1/* 2 * VP9 compatible video decoder 3 * 4 * Copyright (C) 2013 Ronald S. Bultje <rsbultje gmail com> 5 * Copyright (C) 2013 Clément Bœsch <u pkh me> 6 * 7 * This file is part of FFmpeg. 8 * 9 * FFmpeg is free software; you can redistribute it and/or 10 * modify it under the terms of the GNU Lesser General Public 11 * License as published by the Free Software Foundation; either 12 * version 2.1 of the License, or (at your option) any later version. 13 * 14 * FFmpeg is distributed in the hope that it will be useful, 15 * but WITHOUT ANY WARRANTY; without even the implied warranty of 16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 17 * Lesser General Public License for more details. 18 * 19 * You should have received a copy of the GNU Lesser General Public 20 * License along with FFmpeg; if not, write to the Free Software 21 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 22 */ 23 24#ifndef AVCODEC_VP9DSP_H 25#define AVCODEC_VP9DSP_H 26 27#include <stddef.h> 28#include <stdint.h> 29 30#include "libavcodec/vp9.h" 31 32typedef void (*vp9_mc_func)(uint8_t *dst, ptrdiff_t dst_stride, 33 const uint8_t *ref, ptrdiff_t ref_stride, 34 int h, int mx, int my); 35typedef void (*vp9_scaled_mc_func)(uint8_t *dst, ptrdiff_t dst_stride, 36 const uint8_t *ref, ptrdiff_t ref_stride, 37 int h, int mx, int my, int dx, int dy); 38 39typedef struct VP9DSPContext { 40 /* 41 * dimension 1: 0=4x4, 1=8x8, 2=16x16, 3=32x32 42 * dimension 2: intra prediction modes 43 * 44 * dst/left/top is aligned by transform-size (i.e. 4, 8, 16 or 32 pixels) 45 * stride is aligned by 16 pixels 46 * top[-1] is top/left; top[4,7] is top-right for 4x4 47 */ 48 // FIXME(rbultje) maybe replace left/top pointers with HAVE_TOP/ 49 // HAVE_LEFT/HAVE_TOPRIGHT flags instead, and then handle it in-place? 50 // also needs to fit in with what H.264/VP8/etc do 51 void (*intra_pred[N_TXFM_SIZES][N_INTRA_PRED_MODES])(uint8_t *dst, 52 ptrdiff_t stride, 53 const uint8_t *left, 54 const uint8_t *top); 55 56 /* 57 * dimension 1: 0=4x4, 1=8x8, 2=16x16, 3=32x32, 4=lossless (3-4=dct only) 58 * dimension 2: 0=dct/dct, 1=dct/adst, 2=adst/dct, 3=adst/adst 59 * 60 * dst is aligned by transform-size (i.e. 4, 8, 16 or 32 pixels) 61 * stride is aligned by 16 pixels 62 * block is 16-byte aligned 63 * eob indicates the position (+1) of the last non-zero coefficient, 64 * in scan-order. This can be used to write faster versions, e.g. a 65 * dc-only 4x4/8x8/16x16/32x32, or a 4x4-only (eob<10) 8x8/16x16/32x32, 66 * etc. 67 */ 68 // FIXME also write idct_add_block() versions for whole (inter) pred 69 // blocks, so we can do 2 4x4s at once 70 void (*itxfm_add[N_TXFM_SIZES + 1][N_TXFM_TYPES])(uint8_t *dst, 71 ptrdiff_t stride, 72 int16_t *block, int eob); 73 74 /* 75 * dimension 1: width of filter (0=4, 1=8, 2=16) 76 * dimension 2: 0=col-edge filter (h), 1=row-edge filter (v) 77 * 78 * dst/stride are aligned by 8 79 */ 80 void (*loop_filter_8[3][2])(uint8_t *dst, ptrdiff_t stride, 81 int mb_lim, int lim, int hev_thr); 82 83 /* 84 * dimension 1: 0=col-edge filter (h), 1=row-edge filter (v) 85 * 86 * The width of filter is assumed to be 16; dst/stride are aligned by 16 87 */ 88 void (*loop_filter_16[2])(uint8_t *dst, ptrdiff_t stride, 89 int mb_lim, int lim, int hev_thr); 90 91 /* 92 * dimension 1/2: width of filter (0=4, 1=8) for each filter half 93 * dimension 3: 0=col-edge filter (h), 1=row-edge filter (v) 94 * 95 * dst/stride are aligned by operation size 96 * this basically calls loop_filter[d1][d3][0](), followed by 97 * loop_filter[d2][d3][0]() on the next 8 pixels 98 * mb_lim/lim/hev_thr contain two values in the lowest two bytes of the 99 * integer. 100 */ 101 // FIXME perhaps a mix4 that operates on 32px (for AVX2) 102 void (*loop_filter_mix2[2][2][2])(uint8_t *dst, ptrdiff_t stride, 103 int mb_lim, int lim, int hev_thr); 104 105 /* 106 * dimension 1: hsize (0: 64, 1: 32, 2: 16, 3: 8, 4: 4) 107 * dimension 2: filter type (0: smooth, 1: regular, 2: sharp, 3: bilin) 108 * dimension 3: averaging type (0: put, 1: avg) 109 * dimension 4: x subpel interpolation (0: none, 1: 8tap/bilin) 110 * dimension 5: y subpel interpolation (0: none, 1: 8tap/bilin) 111 * 112 * dst/stride are aligned by hsize 113 */ 114 vp9_mc_func mc[5][N_FILTERS][2][2][2]; 115 116 /* 117 * for scalable MC, first 3 dimensions identical to above, the other two 118 * don't exist since it changes per stepsize. 119 */ 120 vp9_scaled_mc_func smc[5][N_FILTERS][2]; 121} VP9DSPContext; 122 123extern const int16_t ff_vp9_subpel_filters[3][16][8]; 124 125void ff_vp9dsp_init(VP9DSPContext *dsp, int bpp, int bitexact); 126 127void ff_vp9dsp_init_8(VP9DSPContext *dsp); 128void ff_vp9dsp_init_10(VP9DSPContext *dsp); 129void ff_vp9dsp_init_12(VP9DSPContext *dsp); 130 131void ff_vp9dsp_init_aarch64(VP9DSPContext *dsp, int bpp); 132void ff_vp9dsp_init_arm(VP9DSPContext *dsp, int bpp); 133void ff_vp9dsp_init_x86(VP9DSPContext *dsp, int bpp, int bitexact); 134void ff_vp9dsp_init_mips(VP9DSPContext *dsp, int bpp); 135void ff_vp9dsp_init_loongarch(VP9DSPContext *dsp, int bpp); 136 137#endif /* AVCODEC_VP9DSP_H */ 138