Unified · Split
| @@ -1,7 +1,8 @@ | ||
| 1 | 1 | clean:: |
| 2 | 2 | $(RM) $(CLEANSUFFIXES:%=libavcodec/vvc/%) |
| 3 | 3 | |
| 4 | −OBJS-$(CONFIG_VVC_DECODER) += vvc/vvc_cabac.o \ | |
| 4 | +OBJS-$(CONFIG_VVC_DECODER) += vvc/vvcdsp.o \ | |
| 5 | + vvc/vvc_cabac.o \ | |
| 5 | 6 | vvc/vvc_ctu.o \ |
| 6 | 7 | vvc/vvc_data.o \ |
| 7 | 8 | vvc/vvc_filter.o \ |
| @@ -0,0 +1,141 @@ | ||
| 1 | +/* | |
| 2 | + * VVC DSP | |
| 3 | + * | |
| 4 | + * Copyright (C) 2021 Nuo Mi | |
| 5 | + * | |
| 6 | + * This file is part of FFmpeg. | |
| 7 | + * | |
| 8 | + * FFmpeg is free software; you can redistribute it and/or | |
| 9 | + * modify it under the terms of the GNU Lesser General Public | |
| 10 | + * License as published by the Free Software Foundation; either | |
| 11 | + * version 2.1 of the License, or (at your option) any later version. | |
| 12 | + * | |
| 13 | + * FFmpeg is distributed in the hope that it will be useful, | |
| 14 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 15 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 16 | + * Lesser General Public License for more details. | |
| 17 | + * | |
| 18 | + * You should have received a copy of the GNU Lesser General Public | |
| 19 | + * License along with FFmpeg; if not, write to the Free Software | |
| 20 | + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
| 21 | + */ | |
| 22 | + | |
| 23 | +#include "vvcdsp.h" | |
| 24 | +#include "vvc_ctu.h" | |
| 25 | +#include "vvc_itx_1d.h" | |
| 26 | + | |
| 27 | +#define VVC_SIGN(v) (v < 0 ? -1 : !!v) | |
| 28 | + | |
| 29 | +static void av_always_inline pad_int16(int16_t *_dst, const ptrdiff_t dst_stride, const int width, const int height) | |
| 30 | +{ | |
| 31 | + const int padded_width = width + 2; | |
| 32 | + int16_t *dst; | |
| 33 | + for (int y = 0; y < height; y++) { | |
| 34 | + dst = _dst + y * dst_stride; | |
| 35 | + for (int x = 0; x < width; x++) { | |
| 36 | + dst[-1] = dst[0]; | |
| 37 | + dst[width] = dst[width - 1]; | |
| 38 | + } | |
| 39 | + } | |
| 40 | + | |
| 41 | + _dst--; | |
| 42 | + //top | |
| 43 | + memcpy(_dst - dst_stride, _dst, padded_width * sizeof(int16_t)); | |
| 44 | + //bottom | |
| 45 | + _dst += dst_stride * height; | |
| 46 | + memcpy(_dst, _dst - dst_stride, padded_width * sizeof(int16_t)); | |
| 47 | +} | |
| 48 | + | |
| 49 | +static int vvc_sad(const int16_t *src0, const int16_t *src1, int dx, int dy, | |
| 50 | + const int block_w, const int block_h) | |
| 51 | +{ | |
| 52 | + int sad = 0; | |
| 53 | + dx -= 2; | |
| 54 | + dy -= 2; | |
| 55 | + src0 += (2 + dy) * MAX_PB_SIZE + 2 + dx; | |
| 56 | + src1 += (2 - dy) * MAX_PB_SIZE + 2 - dx; | |
| 57 | + for (int y = 0; y < block_h; y += 2) { | |
| 58 | + for (int x = 0; x < block_w; x++) { | |
| 59 | + sad += FFABS(src0[x] - src1[x]); | |
| 60 | + } | |
| 61 | + src0 += 2 * MAX_PB_SIZE; | |
| 62 | + src1 += 2 * MAX_PB_SIZE; | |
| 63 | + } | |
| 64 | + return sad; | |
| 65 | +} | |
| 66 | + | |
| 67 | +#define itx_fn(type, s) \ | |
| 68 | +static void itx_##type##_##s(int *coeffs, ptrdiff_t step, size_t nz) \ | |
| 69 | +{ \ | |
| 70 | + ff_vvc_inv_##type##_##s(coeffs, step, nz); \ | |
| 71 | +} | |
| 72 | + | |
| 73 | +#define itx_fn_common(type) \ | |
| 74 | + itx_fn(type, 4); \ | |
| 75 | + itx_fn(type, 8); \ | |
| 76 | + itx_fn(type, 16); \ | |
| 77 | + itx_fn(type, 32); \ | |
| 78 | + | |
| 79 | +itx_fn_common(dct2); | |
| 80 | +itx_fn_common(dst7); | |
| 81 | +itx_fn_common(dct8); | |
| 82 | +itx_fn(dct2, 2); | |
| 83 | +itx_fn(dct2, 64); | |
| 84 | + | |
| 85 | +typedef struct IntraEdgeParams { | |
| 86 | + uint8_t* top; | |
| 87 | + uint8_t* left; | |
| 88 | + int filter_flag; | |
| 89 | + | |
| 90 | + uint16_t left_array[3 * MAX_TB_SIZE + 3]; | |
| 91 | + uint16_t filtered_left_array[3 * MAX_TB_SIZE + 3]; | |
| 92 | + uint16_t top_array[3 * MAX_TB_SIZE + 3]; | |
| 93 | + uint16_t filtered_top_array[3 * MAX_TB_SIZE + 3]; | |
| 94 | +} IntraEdgeParams; | |
| 95 | + | |
| 96 | +#define PROF_BORDER_EXT 1 | |
| 97 | +#define PROF_BLOCK_SIZE (AFFINE_MIN_BLOCK_SIZE + PROF_BORDER_EXT * 2) | |
| 98 | +#define BDOF_BORDER_EXT 1 | |
| 99 | + | |
| 100 | +#define BDOF_PADDED_SIZE (16 + BDOF_BORDER_EXT * 2) | |
| 101 | +#define BDOF_BLOCK_SIZE 4 | |
| 102 | +#define BDOF_GRADIENT_SIZE (BDOF_BLOCK_SIZE + BDOF_BORDER_EXT * 2) | |
| 103 | + | |
| 104 | +#define BIT_DEPTH 8 | |
| 105 | +#include "vvcdsp_template.c" | |
| 106 | +#undef BIT_DEPTH | |
| 107 | + | |
| 108 | +#define BIT_DEPTH 10 | |
| 109 | +#include "vvcdsp_template.c" | |
| 110 | +#undef BIT_DEPTH | |
| 111 | + | |
| 112 | +#define BIT_DEPTH 12 | |
| 113 | +#include "vvcdsp_template.c" | |
| 114 | +#undef BIT_DEPTH | |
| 115 | + | |
| 116 | +void ff_vvc_dsp_init(VVCDSPContext *vvcdsp, int bit_depth) | |
| 117 | +{ | |
| 118 | +#undef FUNC | |
| 119 | +#define FUNC(a, depth) a ## _ ## depth | |
| 120 | + | |
| 121 | +#define VVC_DSP(depth) \ | |
| 122 | + FUNC(ff_vvc_inter_dsp_init, depth)(&vvcdsp->inter); \ | |
| 123 | + FUNC(ff_vvc_intra_dsp_init, depth)(&vvcdsp->intra); \ | |
| 124 | + FUNC(ff_vvc_itx_dsp_init, depth)(&vvcdsp->itx); \ | |
| 125 | + FUNC(ff_vvc_lmcs_dsp_init, depth)(&vvcdsp->lmcs); \ | |
| 126 | + FUNC(ff_vvc_lf_dsp_init, depth)(&vvcdsp->lf); \ | |
| 127 | + FUNC(ff_vvc_sao_dsp_init, depth)(&vvcdsp->sao); \ | |
| 128 | + FUNC(ff_vvc_alf_dsp_init, depth)(&vvcdsp->alf); \ | |
| 129 | + | |
| 130 | + switch (bit_depth) { | |
| 131 | + case 12: | |
| 132 | + VVC_DSP(12); | |
| 133 | + break; | |
| 134 | + case 10: | |
| 135 | + VVC_DSP(10); | |
| 136 | + break; | |
| 137 | + default: | |
| 138 | + VVC_DSP(8); | |
| 139 | + break; | |
| 140 | + } | |
| 141 | +} | |
| @@ -0,0 +1,120 @@ | ||
| 1 | +/* | |
| 2 | + * VVC transform and residual DSP | |
| 3 | + * | |
| 4 | + * Copyright (C) 2021 Nuo Mi | |
| 5 | + * | |
| 6 | + * This file is part of FFmpeg. | |
| 7 | + * | |
| 8 | + * FFmpeg is free software; you can redistribute it and/or | |
| 9 | + * modify it under the terms of the GNU Lesser General Public | |
| 10 | + * License as published by the Free Software Foundation; either | |
| 11 | + * version 2.1 of the License, or (at your option) any later version. | |
| 12 | + * | |
| 13 | + * FFmpeg is distributed in the hope that it will be useful, | |
| 14 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 15 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 16 | + * Lesser General Public License for more details. | |
| 17 | + * | |
| 18 | + * You should have received a copy of the GNU Lesser General Public | |
| 19 | + * License along with FFmpeg; if not, write to the Free Software | |
| 20 | + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
| 21 | + */ | |
| 22 | +#include "libavutil/frame.h" | |
| 23 | +#include "libavcodec/bit_depth_template.c" | |
| 24 | + | |
| 25 | +#include "vvcdec.h" | |
| 26 | +#include "vvc_data.h" | |
| 27 | + | |
| 28 | +#include "vvc_inter_template.c" | |
| 29 | +#include "vvc_intra_template.c" | |
| 30 | +#include "vvc_filter_template.c" | |
| 31 | + | |
| 32 | +static void FUNC(add_residual)(uint8_t *_dst, const int *res, | |
| 33 | + const int w, const int h, const ptrdiff_t _stride) | |
| 34 | +{ | |
| 35 | + pixel *dst = (pixel *)_dst; | |
| 36 | + | |
| 37 | + const int stride = _stride / sizeof(pixel); | |
| 38 | + | |
| 39 | + for (int y = 0; y < h; y++) { | |
| 40 | + for (int x = 0; x < w; x++) { | |
| 41 | + dst[x] = av_clip_pixel(dst[x] + *res); | |
| 42 | + res++; | |
| 43 | + } | |
| 44 | + dst += stride; | |
| 45 | + } | |
| 46 | +} | |
| 47 | + | |
| 48 | +static void FUNC(add_residual_joint)(uint8_t *_dst, const int *res, | |
| 49 | + const int w, const int h, const ptrdiff_t _stride, const int c_sign, const int shift) | |
| 50 | +{ | |
| 51 | + pixel *dst = (pixel *)_dst; | |
| 52 | + | |
| 53 | + const int stride = _stride / sizeof(pixel); | |
| 54 | + | |
| 55 | + for (int y = 0; y < h; y++) { | |
| 56 | + for (int x = 0; x < w; x++) { | |
| 57 | + const int r = ((*res) * c_sign) >> shift; | |
| 58 | + dst[x] = av_clip_pixel(dst[x] + r); | |
| 59 | + res++; | |
| 60 | + } | |
| 61 | + dst += stride; | |
| 62 | + } | |
| 63 | +} | |
| 64 | + | |
| 65 | +static void FUNC(pred_residual_joint)(int *buf, const int w, const int h, | |
| 66 | + const int c_sign, const int shift) | |
| 67 | +{ | |
| 68 | + for (int y = 0; y < h; y++) { | |
| 69 | + for (int x = 0; x < w; x++) { | |
| 70 | + *buf = ((*buf) * c_sign) >> shift; | |
| 71 | + buf++; | |
| 72 | + } | |
| 73 | + } | |
| 74 | +} | |
| 75 | + | |
| 76 | +static void FUNC(transform_bdpcm)(int *coeffs, const int width, const int height, | |
| 77 | + const int vertical, const int log2_transform_range) | |
| 78 | +{ | |
| 79 | + int x, y; | |
| 80 | + | |
| 81 | + if (vertical) { | |
| 82 | + coeffs += width; | |
| 83 | + for (y = 0; y < height - 1; y++) { | |
| 84 | + for (x = 0; x < width; x++) | |
| 85 | + coeffs[x] = av_clip_intp2(coeffs[x] + coeffs[x - width], log2_transform_range); | |
| 86 | + coeffs += width; | |
| 87 | + } | |
| 88 | + } else { | |
| 89 | + for (y = 0; y < height; y++) { | |
| 90 | + for (x = 1; x < width; x++) | |
| 91 | + coeffs[x] = av_clip_intp2(coeffs[x] + coeffs[x - 1], log2_transform_range); | |
| 92 | + coeffs += width; | |
| 93 | + } | |
| 94 | + } | |
| 95 | +} | |
| 96 | + | |
| 97 | +static void FUNC(ff_vvc_itx_dsp_init)(VVCItxDSPContext *const itx) | |
| 98 | +{ | |
| 99 | +#define VVC_ITX(TYPE, type, s) \ | |
| 100 | + itx->itx[TYPE][TX_SIZE_##s] = itx_##type##_##s; \ | |
| 101 | + | |
| 102 | +#define VVC_ITX_COMMON(TYPE, type) \ | |
| 103 | + VVC_ITX(TYPE, type, 4); \ | |
| 104 | + VVC_ITX(TYPE, type, 8); \ | |
| 105 | + VVC_ITX(TYPE, type, 16); \ | |
| 106 | + VVC_ITX(TYPE, type, 32); | |
| 107 | + | |
| 108 | + itx->add_residual = FUNC(add_residual); | |
| 109 | + itx->add_residual_joint = FUNC(add_residual_joint); | |
| 110 | + itx->pred_residual_joint = FUNC(pred_residual_joint); | |
| 111 | + itx->transform_bdpcm = FUNC(transform_bdpcm); | |
| 112 | + VVC_ITX(DCT2, dct2, 2) | |
| 113 | + VVC_ITX(DCT2, dct2, 64) | |
| 114 | + VVC_ITX_COMMON(DCT2, dct2) | |
| 115 | + VVC_ITX_COMMON(DCT8, dct8) | |
| 116 | + VVC_ITX_COMMON(DST7, dst7) | |
| 117 | + | |
| 118 | +#undef VVC_ITX | |
| 119 | +#undef VVC_ITX_COMMON | |
| 120 | +} | |