Subversion Repositories Kolibri OS

Rev

Blame | Last modification | View Log | RSS feed

  1. /*
  2.  * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
  3.  *
  4.  * This file is part of FFmpeg.
  5.  *
  6.  * FFmpeg is free software; you can redistribute it and/or
  7.  * modify it under the terms of the GNU Lesser General Public
  8.  * License as published by the Free Software Foundation; either
  9.  * version 2.1 of the License, or (at your option) any later version.
  10.  *
  11.  * FFmpeg is distributed in the hope that it will be useful,
  12.  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13.  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14.  * Lesser General Public License for more details.
  15.  *
  16.  * You should have received a copy of the GNU Lesser General Public
  17.  * License along with FFmpeg; if not, write to the Free Software
  18.  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19.  */
  20.  
  21. #ifndef SWSCALE_SWSCALE_INTERNAL_H
  22. #define SWSCALE_SWSCALE_INTERNAL_H
  23.  
  24. #include "config.h"
  25.  
  26. #if HAVE_ALTIVEC_H
  27. #include <altivec.h>
  28. #endif
  29.  
  30. #include "version.h"
  31.  
  32. #include "libavutil/avassert.h"
  33. #include "libavutil/avutil.h"
  34. #include "libavutil/common.h"
  35. #include "libavutil/intreadwrite.h"
  36. #include "libavutil/log.h"
  37. #include "libavutil/pixfmt.h"
  38. #include "libavutil/pixdesc.h"
  39.  
  40. #define STR(s) AV_TOSTRING(s) // AV_STRINGIFY is too long
  41.  
  42. #define YUVRGB_TABLE_HEADROOM 512
  43. #define YUVRGB_TABLE_LUMA_HEADROOM 512
  44.  
  45. #define MAX_FILTER_SIZE SWS_MAX_FILTER_SIZE
  46.  
  47. #define DITHER1XBPP
  48.  
  49. #if HAVE_BIGENDIAN
  50. #define ALT32_CORR (-1)
  51. #else
  52. #define ALT32_CORR   1
  53. #endif
  54.  
  55. #if ARCH_X86_64
  56. #   define APCK_PTR2  8
  57. #   define APCK_COEF 16
  58. #   define APCK_SIZE 24
  59. #else
  60. #   define APCK_PTR2  4
  61. #   define APCK_COEF  8
  62. #   define APCK_SIZE 16
  63. #endif
  64.  
  65. #define RETCODE_USE_CASCADE -12345
  66.  
  67. struct SwsContext;
  68.  
  69. typedef enum SwsDither {
  70.     SWS_DITHER_NONE = 0,
  71.     SWS_DITHER_AUTO,
  72.     SWS_DITHER_BAYER,
  73.     SWS_DITHER_ED,
  74.     SWS_DITHER_A_DITHER,
  75.     SWS_DITHER_X_DITHER,
  76.     NB_SWS_DITHER,
  77. } SwsDither;
  78.  
  79. typedef enum SwsAlphaBlend {
  80.     SWS_ALPHA_BLEND_NONE  = 0,
  81.     SWS_ALPHA_BLEND_UNIFORM,
  82.     SWS_ALPHA_BLEND_CHECKERBOARD,
  83.     SWS_ALPHA_BLEND_NB,
  84. } SwsAlphaBlend;
  85.  
  86. typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
  87.                        int srcStride[], int srcSliceY, int srcSliceH,
  88.                        uint8_t *dst[], int dstStride[]);
  89.  
  90. /**
  91.  * Write one line of horizontally scaled data to planar output
  92.  * without any additional vertical scaling (or point-scaling).
  93.  *
  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
  101.  */
  102. typedef void (*yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW,
  103.                                const uint8_t *dither, int offset);
  104.  
  105. /**
  106.  * Write one line of horizontally scaled data to planar output
  107.  * with multi-point vertical scaling between input pixels.
  108.  *
  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
  117.  */
  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);
  121.  
  122. /**
  123.  * Write one line of horizontally scaled chroma to interleaved output
  124.  * with multi-point vertical scaling between input pixels.
  125.  *
  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
  136.  */
  137. typedef void (*yuv2interleavedX_fn)(struct SwsContext *c,
  138.                                     const int16_t *chrFilter,
  139.                                     int chrFilterSize,
  140.                                     const int16_t **chrUSrc,
  141.                                     const int16_t **chrVSrc,
  142.                                     uint8_t *dest, int dstW);
  143.  
  144. /**
  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.
  148.  *
  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
  159.  *                uint16_t
  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.
  172.  */
  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);
  178. /**
  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.
  181.  *
  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
  192.  *                uint16_t
  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
  197.  *                4096 - yalpha
  198.  * @param uvalpha chroma scaling coefficient for the second input line. The
  199.  *                first line's coefficients can be calculated by using
  200.  *                4096 - uvalpha
  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.
  205.  */
  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],
  210.                                uint8_t *dest,
  211.                                int dstW, int yalpha, int uvalpha, int y);
  212. /**
  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.
  215.  *
  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
  230.  *                      uint16_t
  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.
  237.  */
  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,
  244.                                int dstW, int y);
  245.  
  246. /**
  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.
  249.  *
  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
  264.  *                      uint16_t
  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.
  271.  */
  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,
  278.                             int dstW, int y);
  279.  
  280. struct SwsSlice;
  281. struct SwsFilterDescriptor;
  282.  
  283. /* This struct should be aligned on at least a 32-byte boundary. */
  284. typedef struct SwsContext {
  285.     /**
  286.      * info on struct for av_log
  287.      */
  288.     const AVClass *av_class;
  289.  
  290.     /**
  291.      * Note that src, dst, srcStride, dstStride will be copied in the
  292.      * sws_scale() wrapper so they can be freely modified here.
  293.      */
  294.     SwsFunc swscale;
  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.
  308.     int dstBpc, srcBpc;
  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.
  316.  
  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.
  320.      */
  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.  
  327.     double gamma_value;
  328.     int gamma_flag;
  329.     int is_internal_gamma;
  330.     uint16_t *gamma;
  331.     uint16_t *inv_gamma;
  332.  
  333.     int numDesc;
  334.     int descIndex[2];
  335.     int numSlice;
  336.     struct SwsSlice *slice;
  337.     struct SwsFilterDescriptor *desc;
  338.  
  339.     uint32_t pal_yuv[256];
  340.     uint32_t pal_rgb[256];
  341.  
  342.     /**
  343.      * @name Scaled horizontal lines ring buffer.
  344.      * The horizontal scaler keeps just enough scaled lines in a ring buffer
  345.      * so they may be passed to the vertical scaler. The pointers to the
  346.      * allocated buffers for each line are duplicated in sequence in the ring
  347.      * buffer to simplify indexing and avoid wrapping around between lines
  348.      * inside the vertical scaler code. The wrapping is done before the
  349.      * vertical scaler is called.
  350.      */
  351.     //@{
  352.     int16_t **lumPixBuf;          ///< Ring buffer for scaled horizontal luma   plane lines to be fed to the vertical scaler.
  353.     int16_t **chrUPixBuf;         ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
  354.     int16_t **chrVPixBuf;         ///< Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
  355.     int16_t **alpPixBuf;          ///< Ring buffer for scaled horizontal alpha  plane lines to be fed to the vertical scaler.
  356.     int vLumBufSize;              ///< Number of vertical luma/alpha lines allocated in the ring buffer.
  357.     int vChrBufSize;              ///< Number of vertical chroma     lines allocated in the ring buffer.
  358.     int lastInLumBuf;             ///< Last scaled horizontal luma/alpha line from source in the ring buffer.
  359.     int lastInChrBuf;             ///< Last scaled horizontal chroma     line from source in the ring buffer.
  360.     int lumBufIndex;              ///< Index in ring buffer of the last scaled horizontal luma/alpha line from source.
  361.     int chrBufIndex;              ///< Index in ring buffer of the last scaled horizontal chroma     line from source.
  362.     //@}
  363.  
  364.     uint8_t *formatConvBuffer;
  365.  
  366.     /**
  367.      * @name Horizontal and vertical filters.
  368.      * To better understand the following fields, here is a pseudo-code of
  369.      * their usage in filtering a horizontal line:
  370.      * @code
  371.      * for (i = 0; i < width; i++) {
  372.      *     dst[i] = 0;
  373.      *     for (j = 0; j < filterSize; j++)
  374.      *         dst[i] += src[ filterPos[i] + j ] * filter[ filterSize * i + j ];
  375.      *     dst[i] >>= FRAC_BITS; // The actual implementation is fixed-point.
  376.      * }
  377.      * @endcode
  378.      */
  379.     //@{
  380.     int16_t *hLumFilter;          ///< Array of horizontal filter coefficients for luma/alpha planes.
  381.     int16_t *hChrFilter;          ///< Array of horizontal filter coefficients for chroma     planes.
  382.     int16_t *vLumFilter;          ///< Array of vertical   filter coefficients for luma/alpha planes.
  383.     int16_t *vChrFilter;          ///< Array of vertical   filter coefficients for chroma     planes.
  384.     int32_t *hLumFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
  385.     int32_t *hChrFilterPos;       ///< Array of horizontal filter starting positions for each dst[i] for chroma     planes.
  386.     int32_t *vLumFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for luma/alpha planes.
  387.     int32_t *vChrFilterPos;       ///< Array of vertical   filter starting positions for each dst[i] for chroma     planes.
  388.     int hLumFilterSize;           ///< Horizontal filter size for luma/alpha pixels.
  389.     int hChrFilterSize;           ///< Horizontal filter size for chroma     pixels.
  390.     int vLumFilterSize;           ///< Vertical   filter size for luma/alpha pixels.
  391.     int vChrFilterSize;           ///< Vertical   filter size for chroma     pixels.
  392.     //@}
  393.  
  394.     int lumMmxextFilterCodeSize;  ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes.
  395.     int chrMmxextFilterCodeSize;  ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
  396.     uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
  397.     uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
  398.  
  399.     int canMMXEXTBeUsed;
  400.     int warned_unuseable_bilinear;
  401.  
  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
  412. #define RY_IDX 0
  413. #define GY_IDX 1
  414. #define BY_IDX 2
  415. #define RU_IDX 3
  416. #define GU_IDX 4
  417. #define BU_IDX 5
  418. #define RV_IDX 6
  419. #define GV_IDX 7
  420. #define BV_IDX 8
  421. #define RGB2YUV_SHIFT 15
  422.  
  423.     int *dither_error[4];
  424.  
  425.     //Colorspace stuff
  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).
  431.     int src0Alpha;
  432.     int dst0Alpha;
  433.     int srcXYZ;
  434.     int dstXYZ;
  435.     int src_h_chr_pos;
  436.     int dst_h_chr_pos;
  437.     int src_v_chr_pos;
  438.     int dst_v_chr_pos;
  439.     int yuv2rgb_y_offset;
  440.     int yuv2rgb_y_coeff;
  441.     int yuv2rgb_v2r_coeff;
  442.     int yuv2rgb_v2g_coeff;
  443.     int yuv2rgb_u2g_coeff;
  444.     int yuv2rgb_u2b_coeff;
  445.  
  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
  471.  
  472.     DECLARE_ALIGNED(8, uint64_t, redDither);
  473.     DECLARE_ALIGNED(8, uint64_t, greenDither);
  474.     DECLARE_ALIGNED(8, uint64_t, blueDither);
  475.  
  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];
  500.  
  501.     const uint8_t *chrDither8, *lumDither8;
  502.  
  503. #if HAVE_ALTIVEC
  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;
  512. #endif
  513.  
  514.     int use_mmx_vfilter;
  515.  
  516. /* pre defined color-spaces gamma */
  517. #define XYZ_GAMMA (2.6f)
  518. #define RGB_GAMMA (2.2f)
  519.     int16_t *xyzgamma;
  520.     int16_t *rgbgamma;
  521.     int16_t *xyzgammainv;
  522.     int16_t *rgbgammainv;
  523.     int16_t xyz2rgb_matrix[3][4];
  524.     int16_t rgb2xyz_matrix[3][4];
  525.  
  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;
  534.  
  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);
  545.  
  546.     /**
  547.      * Functions to read planar input, such as planar RGB, and convert
  548.      * internally to Y/UV/A.
  549.      */
  550.     /** @{ */
  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);
  555.     /** @} */
  556.  
  557.     /**
  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.
  574.      */
  575.     /** @{ */
  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,
  582.                          int srcW, int xInc);
  583.     /** @} */
  584.  
  585.     /**
  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.
  588.      *
  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.
  614.      */
  615.     /** @{ */
  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);
  622.     /** @} */
  623.  
  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);
  628.  
  629.     int needs_hcscale; ///< Set if there are chroma planes to be converted.
  630.  
  631.     SwsDither dither;
  632.  
  633.     SwsAlphaBlend alphablend;
  634. } SwsContext;
  635. //FIXME check init (where 0)
  636.  
  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);
  643.  
  644. void ff_updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
  645.                            int lastInLumBuf, int lastInChrBuf);
  646.  
  647. av_cold void ff_sws_init_range_convert(SwsContext *c);
  648.  
  649. SwsFunc ff_yuv2rgb_init_x86(SwsContext *c);
  650. SwsFunc ff_yuv2rgb_init_ppc(SwsContext *c);
  651.  
  652. static av_always_inline int is16BPS(enum AVPixelFormat pix_fmt)
  653. {
  654.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  655.     av_assert0(desc);
  656.     return desc->comp[0].depth_minus1 == 15;
  657. }
  658.  
  659. static av_always_inline int is9_OR_10BPS(enum AVPixelFormat pix_fmt)
  660. {
  661.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  662.     av_assert0(desc);
  663.     return desc->comp[0].depth_minus1 >= 8 && desc->comp[0].depth_minus1 <= 13;
  664. }
  665.  
  666. #define isNBPS(x) is9_OR_10BPS(x)
  667.  
  668. static av_always_inline int isBE(enum AVPixelFormat pix_fmt)
  669. {
  670.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  671.     av_assert0(desc);
  672.     return desc->flags & AV_PIX_FMT_FLAG_BE;
  673. }
  674.  
  675. static av_always_inline int isYUV(enum AVPixelFormat pix_fmt)
  676. {
  677.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  678.     av_assert0(desc);
  679.     return !(desc->flags & AV_PIX_FMT_FLAG_RGB) && desc->nb_components >= 2;
  680. }
  681.  
  682. static av_always_inline int isPlanarYUV(enum AVPixelFormat pix_fmt)
  683. {
  684.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  685.     av_assert0(desc);
  686.     return ((desc->flags & AV_PIX_FMT_FLAG_PLANAR) && isYUV(pix_fmt));
  687. }
  688.  
  689. static av_always_inline int isRGB(enum AVPixelFormat pix_fmt)
  690. {
  691.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  692.     av_assert0(desc);
  693.     return (desc->flags & AV_PIX_FMT_FLAG_RGB);
  694. }
  695.  
  696. #if 0 // FIXME
  697. #define isGray(x) \
  698.     (!(av_pix_fmt_desc_get(x)->flags & AV_PIX_FMT_FLAG_PAL) && \
  699.      av_pix_fmt_desc_get(x)->nb_components <= 2)
  700. #else
  701. #define isGray(x)                      \
  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)
  708. #endif
  709.  
  710. #define isRGBinInt(x) \
  711.     (           \
  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   \
  730.     )
  731. #define isBGRinInt(x) \
  732.     (           \
  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   \
  751.     )
  752.  
  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       \
  761.     )
  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       \
  770.     )
  771.  
  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 \
  785.     )
  786.  
  787. #define isAnyRGB(x) \
  788.     (           \
  789.           isBayer(x)          ||    \
  790.           isRGBinInt(x)       ||    \
  791.           isBGRinInt(x)       ||    \
  792.           isRGB(x)      \
  793.     )
  794.  
  795. static av_always_inline int isALPHA(enum AVPixelFormat pix_fmt)
  796. {
  797.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  798.     av_assert0(desc);
  799.     if (pix_fmt == AV_PIX_FMT_PAL8)
  800.         return 1;
  801.     return desc->flags & AV_PIX_FMT_FLAG_ALPHA;
  802. }
  803.  
  804. #if 1
  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    \
  815.         ||  isRGBinInt(x)           \
  816.         ||  isBGRinInt(x)           \
  817.     )
  818. #else
  819. static av_always_inline int isPacked(enum AVPixelFormat pix_fmt)
  820. {
  821.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  822.     av_assert0(desc);
  823.     return ((desc->nb_components >= 2 && !(desc->flags & AV_PIX_FMT_FLAG_PLANAR)) ||
  824.             pix_fmt == AV_PIX_FMT_PAL8);
  825. }
  826.  
  827. #endif
  828. static av_always_inline int isPlanar(enum AVPixelFormat pix_fmt)
  829. {
  830.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  831.     av_assert0(desc);
  832.     return (desc->nb_components >= 2 && (desc->flags & AV_PIX_FMT_FLAG_PLANAR));
  833. }
  834.  
  835. static av_always_inline int isPackedRGB(enum AVPixelFormat pix_fmt)
  836. {
  837.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  838.     av_assert0(desc);
  839.     return ((desc->flags & (AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB)) == AV_PIX_FMT_FLAG_RGB);
  840. }
  841.  
  842. static av_always_inline int isPlanarRGB(enum AVPixelFormat pix_fmt)
  843. {
  844.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  845.     av_assert0(desc);
  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));
  848. }
  849.  
  850. static av_always_inline int usePal(enum AVPixelFormat pix_fmt)
  851. {
  852.     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
  853.     av_assert0(desc);
  854.     return (desc->flags & AV_PIX_FMT_FLAG_PAL) || (desc->flags & AV_PIX_FMT_FLAG_PSEUDOPAL);
  855. }
  856.  
  857. extern const uint64_t ff_dither4[2];
  858. extern const uint64_t ff_dither8[2];
  859.  
  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];
  867.  
  868. extern const int32_t ff_yuv2rgb_coeffs[8][4];
  869.  
  870. extern const AVClass sws_context_class;
  871.  
  872. /**
  873.  * Set c->swscale to an unscaled converter if one exists for the specific
  874.  * source and destination formats, bit depths, flags, etc.
  875.  */
  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);
  879.  
  880. /**
  881.  * Return function pointer to fastest main scaler path function depending
  882.  * on architecture and available optimizations.
  883.  */
  884. SwsFunc ff_getSwsFunc(SwsContext *c);
  885.  
  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);
  897.  
  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,
  905.                            int numSplits);
  906. void ff_hyscale_fast_mmxext(SwsContext *c, int16_t *dst,
  907.                             int dstWidth, const uint8_t *src,
  908.                             int srcW, int xInc);
  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);
  912.  
  913. /**
  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.
  917.  *
  918.  * @see sws_getContext()
  919.  */
  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);
  923.  
  924. int ff_sws_alphablendaway(SwsContext *c, const uint8_t *src[],
  925.                           int srcStride[], int srcSliceY, int srcSliceH,
  926.                           uint8_t *dst[], int dstStride[]);
  927.  
  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)
  930. {
  931.     int i, j;
  932.     uint8_t *ptr = plane + stride * y;
  933.     int v = alpha ? 0xFFFF>>(15-bits) : (1<<bits);
  934.     for (i = 0; i < height; i++) {
  935. #define FILL(wfunc) \
  936.         for (j = 0; j < width; j++) {\
  937.             wfunc(ptr+2*j, v);\
  938.         }
  939.         if (big_endian) {
  940.             FILL(AV_WB16);
  941.         } else {
  942.             FILL(AV_WL16);
  943.         }
  944.         ptr += stride;
  945.     }
  946. }
  947.  
  948. #define MAX_SLICE_PLANES 4
  949.  
  950. /// Slice plane
  951. typedef struct SwsPlane
  952. {
  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
  958. } SwsPlane;
  959.  
  960. /**
  961.  * Struct which defines a slice of an image to be scaled or a output for
  962.  * a scaled slice.
  963.  * A slice can also be used as intermediate ring buffer for scaling steps.
  964.  */
  965. typedef struct SwsSlice
  966. {
  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
  974. } SwsSlice;
  975.  
  976. /**
  977.  * Struct which holds all necessary data for processing a slice.
  978.  * A processing step can be a color conversion or horizontal/vertical scaling.
  979.  */
  980. typedef struct SwsFilterDescriptor
  981. {
  982.     SwsSlice *src;  ///< Source slice
  983.     SwsSlice *dst;  ///< Output slice
  984.  
  985.     int alpha;      ///< Flag for processing alpha channel
  986.     void *instance; ///< Filter instance data
  987.  
  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;
  991.  
  992. /// Color conversion instance data
  993. typedef struct ColorContext
  994. {
  995.     uint32_t *pal;
  996. } ColorContext;
  997.  
  998. /// Scaler instance data
  999. typedef struct FilterContext
  1000. {
  1001.     uint16_t *filter;
  1002.     int *filter_pos;
  1003.     int filter_size;
  1004.     int xInc;
  1005. } FilterContext;
  1006.  
  1007. // warp input lines in the form (src + width*i + j) to slice format (line[i][j])
  1008. // relative=true means first line src[x][0] otherwise first line is src[x][lum/crh Y]
  1009. 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);
  1010.  
  1011. // Initialize scaler filter descriptor chain
  1012. int ff_init_filters(SwsContext *c);
  1013.  
  1014. // Free all filter data
  1015. int ff_free_filters(SwsContext *c);
  1016.  
  1017. /*
  1018.  function for applying ring buffer logic into slice s
  1019.  It checks if the slice can hold more @lum lines, if yes
  1020.  do nothing otherwise remove @lum least used lines.
  1021.  It applies the same procedure for @chr lines.
  1022. */
  1023. int ff_rotate_slice(SwsSlice *s, int lum, int chr);
  1024.  
  1025. /// initializes gamma conversion descriptor
  1026. int ff_init_gamma_convert(SwsFilterDescriptor *desc, SwsSlice * src, uint16_t *table);
  1027.  
  1028. /// initializes lum pixel format conversion descriptor
  1029. int ff_init_desc_fmt_convert(SwsFilterDescriptor *desc, SwsSlice * src, SwsSlice *dst, uint32_t *pal);
  1030.  
  1031. /// initializes lum horizontal scaling descriptor
  1032. int ff_init_desc_hscale(SwsFilterDescriptor *desc, SwsSlice *src, SwsSlice *dst, uint16_t *filter, int * filter_pos, int filter_size, int xInc);
  1033.  
  1034. /// initializes chr pixel format conversion descriptor
  1035. int ff_init_desc_cfmt_convert(SwsFilterDescriptor *desc, SwsSlice * src, SwsSlice *dst, uint32_t *pal);
  1036.  
  1037. /// initializes chr horizontal scaling descriptor
  1038. int ff_init_desc_chscale(SwsFilterDescriptor *desc, SwsSlice *src, SwsSlice *dst, uint16_t *filter, int * filter_pos, int filter_size, int xInc);
  1039.  
  1040. int ff_init_desc_no_chr(SwsFilterDescriptor *desc, SwsSlice * src, SwsSlice *dst);
  1041.  
  1042. /// initializes vertical scaling descriptors
  1043. int ff_init_vscale(SwsContext *c, SwsFilterDescriptor *desc, SwsSlice *src, SwsSlice *dst);
  1044.  
  1045. /// setup vertical scaler functions
  1046. void ff_init_vscale_pfn(SwsContext *c, yuv2planar1_fn yuv2plane1, yuv2planarX_fn yuv2planeX,
  1047.     yuv2interleavedX_fn yuv2nv12cX, yuv2packed1_fn yuv2packed1, yuv2packed2_fn yuv2packed2,
  1048.     yuv2packedX_fn yuv2packedX, yuv2anyX_fn yuv2anyX, int use_mmx);
  1049.  
  1050. //number of extra lines to process
  1051. #define MAX_LINES_AHEAD 4
  1052.  
  1053. // enable use of refactored scaler code
  1054. #define NEW_FILTER
  1055.  
  1056. #endif /* SWSCALE_SWSCALE_INTERNAL_H */
  1057.