define APCK_SIZE 16
  57 #endif
  58
  59 struct SwsContext;
  60
  61 typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
  62                        int srcStride[], int srcSliceY, int srcSliceH,
  63                        uint8_t *dst[], int dstStride[]);
  64
  65 /**
  66  * Write one line of horizontally scaled data to planar output
  67  * without any additional vertical scaling (or point-scaling).
  68  *
  69  * @param src     scaled source data, 15bit for 8-10bit output,
  70  *                19-bit for 16bit output (in int32_t)
  71  * @param dest    pointer to the output plane. For >8bit
  72  *                output, this is in uint16_t
  73  * @param dstW    width of destination in pixels
  74  * @param dither  ordered dither array of type int16_t and size 8
  75  * @param offset  Dither offset
  76  */
  77 typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
  78                                const uint8_t *dither, int offset);
  79
  80 /**
  81  * Write one line of horizontally scaled data to planar output
  82  * with multi-point vertical scaling between input pixels.
  83  *
  84  * @param filter        vertical luma/alpha scaling coefficients, 12bit [0,4096]
  85  * @param src           scaled luma (Y) or alpha (A) source data, 15bit for 8-10bit output,
  86  *                      19-bit for 16bit output (in int32_t)
  87  * @param filterSize    number of vertical input lines to scale
  88  * @param dest          pointer to output plane. For >8bit
  89  *                      output, this is in uint16_t
  90  * @param dstW          width of destination pixels
  91  * @param offset        Dither offset
  92  */
  93 typedef void (*yuv2planarX_fn)(const int16_t *filter, int filterSize,
  94                                const int16_t **src, uint8_t *dest, int dstW,
  95                                const uint8_t *dither, int offset);
  96
  97 /**
  98  * Write one line of horizontally scaled chroma to interleaved output
  99  * with multi-point vertical scaling between input pixels.
 100  *
 101  * @param c             SWS scaling context
 102  * @param chrFilter     vertical chroma scaling coefficients, 12bit [0,4096]
 103  * @param chrUSrc       scaled chroma (U) source data, 15bit for 8-10bit output,
 104  *                      19-bit for 16bit output (in int32_t)
 105  * @param chrVSrc       scaled chroma (V) source data, 15bit for 8-10bit output,
 106  *                      19-bit for 16bit output (in int32_t)
 107  * @param chrFilterSize number of vertical chroma input lines to scale
 108  * @param dest          pointer to the output plane. For >8bit
 109  *                      output, this is in uint16_t
 110  * @param dstW          width of chroma planes
 111  */
 112 typedef void (*yuv2interleavedX_fn)(struct SwsContext *c,
 113                                     const int16_t *chrFilter,
 114                                     int chrFilterSize,
 115                                     const int16_t **chrUSrc,
 116                                     const int16_t **chrVSrc,
 117                                     uint8_t *dest, int dstW);
 118
 119 /**
 120  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
 121  * output without any additional vertical scaling (or point-scaling). Note
 122  * that this function may do chroma scaling, see the "uvalpha" argument.
 123  *
 124  * @param c       SWS scaling context
 125  * @param lumSrc  scaled luma (Y) source data, 15bit for 8-10bit output,
 126  *                19-bit for 16bit output (in int32_t)
 127  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
 128  *                19-bit for 16bit output (in int32_t)
 129  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
 130  *                19-bit for 16bit output (in int32_t)
 131  * @param alpSrc  scaled alpha (A) source data, 15bit for 8-10bit output,
 132  *                19-bit for 16bit output (in int32_t)
 133  * @param dest    pointer to the output plane. For 16bit output, this is
 134  *                uint16_t
 135  * @param dstW    width of lumSrc and alpSrc in pixels, number of pixels
 136  *                to write into dest[]
 137  * @param uvalpha chroma scaling coefficient for the second line of chroma
 138  *                pixels, either 2048 or 0. If 0, one chroma input is used
 139  *                for 2 output pixels (or if the SWS_FLAG_FULL_CHR_INT flag
 140  *                is set, it generates 1 output pixel). If 2048, two chroma
 141  *                input pixels should be averaged for 2 output pixels (this
 142  *                only happens if SWS_FLAG_FULL_CHR_INT is not set)
 143  * @param y       vertical line number for this output. This does not need
 144  *                to be used to calculate the offset in the destination,
 145  *                but can be used to generate comfort noise using dithering
 146  *                for some output formats.
 147  */
 148 typedef void (*yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc,
 149                                const int16_t *chrUSrc[2],
 150                                const int16_t *chrVSrc[2],
 151                                const int16_t *alpSrc, uint8_t *dest,
 152                                int dstW, int uvalpha, int y);
 153 /**
 154  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
 155  * output by doing bilinear scaling between two input lines.
 156  *
 157  * @param c       SWS scaling context
 158  * @param lumSrc  scaled luma (Y) source data, 15bit for 8-10bit output,
 159  *                19-bit for 16bit output (in int32_t)
 160  * @param chrUSrc scaled chroma (U) source data, 15bit for 8-10bit output,
 161  *                19-bit for 16bit output (in int32_t)
 162  * @param chrVSrc scaled chroma (V) source data, 15bit for 8-10bit output,
 163  *                19-bit for 16bit output (in int32_t)
 164  * @param alpSrc  scaled alpha (A) source data, 15bit for 8-10bit output,
 165  *                19-bit for 16bit output (in int32_t)
 166  * @param dest    pointer to the output plane. For 16bit output, this is
 167  *                uint16_t
 168  * @param dstW    width of lumSrc and alpSrc in pixels, number of pixels
 169  *                to write into dest[]
 170  * @param yalpha  luma/alpha scaling coefficients for the second input line.
 171  *                The first line's coefficients can be calculated by using
 172  *                4096 - yalpha
 173  * @param uvalpha chroma scaling coefficient for the second input line. The
 174  *                first line's coefficients can be calculated by using
 175  *                4096 - uvalpha
 176  * @param y       vertical line number for this output. This does not need
 177  *                to be used to calculate the offset in the destination,
 178  *                but can be used to generate comfort noise using dithering
 179  *                for some output formats.
 180  */
 181 typedef void (*yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2],
 182                                const int16_t *chrUSrc[2],
 183                                const int16_t *chrVSrc[2],
 184                                const int16_t *alpSrc[2],
 185                                uint8_t *dest,
 186                                int dstW, int yalpha, int uvalpha, int y);
 187 /**
 188  * Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB
 189  * output by doing multi-point vertical scaling between input pixels.
 190  *
 191  * @param c             SWS scaling context
 192  * @param lumFilter     vertical luma/alpha scaling coefficients, 12bit [0,4096]
 193  * @param lumSrc        scaled luma (Y) source data, 15bit for 8-10bit output,
 194  *                      19-bit for 16bit output (in int32_t)
 195  * @param lumFilterSize number of vertical luma/alpha input lines to scale
 196  * @param chrFilter     vertical chroma scaling coefficients, 12bit [0,4096]
 197  * @param chrUSrc       scaled chroma (U) source data, 15bit for 8-10bit output,
 198  *                      19-bit for 16bit output (in int32_t)
 199  * @param chrVSrc       scaled chroma (V) source data, 15bit for 8-10bit output,
 200  *                      19-bit for 16bit output (in int32_t)
 201  * @param chrFilterSize number of vertical chroma input lines to scale
 202  * @param alpSrc        scaled alpha (A) source data, 15bit for 8-10bit output,
 203  *                      19-bit for 16bit output (in int32_t)
 204  * @param dest          pointer to the output plane. For 16bit output, this is
 205  *                      uint16_t
 206  * @param dstW          width of lumSrc and alpSrc in pixels, number of pixels
 207  *                      to write into dest[]
 208  * @param y             vertical line number for this output. This does not need
 209  *                      to be used to calculate the offset in the destination,
 210  *                      but can be used to generate comfort noise using dithering
 211  *                      or some output formats.
 212  */
 213 typedef void (*yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter,
 214                                const int16_t **lumSrc, int lumFilterSize,
 215                                const int16_t *chrFilter,
 216                                const int16_t **chrUSrc,
 217                                const int16_t **chrVSrc, int chrFilterSize,
 218                                const int16_t **alpSrc, uint8_t *dest,
 219                                int dstW, int y);
 220
 221 /**
 222  * Write one line of horizontally scaled Y/U/V/A to YUV/RGB
 223  * output by doing multi-point vertical scaling between input pixels.
 224  *
 225  * @param c             SWS scaling context
 226  * @param lumFilter     vertical luma/alpha scaling coefficients, 12bit [0,4096]
 227  * @param lumSrc        scaled luma (Y) source data, 15bit for 8-10bit output,
 228  *                      19-bit for 16bit output (in int32_t)
 229  * @param lumFilterSize number of vertical luma/alpha input lines to scale
 230  * @param chrFilter     vertical chroma scaling coefficients, 12bit [0,4096]
 231  * @param chrUSrc       scaled chroma (U) source data, 15bit for 8-10bit output,
 232  *                      19-bit for 16bit output (in int32_t)
 233  * @param chrVSrc       scaled chroma (V) source data, 15bit for 8-10bit output,
 234  *                      19-bit for 16bit output (in int32_t)
 235  * @param chrFilterSize number of vertical chroma input lines to scale
 236  * @param alpSrc        scaled alpha (A) source data, 15bit for 8-10bit output,
 237  *                      19-bit for 16bit output (in int32_t)
 238  * @param dest          pointer to the output planes. For 16bit output, this is
 239  *                      uint16_t
 240  * @param dstW          width of lumSrc and alpSrc in pixels, number of pixels
 241  *                      to write into dest[]
 242  * @param y             vertical line number for this output. This does not need
 243  *                      to be used to calculate the offset in the destination,
 244  *                      but can be used to generate comfort noise using dithering
 245  *                      or some output formats.
 246  */
 247 typedef void (*yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter,
 248                             const int16_t **lumSrc, int lumFilterSize,
 249                             const int16_t *chrFilter,
 250                             const int16_t **chrUSrc,
 251                             const int16_t **chrVSrc, int chrFilterSize,
 252                             const int16_t **alpSrc, uint8_t **dest,
 253                             int dstW, int y);
 254
 255 /* This struct should be aligned on at least a 32-byte boundary. */
 256 typedef struct SwsContext {
 257     /**
 258      * info on struct for av_log
 259      */
 260     const AVClass *av_class;
 261
 262     /**
 263      * Note that src, dst, srcStride, dstStride will be copied in the
 264      * sws_scale() wrapper so they can be freely modified here.
 265      */
 266     SwsFunc swScale;
 267     int srcW;                     ///< Width  of source      luma/alpha planes.
 268     int srcH;                     ///< Height of source      luma/alpha planes.
 269     int dstH;                     ///< Height of destination luma/alpha planes.
 270     int chrSrcW;                  ///< Width  of source      chroma     planes.
 271     int chrSrcH;                  ///< Height of source      chroma     planes.
 272     int chrDstW;                  ///< Width  of destination chroma     planes.
 273     int chrDstH;                  ///< Height of destination chroma     planes.
 274     int lumXInc, chrXInc;
 275     int lumYInc, chrYInc;
 276     enum AVPixelFormat dstFormat; ///< Destination pixel format.
 277     enum AVPixelFormat srcFormat; ///< Source      pixel format.
 278     int dstFormatBpp;             ///< Number of bits per pixel of the destination pixel format.
 279     int srcFormatBpp;             ///< Number of bits per pixel of the source      pixel format.
 280     int dstBpc, srcBpc;
 281     int chrSrcHSubSample;         ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source      image.
 282     int chrSrcVSubSample;         ///< Binary logarithm of vertical   subsampling factor between luma/alpha and chroma planes in source      image.
 283     int chrDstHSubSample;         ///< Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination image.
 284     int chrDstVSubSample;         ///< Binary logarithm of vertical   subsampling factor between luma/alpha and chroma planes in destination image.
 285     int vChrDrop;                 ///< Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user.
 286     int sliceDir;                 ///< Direction that slices are fed to the scaler (1 = top-to-bottom, -1 = bottom-to-top).
 287     double param[2];              ///< Input parameters for scaling algorithms that need them.
 288
 289     uint32_t pal_yuv[256];
 290     uint32_t pal_rgb[256];
 291
 292     /**
 293      * @name Scaled horizontal lines ring buffer.
 294      * The horizontal scaler keeps just enough scaled lines in a ring buffer
 295      * so they may be passed to the vertical scaler. The pointers to the
 296      * allocated buffers for each line are duplicated in sequence in the ring
 297      * buffer to simplify indexing and avoid wrapping around between lines
 298      * inside the vertical scaler code. The wrapping is done before the
 299      * vertical scaler is called.
 300      */
 301     //@{
 302     int16_t **lumPixBuf;          ///< Ring buffer for scaled horizontal luma   plane lines to be fed to the vertical scaler.
 303     int16_t **chrUPixBuf;         ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
 304     int16_t **chrVPixBuf;         ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
 305     int16_t **alpPixBuf;          ///< Ring buffer for scaled horizontal alpha  plane lines to be fed to the vertical scaler.
 306     int vLumBufSize;              ///< Number of vertical luma/alpha lines allocated in the ring buffer.
 307     int vChrBufSize;              ///< Number of vertical chroma     lines allocated in the ring buffer.
 308     int lastInLumBuf;             ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
 309     int lastInChrBuf;             ///< Last scaled horizontal chroma     line from source in the ring buffer.
 310     int lumBufIndex;              ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
 311     int chrBufIndex;              ///< Index in ring buffer of the last scaled horizontal chroma     line from source.
 312     //@}
 313
 314     uint8_t *formatConvBuffer;
 315
 316     /**
 317      * @name Horizontal and vertical filters.
 318      * To better understand the following fields, here is a pseudo-code of
 319      * their usage in filtering a horizontal line:
 320      * @code
 321      * for (i = 0; i < width; i++) {
 322      *     dst[i] = 0;
 323      *     for (j = 0; j < filterSize; j++)
 324      *         dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
 325      *     dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
 326      * }
 327      * @endcode
 328      */
 329     //@{
 330     int16_t *hLumFilter;          ///< Array of horizontal filter coefficients for luma/alpha planes.
 331     int16_t *hChrFilter;          ///< Array of horizontal filter coefficients for chroma     planes.
 332     int16_t *vLumFilter;          ///< Array of vertical   filter coefficients for luma/alpha planes.
 333     int16_t *vChrFilter;          ///< Array of vertical   filter coefficients for chroma     planes.
 334     int32_t *hLumFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
 335     int32_t *hChrFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for chroma     planes.
 336     int32_t *vLumFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for luma/alpha planes.
 337     int32_t *vChrFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for chroma     planes.
 338     int hLumFilterSize;           ///< Horizontal filter size for luma/alpha pixels.
 339     int hChrFilterSize;           ///< Horizontal filter size for chroma     pixels.
 340     int vLumFilterSize;           ///< Vertical   filter size for luma/alpha pixels.
 341     int vChrFilterSize;           ///< Vertical   filter size for chroma     pixels.
 342     //@}
 343
 344     int lumMmxextFilterCodeSize;  ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
 345     int chrMmxextFilterCodeSize;  ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
 346     uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
 347     uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
 348
 349     int canMMXEXTBeUsed;
 350
 351     int dstY;                     ///< Last destination vertical line output from last slice.
 352     int flags;                    ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
 353     void *yuvTable;             // pointer to the yuv->rgb table start so it can be freed()
 354     uint8_t *table_rV[256];
 355     uint8_t *table_gU[256];
 356     int table_gV[256];
 357     uint8_t *table_bU[256];
 358
 359     //Colorspace stuff
 360     int contrast, brightness, saturation;    // for sws_getColorspaceDetails
 361     int srcColorspaceTable[4];
 362     int dstColorspaceTable[4];
 363     int srcRange;                 ///< 0 = MPG YUV range, 1 = JPG YUV range (source      image).
 364     int dstRange;                 ///< 0 = MPG YUV range, 1 = JPG YUV range (destination image).
 365     int yuv2rgb_y_offset;
 366     int yuv2rgb_y_coeff;
 367     int yuv2rgb_v2r_coeff;
 368     int yuv2rgb_v2g_coeff;
 369     int yuv2rgb_u2g_coeff;
 370     int yuv2rgb_u2b_coeff;
 371
 372 #define RED_DITHER            "0*8"
 373 #define GREEN_DITHER          "1*8"
 374 #define BLUE_DITHER           "2*8"
 375 #define Y_COEFF               "3*8"
 376 #define VR_COEFF              "4*8"
 377 #define UB_COEFF              "5*8"
 378 #define VG_COEFF              "6*8"
 379 #define UG_COEFF              "7*8"
 380 #define Y_OFFSET              "8*8"
 381 #define U_OFFSET              "9*8"
 382 #define V_OFFSET              "10*8"
 383 #define LUM_MMX_FILTER_OFFSET "11*8"
 384 #define CHR_MMX_FILTER_OFFSET "11*8+4*4*256"
 385 #define DSTW_OFFSET           "11*8+4*4*256*2" //do not change, it is hardcoded in the ASM
 386 #define ESP_OFFSET            "11*8+4*4*256*2+8"
 387 #define VROUNDER_OFFSET       "11*8+4*4*256*2+16"
 388 #define U_TEMP                "11*8+4*4*256*2+24"
 389 #define V_TEMP                "11*8+4*4*256*2+32"
 390 #define Y_TEMP                "11*8+4*4*256*2+40"
 391 #define ALP_MMX_FILTER_OFFSET "11*8+4*4*256*2+48"
 392 #define UV_OFF_PX             "11*8+4*4*256*3+48"
 393 #define UV_OFF_BYTE           "11*8+4*4*256*3+56"
 394 #define DITHER16              "11*8+4*4*256*3+64"
 395 #define DITHER32              "11*8+4*4*256*3+80"
 396
 397     DECLARE_ALIGNED(8, uint64_t, redDither);
 398     DECLARE_ALIGNED(8, uint64_t, greenDither);
 399     DECLARE_ALIGNED(8, uint64_t, blueDither);
 400
 401     DECLARE_ALIGNED(8, uint64_t, yCoeff);
 402     DECLARE_ALIGNED(8, uint64_t, vrCoeff);
 403     DECLARE_ALIGNED(8, uint64_t, ubCoeff);
 404     DECLARE_ALIGNED(8, uint64_t, vgCoeff);
 405     DECLARE_ALIGNED(8, uint64_t, ugCoeff);
 406     DECLARE_ALIGNED(8, uint64_t, yOffset);
 407     DECLARE_ALIGNED(8, uint64_t, uOffset);
 408     DECLARE_ALIGNED(8, uint64_t, vOffset);
 409     int32_t lumMmxFilter[4 * MAX_FILTER_SIZE];
 410     int32_t chrMmxFilter[4 * MAX_FILTER_SIZE];
 411     int dstW;                     ///< Width  of destination luma/alpha planes.
 412     DECLARE_ALIGNED(8, uint64_t, esp);
 413     DECLARE_ALIGNED(8, uint64_t, vRounder);
 414     DECLARE_ALIGNED(8, uint64_t, u_temp);
 415     DECLARE_ALIGNED(8, uint64_t, v_temp);
 416     DECLARE_ALIGNED(8, uint64_t, y_temp);
 417     int32_t alpMmxFilter[4 * MAX_FILTER_SIZE];
 418     // alignment of these values is not necessary, but merely here
 419     // to maintain the same offset across x8632 and x86-64. Once we
 420     // use proper offset macros in the asm, they can be removed.
 421     DECLARE_ALIGNED(8, ptrdiff_t, uv_off_px);   ///< offset (in pixels) between u and v planes
 422     DECLARE_ALIGNED(8, ptrdiff_t, uv_off_byte); ///< offset (in bytes) between u and v planes
 423     DECLARE_ALIGNED(8, uint16_t, dither16)[8];
 424     DECLARE_ALIGNED(8, uint32_t, dither32)[8];
 425
 426     const uint8_t *chrDither8, *lumDither8;
 427
 428 #if HAVE_ALTIVEC
 429     vector signed short   CY;
 430     vector signed short   CRV;
 431     vector signed short   CBU;
 432     vector signed short   CGU;
 433     vector signed short   CGV;
 434     vector signed short   OY;
 435     vector unsigned short CSHIFT;
 436     vector signed short  *vYCoeffsBank, *vCCoeffsBank;
 437 #endif
 438
 439 #if ARCH_BFIN
 440     DECLARE_ALIGNED(4, uint32_t, oy);
 441     DECLARE_ALIGNED(4, uint32_t, oc);
 442     DECLARE_ALIGNED(4, uint32_t, zero);
 443     DECLARE_ALIGNED(4, uint32_t, cy);
 444     DECLARE_ALIGNED(4, uint32_t, crv);
 445     DECLARE_ALIGNED(4, uint32_t, rmask);
 446     DECLARE_ALIGNED(4, uint32_t, cbu);
 447     DECLARE_ALIGNED(4, uint32_t, bmask);
 448     DECLARE_ALIGNED(4, uint32_t, cgu);
 449     DECLARE_ALIGNED(4, uint32_t, cgv);
 450     DECLARE_ALIGNED(4, uint32_t, gmask);
 451 #endif
 452
 453 #if HAVE_VIS
 454     DECLARE_ALIGNED(8, uint64_t, sparc_coeffs)[10];
 455 #endif
 456
 457     /* function pointers for swScale() */
 458     yuv2planar1_fn yuv2plane1;
 459     yuv2planarX_fn yuv2planeX;
 460     yuv2interleavedX_fn yuv2nv12cX;
 461     yuv2packed1_fn yuv2packed1;
 462     yuv2packed2_fn yuv2packed2;
 463     yuv2packedX_fn yuv2packedX;
 464     yuv2anyX_fn yuv2anyX;
 465
 466     /// Unscaled conversion of luma plane to YV12 for horizontal scaler.
 467     void (*lumToYV12)(uint8_t *dst, const uint8_t *src,
 468                       int width, uint32_t *pal);
 469     /// Unscaled conversion of alpha plane to YV12 for horizontal scaler.
 470     void (*alpToYV12)(uint8_t *dst, const uint8_t *src,
 471                       int width, uint32_t *pal);
 472     /// Unscaled conversion of chroma planes to YV12 for horizontal scaler.
 473     void (*chrToYV12)(uint8_t *dstU, uint8_t *dstV,
 474                       const uint8_t *src1, const uint8_t *src2,
 475                       int width, uint32_t *pal);
 476
 477     /**
 478      * Functions to read planar input, such as planar RGB, and convert
 479      * internally to Y/UV.
 480      */
 481     /** @{ */
 482     void (*readLumPlanar)(uint8_t *dst, const uint8_t *src[4], int width);
 483     void (*readChrPlanar)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4],
 484                           int width);
 485     /** @} */
 486
 487     /**
 488      * Scale one horizontal line of input data using a bilinear filter
 489      * to produce one line of output data. Compared to SwsContext->hScale(),
 490      * please take note of the following caveats when using these:
 491      * - Scaling is done using only 7bit instead of 14bit coefficients.
 492      * - You can use no more than 5 input pixels to produce 4 output
 493      *   pixels. Therefore, this filter should not be used for downscaling
 494      *   by more than ~20% in width (because that equals more than 5/4th
 495      *   downscaling and thus more than 5 pixels input per 4 pixels output).
 496      * - In general, bilinear filters create artifacts during downscaling
 497      *   (even when <20%), because one output pixel will span more than one
 498      *   input pixel, and thus some pixels will need edges of both neighbor
 499      *   pixels to interpolate the output pixel. Since you can use at most
 500      *   two input pixels per output pixel in bilinear scaling, this is
 501      *   impossible and thus downscaling by any size will create artifacts.
 502      * To enable this type of scaling, set SWS_FLAG_FAST_BILINEAR
 503      * in SwsContext->flags.
 504      */
 505     /** @{ */
 506     void (*hyscale_fast)(struct SwsContext *c,
 507                          int16_t *dst, int dstWidth,
 508                          const uint8_t *src, int srcW, int xInc);
 509     void (*hcscale_fast)(struct SwsContext *c,
 510                          int16_t *dst1, int16_t *dst2, int dstWidth,
 511                          const uint8_t *src1, const uint8_t *src2,
 512                          int srcW, int xInc);
 513     /** @} */
 514
 515     /**
 516      * Scale one horizontal line of input data using a filter over the input
 517      * lines, to produce one (differently sized) line of output data.
 518      *
 519      * @param dst        pointer to destination buffer for horizontally scaled
 520      *                   data. If the number of bits per component of one
 521      *                   destination pixel (SwsContext->dstBpc) is <= 10, data
 522      *                   will be 15bpc in 16bits (int16_t) width. Else (i.e.
 523      *                   SwsContext->dstBpc == 16), data will be 19bpc in
 524      *                   32bits (int32_t) width.
 525      * @param dstW       width of destination image
 526      * @param src        pointer to source data to be scaled. If the number of
 527      *                   bits per component of a source pixel (SwsContext->srcBpc)
 528      *                   is 8, this is 8bpc in 8bits (uint8_t) width. Else
 529      *                   (i.e. SwsContext->dstBpc > 8), this is native depth
 530      *                   in 16bits (uint16_t) width. In other words, for 9-bit
 531      *                   YUV input, this is 9bpc, for 10-bit YUV input, this is
 532      *                   10bpc, and for 16-bit RGB or YUV, this is 16bpc.
 533      * @param filter     filter coefficients to be used per output pixel for
 534      *                   scaling. This contains 14bpp filtering coefficients.
 535      *                   Guaranteed to contain dstW * filterSize entries.
 536      * @param filterPos  position of the first input pixel to be used for
 537      *                   each output pixel during scaling. Guaranteed to
 538      *                   contain dstW entries.
 539      * @param filterSize the number of input coefficients to be used (and
 540      *                   thus the number of input pixels to be used) for
 541      *                   creating a single output pixel. Is aligned to 4
 542      *                   (and input coefficients thus padded with zeroes)
 543      *                   to simplify creating SIMD code.
 544      */
 545     /** @{ */
 546     void (*hyScale)(struct SwsContext *c, int16_t *dst, int dstW,
 547                     const uint8_t *src, const int16_t *filter,
 548                     const int32_t *filterPos, int filterSize);
 549     void (*hcScale)(struct SwsContext *c, int16_t *dst, int dstW,
 550                     const uint8_t *src, const int16_t *filter,
 551                     const int32_t *filterPos, int filterSize);
 552     /** @} */
 553
 554     /// Color range conversion function for luma plane if needed.
 555     void (*lumConvertRange)(int16_t *dst, int width);
 556     /// Color range conversion function for chroma planes if needed.
 557     void (*chrConvertRange)(int16_t *dst1, int16_t *dst2, int width);
 558
 559     int needs_hcscale; ///< Set if there are chroma planes to be converted.
 560 } SwsContext;
 561 //FIXME check init (where 0)
 562
 563 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c);
 564 int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
 565                              int fullRange, int brightness,
 566                              int contrast, int saturation);
 567
 568 void ff_yuv2rgb_init_tables_altivec(SwsContext *c, const int inv_table[4],
 569                                     int brightness, int contrast, int saturation);
 570 void updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
 571                            int lastInLumBuf, int lastInChrBuf);
 572
 573 SwsFunc ff_yuv2rgb_init_mmx(SwsContext *c);
 574 SwsFunc ff_yuv2rgb_init_vis(SwsContext *c);
 575 SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c);
 576 SwsFunc ff_yuv2rgb_get_func_ptr_bfin(SwsContext *c);
 577 void ff_bfin_get_unscaled_swscale(SwsContext *c);
 578
 579 const char *sws_format_name(enum AVPixelFormat format);
 580
 581 static av_always_inline int is16BPS(enum AVPixelFormat pix_fmt)
 582 {
 583     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 584     av_assert0(desc);
 585     return desc->comp[0].depth_minus1 == 15;
 586 }
 587
 588 static av_always_inline int is9_OR_10BPS(enum AVPixelFormat pix_fmt)
 589 {
 590     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 591     av_assert0(desc);
 592     return desc->comp[0].depth_minus1 == 8 || desc->comp[0].depth_minus1 == 9;
 593 }
 594
 595 static av_always_inline int isBE(enum AVPixelFormat pix_fmt)
 596 {
 597     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 598     av_assert0(desc);
 599     return desc->flags & PIX_FMT_BE;
 600 }
 601
 602 static av_always_inline int isYUV(enum AVPixelFormat pix_fmt)
 603 {
 604     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 605     av_assert0(desc);
 606     return !(desc->flags & PIX_FMT_RGB) && desc->nb_components >= 2;
 607 }
 608
 609 static av_always_inline int isPlanarYUV(enum AVPixelFormat pix_fmt)
 610 {
 611     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 612     av_assert0(desc);
 613     return ((desc->flags & PIX_FMT_PLANAR) && isYUV(pix_fmt));
 614 }
 615
 616 static av_always_inline int isRGB(enum AVPixelFormat pix_fmt)
 617 {
 618     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 619     av_assert0(desc);
 620     return (desc->flags & PIX_FMT_RGB);
 621 }
 622
 623 #if 0 // FIXME
 624 #define isGray(x) \
 625     (!(av_pix_fmt_descriptors[x].flags & PIX_FMT_PAL) && \
 626      av_pix_fmt_descriptors[x].nb_components <= 2)
 627 #else
 628 #define isGray(x)                      \
 629     ((x) == AV_PIX_FMT_GRAY8       ||  \
 630      (x) == AV_PIX_FMT_Y400A       ||  \
 631      (x) == AV_PIX_FMT_GRAY16BE    ||  \
 632      (x) == AV_PIX_FMT_GRAY16LE)
 633 #endif
 634
 635 #define isRGBinInt(x)                  \
 636     ((x) == AV_PIX_FMT_RGB48BE     ||  \
 637      (x) == AV_PIX_FMT_RGB48LE     ||  \
 638      (x) == AV_PIX_FMT_RGB32       ||  \
 639      (x) == AV_PIX_FMT_RGB32_1     ||  \
 640      (x) == AV_PIX_FMT_RGB24       ||  \
 641      (x) == AV_PIX_FMT_RGB565BE    ||  \
 642      (x) == AV_PIX_FMT_RGB565LE    ||  \
 643      (x) == AV_PIX_FMT_RGB555BE    ||  \
 644      (x) == AV_PIX_FMT_RGB555LE    ||  \
 645      (x) == AV_PIX_FMT_RGB444BE    ||  \
 646      (x) == AV_PIX_FMT_RGB444LE    ||  \
 647      (x) == AV_PIX_FMT_RGB8        ||  \
 648      (x) == AV_PIX_FMT_RGB4        ||  \
 649      (x) == AV_PIX_FMT_RGB4_BYTE   ||  \
 650      (x) == AV_PIX_FMT_MONOBLACK   ||  \
 651      (x) == AV_PIX_FMT_MONOWHITE)
 652
 653 #define isBGRinInt(x)                  \
 654     ((x) == AV_PIX_FMT_BGR48BE     ||  \
 655      (x) == AV_PIX_FMT_BGR48LE     ||  \
 656      (x) == AV_PIX_FMT_BGR32       ||  \
 657      (x) == AV_PIX_FMT_BGR32_1     ||  \
 658      (x) == AV_PIX_FMT_BGR24       ||  \
 659      (x) == AV_PIX_FMT_BGR565BE    ||  \
 660      (x) == AV_PIX_FMT_BGR565LE    ||  \
 661      (x) == AV_PIX_FMT_BGR555BE    ||  \
 662      (x) == AV_PIX_FMT_BGR555LE    ||  \
 663      (x) == AV_PIX_FMT_BGR444BE    ||  \
 664      (x) == AV_PIX_FMT_BGR444LE    ||  \
 665      (x) == AV_PIX_FMT_BGR8        ||  \
 666      (x) == AV_PIX_FMT_BGR4        ||  \
 667      (x) == AV_PIX_FMT_BGR4_BYTE   ||  \
 668      (x) == AV_PIX_FMT_MONOBLACK   ||  \
 669      (x) == AV_PIX_FMT_MONOWHITE)
 670
 671 #define isAnyRGB(x)                    \
 672     (isRGBinInt(x)              ||     \
 673      isBGRinInt(x))
 674
 675 static av_always_inline int isALPHA(enum AVPixelFormat pix_fmt)
 676 {
 677     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 678     av_assert0(desc);
 679     return desc->nb_components == 2 || desc->nb_components == 4;
 680 }
 681
 682 static av_always_inline int isPacked(enum AVPixelFormat pix_fmt)
 683 {
 684     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 685     av_assert0(desc);
 686     return ((desc->nb_components >= 2 && !(desc->flags & PIX_FMT_PLANAR)) ||
 687             pix_fmt == AV_PIX_FMT_PAL8);
 688 }
 689
 690 static av_always_inline int isPlanar(enum AVPixelFormat pix_fmt)
 691 {
 692     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 693     av_assert0(desc);
 694     return (desc->nb_components >= 2 && (desc->flags & PIX_FMT_PLANAR));
 695 }
 696
 697 static av_always_inline int isPackedRGB(enum AVPixelFormat pix_fmt)
 698 {
 699     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 700     av_assert0(desc);
 701     return ((desc->flags & (PIX_FMT_PLANAR | PIX_FMT_RGB)) == PIX_FMT_RGB);
 702 }
 703
 704 static av_always_inline int isPlanarRGB(enum AVPixelFormat pix_fmt)
 705 {
 706     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 707     av_assert0(desc);
 708     return ((desc->flags & (PIX_FMT_PLANAR | PIX_FMT_RGB)) ==
 709             (PIX_FMT_PLANAR | PIX_FMT_RGB));
 710 }
 711
 712 static av_always_inline int usePal(enum AVPixelFormat pix_fmt)
 713 {
 714     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
 715     av_assert0(desc);
 716     return ((desc->flags & PIX_FMT_PAL) || (desc->flags & PIX_FMT_PSEUDOPAL) ||
 717             pix_fmt == AV_PIX_FMT_Y400A);
 718 }
 719
 720 extern const uint64_t ff_dither4[2];
 721 extern const uint64_t ff_dither8[2];
 722
 723 extern const AVClass sws_context_class;
 724
 725 /**
 726  * Set c->swScale to an unscaled converter if one exists for the specific
 727  * source and destination formats, bit depths, flags, etc.
 728  */
 729 void ff_get_unscaled_swscale(SwsContext *c);
 730
 731 void ff_swscale_get_unscaled_altivec(SwsContext *c);
 732
 733 /**
 734  * Return function pointer to fastest main scaler path function depending
 735  * on architecture and available optimizations.
 736  */
 737 SwsFunc ff_getSwsFunc(SwsContext *c);
 738
 739 void ff_sws_init_input_funcs(SwsContext *c);
 740 void ff_sws_init_output_funcs(SwsContext *c,
 741                               yuv2planar1_fn *yuv2plane1,
 742                               yuv2planarX_fn *yuv2planeX,
 743                               yuv2interleavedX_fn *yuv2nv12cX,
 744                               yuv2packed1_fn *yuv2packed1,
 745                               yuv2packed2_fn *yuv2packed2,
 746                               yuv2packedX_fn *yuv2packedX,
 747                               yuv2anyX_fn *yuv2anyX);
 748 void ff_sws_init_swScale_altivec(SwsContext *c);
 749 void ff_sws_init_swScale_mmx(SwsContext *c);
 750
 751 #endif /* SWSCALE_SWSCALE_INTERNAL_H */