diff options
Diffstat (limited to 'libavcodec/hevcdsp_template.c')
-rw-r--r-- | libavcodec/hevcdsp_template.c | 1849 |
1 files changed, 1124 insertions, 725 deletions
diff --git a/libavcodec/hevcdsp_template.c b/libavcodec/hevcdsp_template.c index 8dcc83d..8a843f5 100644 --- a/libavcodec/hevcdsp_template.c +++ b/libavcodec/hevcdsp_template.c @@ -3,20 +3,20 @@ * * Copyright (C) 2012 - 2013 Guillaume Martres * - * This file is part of Libav. + * This file is part of FFmpeg. * - * Libav is free software; you can redistribute it and/or + * FFmpeg is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * - * Libav is distributed in the hope that it will be useful, + * FFmpeg is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public - * License along with Libav; if not, write to the Free Software + * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ @@ -24,8 +24,10 @@ #include "hevc.h" #include "bit_depth_template.c" +#include "hevcdsp.h" -static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int size, + +static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int width, int height, GetBitContext *gb, int pcm_bit_depth) { int x, y; @@ -33,14 +35,14 @@ static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int size, stride /= sizeof(pixel); - for (y = 0; y < size; y++) { - for (x = 0; x < size; x++) + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) dst[x] = get_bits(gb, pcm_bit_depth) << (BIT_DEPTH - pcm_bit_depth); dst += stride; } } -static void FUNC(transquant_bypass4x4)(uint8_t *_dst, int16_t *coeffs, +static void FUNC(transform_add4x4)(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride) { int x, y; @@ -57,7 +59,7 @@ static void FUNC(transquant_bypass4x4)(uint8_t *_dst, int16_t *coeffs, } } -static void FUNC(transquant_bypass8x8)(uint8_t *_dst, int16_t *coeffs, +static void FUNC(transform_add8x8)(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride) { int x, y; @@ -74,7 +76,7 @@ static void FUNC(transquant_bypass8x8)(uint8_t *_dst, int16_t *coeffs, } } -static void FUNC(transquant_bypass16x16)(uint8_t *_dst, int16_t *coeffs, +static void FUNC(transform_add16x16)(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride) { int x, y; @@ -91,7 +93,7 @@ static void FUNC(transquant_bypass16x16)(uint8_t *_dst, int16_t *coeffs, } } -static void FUNC(transquant_bypass32x32)(uint8_t *_dst, int16_t *coeffs, +static void FUNC(transform_add32x32)(uint8_t *_dst, int16_t *coeffs, ptrdiff_t stride) { int x, y; @@ -108,24 +110,52 @@ static void FUNC(transquant_bypass32x32)(uint8_t *_dst, int16_t *coeffs, } } -static void FUNC(transform_skip)(uint8_t *_dst, int16_t *coeffs, - ptrdiff_t stride) + +static void FUNC(transform_rdpcm)(int16_t *_coeffs, int16_t log2_size, int mode) { - pixel *dst = (pixel *)_dst; - int shift = 13 - BIT_DEPTH; -#if BIT_DEPTH <= 13 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif + int16_t *coeffs = (int16_t *) _coeffs; + int x, y; + int size = 1 << log2_size; + + if (mode) { + coeffs += size; + for (y = 0; y < size - 1; y++) { + for (x = 0; x < size; x++) + coeffs[x] += coeffs[x - size]; + coeffs += size; + } + } else { + for (y = 0; y < size; y++) { + for (x = 1; x < size; x++) + coeffs[x] += coeffs[x - 1]; + coeffs += size; + } + } +} + +static void FUNC(transform_skip)(int16_t *_coeffs, int16_t log2_size) +{ + int shift = 15 - BIT_DEPTH - log2_size; int x, y; + int size = 1 << log2_size; + int16_t *coeffs = _coeffs; - stride /= sizeof(pixel); - for (y = 0; y < 4 * 4; y += 4) { - for (x = 0; x < 4; x++) - dst[x] = av_clip_pixel(dst[x] + ((coeffs[y + x] + offset) >> shift)); - dst += stride; + if (shift > 0) { + int offset = 1 << (shift - 1); + for (y = 0; y < size; y++) { + for (x = 0; x < size; x++) { + *coeffs = (*coeffs + offset) >> shift; + coeffs++; + } + } + } else { + for (y = 0; y < size; y++) { + for (x = 0; x < size; x++) { + *coeffs = *coeffs << -shift; + coeffs++; + } + } } } @@ -149,17 +179,13 @@ static void FUNC(transform_skip)(uint8_t *_dst, int16_t *coeffs, assign(dst[3 * step], 55 * c0 + 29 * c2 - c3); \ } while (0) -static void FUNC(transform_4x4_luma_add)(uint8_t *_dst, int16_t *coeffs, - ptrdiff_t stride) +static void FUNC(transform_4x4_luma)(int16_t *coeffs) { int i; - pixel *dst = (pixel *)_dst; int shift = 7; int add = 1 << (shift - 1); int16_t *src = coeffs; - stride /= sizeof(pixel); - for (i = 0; i < 4; i++) { TR_4x4_LUMA(src, src, 4, SCALE); src++; @@ -168,323 +194,238 @@ static void FUNC(transform_4x4_luma_add)(uint8_t *_dst, int16_t *coeffs, shift = 20 - BIT_DEPTH; add = 1 << (shift - 1); for (i = 0; i < 4; i++) { - TR_4x4_LUMA(dst, coeffs, 1, ADD_AND_SCALE); + TR_4x4_LUMA(coeffs, coeffs, 1, SCALE); coeffs += 4; - dst += stride; } } #undef TR_4x4_LUMA -#define TR_4(dst, src, dstep, sstep, assign) \ - do { \ - const int e0 = transform[8 * 0][0] * src[0 * sstep] + \ - transform[8 * 2][0] * src[2 * sstep]; \ - const int e1 = transform[8 * 0][1] * src[0 * sstep] + \ - transform[8 * 2][1] * src[2 * sstep]; \ - const int o0 = transform[8 * 1][0] * src[1 * sstep] + \ - transform[8 * 3][0] * src[3 * sstep]; \ - const int o1 = transform[8 * 1][1] * src[1 * sstep] + \ - transform[8 * 3][1] * src[3 * sstep]; \ - \ - assign(dst[0 * dstep], e0 + o0); \ - assign(dst[1 * dstep], e1 + o1); \ - assign(dst[2 * dstep], e1 - o1); \ - assign(dst[3 * dstep], e0 - o0); \ +#define TR_4(dst, src, dstep, sstep, assign, end) \ + do { \ + const int e0 = 64 * src[0 * sstep] + 64 * src[2 * sstep]; \ + const int e1 = 64 * src[0 * sstep] - 64 * src[2 * sstep]; \ + const int o0 = 83 * src[1 * sstep] + 36 * src[3 * sstep]; \ + const int o1 = 36 * src[1 * sstep] - 83 * src[3 * sstep]; \ + \ + assign(dst[0 * dstep], e0 + o0); \ + assign(dst[1 * dstep], e1 + o1); \ + assign(dst[2 * dstep], e1 - o1); \ + assign(dst[3 * dstep], e0 - o0); \ } while (0) -static void FUNC(transform_4x4_add)(uint8_t *_dst, int16_t *coeffs, - ptrdiff_t stride) -{ - int i; - pixel *dst = (pixel *)_dst; - int shift = 7; - int add = 1 << (shift - 1); - int16_t *src = coeffs; - - stride /= sizeof(pixel); - - for (i = 0; i < 4; i++) { - TR_4(src, src, 4, 4, SCALE); - src++; - } - - shift = 20 - BIT_DEPTH; - add = 1 << (shift - 1); - for (i = 0; i < 4; i++) { - TR_4(dst, coeffs, 1, 1, ADD_AND_SCALE); - coeffs += 4; - dst += stride; - } -} - -#define TR_8(dst, src, dstep, sstep, assign) \ - do { \ - int i, j; \ - int e_8[4]; \ - int o_8[4] = { 0 }; \ - for (i = 0; i < 4; i++) \ - for (j = 1; j < 8; j += 2) \ - o_8[i] += transform[4 * j][i] * src[j * sstep]; \ - TR_4(e_8, src, 1, 2 * sstep, SET); \ - \ - for (i = 0; i < 4; i++) { \ - assign(dst[i * dstep], e_8[i] + o_8[i]); \ - assign(dst[(7 - i) * dstep], e_8[i] - o_8[i]); \ - } \ +#define TR_8(dst, src, dstep, sstep, assign, end) \ + do { \ + int i, j; \ + int e_8[4]; \ + int o_8[4] = { 0 }; \ + for (i = 0; i < 4; i++) \ + for (j = 1; j < end; j += 2) \ + o_8[i] += transform[4 * j][i] * src[j * sstep]; \ + TR_4(e_8, src, 1, 2 * sstep, SET, 4); \ + \ + for (i = 0; i < 4; i++) { \ + assign(dst[i * dstep], e_8[i] + o_8[i]); \ + assign(dst[(7 - i) * dstep], e_8[i] - o_8[i]); \ + } \ } while (0) -#define TR_16(dst, src, dstep, sstep, assign) \ - do { \ - int i, j; \ - int e_16[8]; \ - int o_16[8] = { 0 }; \ - for (i = 0; i < 8; i++) \ - for (j = 1; j < 16; j += 2) \ - o_16[i] += transform[2 * j][i] * src[j * sstep]; \ - TR_8(e_16, src, 1, 2 * sstep, SET); \ - \ - for (i = 0; i < 8; i++) { \ - assign(dst[i * dstep], e_16[i] + o_16[i]); \ - assign(dst[(15 - i) * dstep], e_16[i] - o_16[i]); \ - } \ +#define TR_16(dst, src, dstep, sstep, assign, end) \ + do { \ + int i, j; \ + int e_16[8]; \ + int o_16[8] = { 0 }; \ + for (i = 0; i < 8; i++) \ + for (j = 1; j < end; j += 2) \ + o_16[i] += transform[2 * j][i] * src[j * sstep]; \ + TR_8(e_16, src, 1, 2 * sstep, SET, 8); \ + \ + for (i = 0; i < 8; i++) { \ + assign(dst[i * dstep], e_16[i] + o_16[i]); \ + assign(dst[(15 - i) * dstep], e_16[i] - o_16[i]); \ + } \ } while (0) -#define TR_32(dst, src, dstep, sstep, assign) \ - do { \ - int i, j; \ - int e_32[16]; \ - int o_32[16] = { 0 }; \ - for (i = 0; i < 16; i++) \ - for (j = 1; j < 32; j += 2) \ - o_32[i] += transform[j][i] * src[j * sstep]; \ - TR_16(e_32, src, 1, 2 * sstep, SET); \ - \ - for (i = 0; i < 16; i++) { \ - assign(dst[i * dstep], e_32[i] + o_32[i]); \ - assign(dst[(31 - i) * dstep], e_32[i] - o_32[i]); \ - } \ +#define TR_32(dst, src, dstep, sstep, assign, end) \ + do { \ + int i, j; \ + int e_32[16]; \ + int o_32[16] = { 0 }; \ + for (i = 0; i < 16; i++) \ + for (j = 1; j < end; j += 2) \ + o_32[i] += transform[j][i] * src[j * sstep]; \ + TR_16(e_32, src, 1, 2 * sstep, SET, end/2); \ + \ + for (i = 0; i < 16; i++) { \ + assign(dst[i * dstep], e_32[i] + o_32[i]); \ + assign(dst[(31 - i) * dstep], e_32[i] - o_32[i]); \ + } \ } while (0) - - -static void FUNC(transform_8x8_add)(uint8_t *_dst, int16_t *coeffs, - ptrdiff_t stride) -{ - int i; - pixel *dst = (pixel *)_dst; - int shift = 7; - int add = 1 << (shift - 1); - int16_t *src = coeffs; - - stride /= sizeof(pixel); - - for (i = 0; i < 8; i++) { - TR_8(src, src, 8, 8, SCALE); - src++; - } - - shift = 20 - BIT_DEPTH; - add = 1 << (shift - 1); - for (i = 0; i < 8; i++) { - TR_8(dst, coeffs, 1, 1, ADD_AND_SCALE); - coeffs += 8; - dst += stride; - } +#define IDCT_VAR4(H) \ + int limit2 = FFMIN(col_limit + 4, H) +#define IDCT_VAR8(H) \ + int limit = FFMIN(col_limit, H); \ + int limit2 = FFMIN(col_limit + 4, H) +#define IDCT_VAR16(H) IDCT_VAR8(H) +#define IDCT_VAR32(H) IDCT_VAR8(H) + +#define IDCT(H) \ +static void FUNC(idct_##H ##x ##H )( \ + int16_t *coeffs, int col_limit) { \ + int i; \ + int shift = 7; \ + int add = 1 << (shift - 1); \ + int16_t *src = coeffs; \ + IDCT_VAR ##H(H); \ + \ + for (i = 0; i < H; i++) { \ + TR_ ## H(src, src, H, H, SCALE, limit2); \ + if (limit2 < H && i%4 == 0 && !!i) \ + limit2 -= 4; \ + src++; \ + } \ + \ + shift = 20 - BIT_DEPTH; \ + add = 1 << (shift - 1); \ + for (i = 0; i < H; i++) { \ + TR_ ## H(coeffs, coeffs, 1, 1, SCALE, limit); \ + coeffs += H; \ + } \ } -static void FUNC(transform_16x16_add)(uint8_t *_dst, int16_t *coeffs, - ptrdiff_t stride) -{ - int i; - pixel *dst = (pixel *)_dst; - int shift = 7; - int add = 1 << (shift - 1); - int16_t *src = coeffs; - - stride /= sizeof(pixel); - - for (i = 0; i < 16; i++) { - TR_16(src, src, 16, 16, SCALE); - src++; - } - - shift = 20 - BIT_DEPTH; - add = 1 << (shift - 1); - for (i = 0; i < 16; i++) { - TR_16(dst, coeffs, 1, 1, ADD_AND_SCALE); - coeffs += 16; - dst += stride; - } +#define IDCT_DC(H) \ +static void FUNC(idct_##H ##x ##H ##_dc)( \ + int16_t *coeffs) { \ + int i, j; \ + int shift = 14 - BIT_DEPTH; \ + int add = 1 << (shift - 1); \ + int coeff = (((coeffs[0] + 1) >> 1) + add) >> shift; \ + \ + for (j = 0; j < H; j++) { \ + for (i = 0; i < H; i++) { \ + coeffs[i+j*H] = coeff; \ + } \ + } \ } -static void FUNC(transform_32x32_add)(uint8_t *_dst, int16_t *coeffs, - ptrdiff_t stride) -{ - int i; - pixel *dst = (pixel *)_dst; - int shift = 7; - int add = 1 << (shift - 1); - int16_t *src = coeffs; +IDCT( 4) +IDCT( 8) +IDCT(16) +IDCT(32) - stride /= sizeof(pixel); +IDCT_DC( 4) +IDCT_DC( 8) +IDCT_DC(16) +IDCT_DC(32) - for (i = 0; i < 32; i++) { - TR_32(src, src, 32, 32, SCALE); - src++; - } - src = coeffs; - shift = 20 - BIT_DEPTH; - add = 1 << (shift - 1); - for (i = 0; i < 32; i++) { - TR_32(dst, coeffs, 1, 1, ADD_AND_SCALE); - coeffs += 32; - dst += stride; - } -} +#undef TR_4 +#undef TR_8 +#undef TR_16 +#undef TR_32 -static void FUNC(sao_band_filter)(uint8_t *_dst, uint8_t *_src, - ptrdiff_t stride, SAOParams *sao, +#undef SET +#undef SCALE +#undef ADD_AND_SCALE + +static void FUNC(sao_band_filter_0)(uint8_t *_dst, uint8_t *_src, + ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao, int *borders, int width, int height, - int c_idx, int class) + int c_idx) { pixel *dst = (pixel *)_dst; pixel *src = (pixel *)_src; int offset_table[32] = { 0 }; int k, y, x; - int chroma = !!c_idx; int shift = BIT_DEPTH - 5; - int *sao_offset_val = sao->offset_val[c_idx]; + int16_t *sao_offset_val = sao->offset_val[c_idx]; int sao_left_class = sao->band_position[c_idx]; - int init_y = 0, init_x = 0; - - stride /= sizeof(pixel); - switch (class) { - case 0: - if (!borders[2]) - width -= (8 >> chroma) + 2; - if (!borders[3]) - height -= (4 >> chroma) + 2; - break; - case 1: - init_y = -(4 >> chroma) - 2; - if (!borders[2]) - width -= (8 >> chroma) + 2; - height = (4 >> chroma) + 2; - break; - case 2: - init_x = -(8 >> chroma) - 2; - width = (8 >> chroma) + 2; - if (!borders[3]) - height -= (4 >> chroma) + 2; - break; - case 3: - init_y = -(4 >> chroma) - 2; - init_x = -(8 >> chroma) - 2; - width = (8 >> chroma) + 2; - height = (4 >> chroma) + 2; - break; - } + stride_dst /= sizeof(pixel); + stride_src /= sizeof(pixel); - dst = dst + (init_y * stride + init_x); - src = src + (init_y * stride + init_x); for (k = 0; k < 4; k++) offset_table[(k + sao_left_class) & 31] = sao_offset_val[k + 1]; for (y = 0; y < height; y++) { for (x = 0; x < width; x++) dst[x] = av_clip_pixel(src[x] + offset_table[src[x] >> shift]); - dst += stride; - src += stride; + dst += stride_dst; + src += stride_src; } } -static void FUNC(sao_band_filter_0)(uint8_t *dst, uint8_t *src, - ptrdiff_t stride, SAOParams *sao, - int *borders, int width, int height, - int c_idx) -{ - FUNC(sao_band_filter)(dst, src, stride, sao, borders, - width, height, c_idx, 0); -} +#define CMP(a, b) ((a) > (b) ? 1 : ((a) == (b) ? 0 : -1)) -static void FUNC(sao_band_filter_1)(uint8_t *dst, uint8_t *src, - ptrdiff_t stride, SAOParams *sao, - int *borders, int width, int height, - int c_idx) -{ - FUNC(sao_band_filter)(dst, src, stride, sao, borders, - width, height, c_idx, 1); -} +static void FUNC(sao_edge_filter)(uint8_t *_dst, uint8_t *_src, + ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao, + int width, int height, + int c_idx, int init_x, int init_y) { -static void FUNC(sao_band_filter_2)(uint8_t *dst, uint8_t *src, - ptrdiff_t stride, SAOParams *sao, - int *borders, int width, int height, - int c_idx) -{ - FUNC(sao_band_filter)(dst, src, stride, sao, borders, - width, height, c_idx, 2); -} + static const uint8_t edge_idx[] = { 1, 2, 0, 3, 4 }; + static const int8_t pos[4][2][2] = { + { { -1, 0 }, { 1, 0 } }, // horizontal + { { 0, -1 }, { 0, 1 } }, // vertical + { { -1, -1 }, { 1, 1 } }, // 45 degree + { { 1, -1 }, { -1, 1 } }, // 135 degree + }; + int16_t *sao_offset_val = sao->offset_val[c_idx]; + int sao_eo_class = sao->eo_class[c_idx]; + pixel *dst = (pixel *)_dst; + pixel *src = (pixel *)_src; -static void FUNC(sao_band_filter_3)(uint8_t *_dst, uint8_t *_src, - ptrdiff_t stride, SAOParams *sao, - int *borders, int width, int height, - int c_idx) -{ - FUNC(sao_band_filter)(_dst, _src, stride, sao, borders, - width, height, c_idx, 3); + int y_stride_src = init_y * stride_src; + int y_stride_dst = init_y * stride_dst; + int pos_0_0 = pos[sao_eo_class][0][0]; + int pos_0_1 = pos[sao_eo_class][0][1]; + int pos_1_0 = pos[sao_eo_class][1][0]; + int pos_1_1 = pos[sao_eo_class][1][1]; + int x, y; + + int y_stride_0_1 = (init_y + pos_0_1) * stride_src; + int y_stride_1_1 = (init_y + pos_1_1) * stride_src; + for (y = init_y; y < height; y++) { + for (x = init_x; x < width; x++) { + int diff0 = CMP(src[x + y_stride_src], src[x + pos_0_0 + y_stride_0_1]); + int diff1 = CMP(src[x + y_stride_src], src[x + pos_1_0 + y_stride_1_1]); + int offset_val = edge_idx[2 + diff0 + diff1]; + dst[x + y_stride_dst] = av_clip_pixel(src[x + y_stride_src] + sao_offset_val[offset_val]); + } + y_stride_src += stride_src; + y_stride_dst += stride_dst; + y_stride_0_1 += stride_src; + y_stride_1_1 += stride_src; + } } static void FUNC(sao_edge_filter_0)(uint8_t *_dst, uint8_t *_src, - ptrdiff_t stride, SAOParams *sao, + ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao, int *borders, int _width, int _height, - int c_idx, uint8_t vert_edge, - uint8_t horiz_edge, uint8_t diag_edge) + int c_idx, uint8_t *vert_edge, + uint8_t *horiz_edge, uint8_t *diag_edge) { int x, y; pixel *dst = (pixel *)_dst; pixel *src = (pixel *)_src; - int chroma = !!c_idx; - int *sao_offset_val = sao->offset_val[c_idx]; + int16_t *sao_offset_val = sao->offset_val[c_idx]; int sao_eo_class = sao->eo_class[c_idx]; int init_x = 0, init_y = 0, width = _width, height = _height; - static const int8_t pos[4][2][2] = { - { { -1, 0 }, { 1, 0 } }, // horizontal - { { 0, -1 }, { 0, 1 } }, // vertical - { { -1, -1 }, { 1, 1 } }, // 45 degree - { { 1, -1 }, { -1, 1 } }, // 135 degree - }; - static const uint8_t edge_idx[] = { 1, 2, 0, 3, 4 }; - -#define CMP(a, b) ((a) > (b) ? 1 : ((a) == (b) ? 0 : -1)) - - stride /= sizeof(pixel); - - if (!borders[2]) - width -= (8 >> chroma) + 2; - if (!borders[3]) - height -= (4 >> chroma) + 2; + stride_dst /= sizeof(pixel); + stride_src /= sizeof(pixel); - dst = dst + (init_y * stride + init_x); - src = src + (init_y * stride + init_x); - init_y = init_x = 0; if (sao_eo_class != SAO_EO_VERT) { if (borders[0]) { int offset_val = sao_offset_val[0]; - int y_stride = 0; for (y = 0; y < height; y++) { - dst[y_stride] = av_clip_pixel(src[y_stride] + offset_val); - y_stride += stride; + dst[y * stride_dst] = av_clip_pixel(src[y * stride_src] + offset_val); } init_x = 1; } if (borders[2]) { int offset_val = sao_offset_val[0]; - int x_stride = width - 1; + int offset = width - 1; for (x = 0; x < height; x++) { - dst[x_stride] = av_clip_pixel(src[x_stride] + offset_val); - x_stride += stride; + dst[x * stride_dst + offset] = av_clip_pixel(src[x * stride_src + offset] + offset_val); } width--; } @@ -497,523 +438,727 @@ static void FUNC(sao_edge_filter_0)(uint8_t *_dst, uint8_t *_src, init_y = 1; } if (borders[3]) { - int offset_val = sao_offset_val[0]; - int y_stride = stride * (height - 1); + int offset_val = sao_offset_val[0]; + int y_stride_dst = stride_dst * (height - 1); + int y_stride_src = stride_src * (height - 1); for (x = init_x; x < width; x++) - dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + offset_val); + dst[x + y_stride_dst] = av_clip_pixel(src[x + y_stride_src] + offset_val); height--; } } - { - int y_stride = init_y * stride; - int pos_0_0 = pos[sao_eo_class][0][0]; - int pos_0_1 = pos[sao_eo_class][0][1]; - int pos_1_0 = pos[sao_eo_class][1][0]; - int pos_1_1 = pos[sao_eo_class][1][1]; - - int y_stride_0_1 = (init_y + pos_0_1) * stride; - int y_stride_1_1 = (init_y + pos_1_1) * stride; - for (y = init_y; y < height; y++) { - for (x = init_x; x < width; x++) { - int diff0 = CMP(src[x + y_stride], src[x + pos_0_0 + y_stride_0_1]); - int diff1 = CMP(src[x + y_stride], src[x + pos_1_0 + y_stride_1_1]); - int offset_val = edge_idx[2 + diff0 + diff1]; - dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + sao_offset_val[offset_val]); - } - y_stride += stride; - y_stride_0_1 += stride; - y_stride_1_1 += stride; - } - } - - { - // Restore pixels that can't be modified - int save_upper_left = !diag_edge && sao_eo_class == SAO_EO_135D && !borders[0] && !borders[1]; - if (vert_edge && sao_eo_class != SAO_EO_VERT) - for (y = init_y+save_upper_left; y< height; y++) - dst[y*stride] = src[y*stride]; - if(horiz_edge && sao_eo_class != SAO_EO_HORIZ) - for(x = init_x+save_upper_left; x<width; x++) - dst[x] = src[x]; - if(diag_edge && sao_eo_class == SAO_EO_135D) - dst[0] = src[0]; - } -#undef CMP + FUNC(sao_edge_filter)((uint8_t *)dst, (uint8_t *)src, stride_dst, stride_src, sao, width, height, c_idx, init_x, init_y); } static void FUNC(sao_edge_filter_1)(uint8_t *_dst, uint8_t *_src, - ptrdiff_t stride, SAOParams *sao, + ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao, int *borders, int _width, int _height, - int c_idx, uint8_t vert_edge, - uint8_t horiz_edge, uint8_t diag_edge) + int c_idx, uint8_t *vert_edge, + uint8_t *horiz_edge, uint8_t *diag_edge) { int x, y; pixel *dst = (pixel *)_dst; pixel *src = (pixel *)_src; - int chroma = !!c_idx; - int *sao_offset_val = sao->offset_val[c_idx]; + int16_t *sao_offset_val = sao->offset_val[c_idx]; int sao_eo_class = sao->eo_class[c_idx]; int init_x = 0, init_y = 0, width = _width, height = _height; - static const int8_t pos[4][2][2] = { - { { -1, 0 }, { 1, 0 } }, // horizontal - { { 0, -1 }, { 0, 1 } }, // vertical - { { -1, -1 }, { 1, 1 } }, // 45 degree - { { 1, -1 }, { -1, 1 } }, // 135 degree - }; - static const uint8_t edge_idx[] = { 1, 2, 0, 3, 4 }; - -#define CMP(a, b) ((a) > (b) ? 1 : ((a) == (b) ? 0 : -1)) - - stride /= sizeof(pixel); - - init_y = -(4 >> chroma) - 2; - if (!borders[2]) - width -= (8 >> chroma) + 2; - height = (4 >> chroma) + 2; + stride_dst /= sizeof(pixel); + stride_src /= sizeof(pixel); - dst = dst + (init_y * stride + init_x); - src = src + (init_y * stride + init_x); - init_y = init_x = 0; if (sao_eo_class != SAO_EO_VERT) { if (borders[0]) { int offset_val = sao_offset_val[0]; - int y_stride = 0; for (y = 0; y < height; y++) { - dst[y_stride] = av_clip_pixel(src[y_stride] + offset_val); - y_stride += stride; + dst[y * stride_dst] = av_clip_pixel(src[y * stride_src] + offset_val); } init_x = 1; } if (borders[2]) { int offset_val = sao_offset_val[0]; - int x_stride = width - 1; + int offset = width - 1; for (x = 0; x < height; x++) { - dst[x_stride] = av_clip_pixel(src[x_stride] + offset_val); - x_stride += stride; + dst[x * stride_dst + offset] = av_clip_pixel(src[x * stride_src + offset] + offset_val); } width--; } } - { - int y_stride = init_y * stride; - int pos_0_0 = pos[sao_eo_class][0][0]; - int pos_0_1 = pos[sao_eo_class][0][1]; - int pos_1_0 = pos[sao_eo_class][1][0]; - int pos_1_1 = pos[sao_eo_class][1][1]; - - int y_stride_0_1 = (init_y + pos_0_1) * stride; - int y_stride_1_1 = (init_y + pos_1_1) * stride; - for (y = init_y; y < height; y++) { - for (x = init_x; x < width; x++) { - int diff0 = CMP(src[x + y_stride], src[x + pos_0_0 + y_stride_0_1]); - int diff1 = CMP(src[x + y_stride], src[x + pos_1_0 + y_stride_1_1]); - int offset_val = edge_idx[2 + diff0 + diff1]; - dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + sao_offset_val[offset_val]); - } - y_stride += stride; - y_stride_0_1 += stride; - y_stride_1_1 += stride; + if (sao_eo_class != SAO_EO_HORIZ) { + if (borders[1]) { + int offset_val = sao_offset_val[0]; + for (x = init_x; x < width; x++) + dst[x] = av_clip_pixel(src[x] + offset_val); + init_y = 1; + } + if (borders[3]) { + int offset_val = sao_offset_val[0]; + int y_stride_dst = stride_dst * (height - 1); + int y_stride_src = stride_src * (height - 1); + for (x = init_x; x < width; x++) + dst[x + y_stride_dst] = av_clip_pixel(src[x + y_stride_src] + offset_val); + height--; } } + FUNC(sao_edge_filter)((uint8_t *)dst, (uint8_t *)src, stride_dst, stride_src, sao, width, height, c_idx, init_x, init_y); + { + int save_upper_left = !diag_edge[0] && sao_eo_class == SAO_EO_135D && !borders[0] && !borders[1]; + int save_upper_right = !diag_edge[1] && sao_eo_class == SAO_EO_45D && !borders[1] && !borders[2]; + int save_lower_right = !diag_edge[2] && sao_eo_class == SAO_EO_135D && !borders[2] && !borders[3]; + int save_lower_left = !diag_edge[3] && sao_eo_class == SAO_EO_45D && !borders[0] && !borders[3]; + // Restore pixels that can't be modified - int save_lower_left = !diag_edge && sao_eo_class == SAO_EO_45D && !borders[0]; - if(vert_edge && sao_eo_class != SAO_EO_VERT) - for(y = init_y; y< height-save_lower_left; y++) - dst[y*stride] = src[y*stride]; - if(horiz_edge && sao_eo_class != SAO_EO_HORIZ) - for(x = init_x+save_lower_left; x<width; x++) - dst[(height-1)*stride+x] = src[(height-1)*stride+x]; - if(diag_edge && sao_eo_class == SAO_EO_45D) - dst[stride*(height-1)] = src[stride*(height-1)]; + if(vert_edge[0] && sao_eo_class != SAO_EO_VERT) { + for(y = init_y+save_upper_left; y< height-save_lower_left; y++) + dst[y*stride_dst] = src[y*stride_src]; + } + if(vert_edge[1] && sao_eo_class != SAO_EO_VERT) { + for(y = init_y+save_upper_right; y< height-save_lower_right; y++) + dst[y*stride_dst+width-1] = src[y*stride_src+width-1]; + } + + if(horiz_edge[0] && sao_eo_class != SAO_EO_HORIZ) { + for(x = init_x+save_upper_left; x < width-save_upper_right; x++) + dst[x] = src[x]; + } + if(horiz_edge[1] && sao_eo_class != SAO_EO_HORIZ) { + for(x = init_x+save_lower_left; x < width-save_lower_right; x++) + dst[(height-1)*stride_dst+x] = src[(height-1)*stride_src+x]; + } + if(diag_edge[0] && sao_eo_class == SAO_EO_135D) + dst[0] = src[0]; + if(diag_edge[1] && sao_eo_class == SAO_EO_45D) + dst[width-1] = src[width-1]; + if(diag_edge[2] && sao_eo_class == SAO_EO_135D) + dst[stride_dst*(height-1)+width-1] = src[stride_src*(height-1)+width-1]; + if(diag_edge[3] && sao_eo_class == SAO_EO_45D) + dst[stride_dst*(height-1)] = src[stride_src*(height-1)]; + } +} #undef CMP + +//////////////////////////////////////////////////////////////////////////////// +// +//////////////////////////////////////////////////////////////////////////////// +static void FUNC(put_hevc_pel_pixels)(int16_t *dst, + uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = src[x] << (14 - BIT_DEPTH); + src += srcstride; + dst += MAX_PB_SIZE; + } } -static void FUNC(sao_edge_filter_2)(uint8_t *_dst, uint8_t *_src, - ptrdiff_t stride, SAOParams *sao, - int *borders, int _width, int _height, - int c_idx, uint8_t vert_edge, - uint8_t horiz_edge, uint8_t diag_edge) +static void FUNC(put_hevc_pel_uni_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + + for (y = 0; y < height; y++) { + memcpy(dst, src, width * sizeof(pixel)); + src += srcstride; + dst += dststride; + } +} + +static void FUNC(put_hevc_pel_bi_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, intptr_t mx, intptr_t my, int width) { int x, y; - pixel *dst = (pixel *)_dst; - pixel *src = (pixel *)_src; - int chroma = !!c_idx; - int *sao_offset_val = sao->offset_val[c_idx]; - int sao_eo_class = sao->eo_class[c_idx]; - int init_x = 0, init_y = 0, width = _width, height = _height; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); - static const int8_t pos[4][2][2] = { - { { -1, 0 }, { 1, 0 } }, // horizontal - { { 0, -1 }, { 0, 1 } }, // vertical - { { -1, -1 }, { 1, 1 } }, // 45 degree - { { 1, -1 }, { -1, 1 } }, // 135 degree - }; - static const uint8_t edge_idx[] = { 1, 2, 0, 3, 4 }; + int shift = 14 + 1 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif -#define CMP(a, b) ((a) > (b) ? 1 : ((a) == (b) ? 0 : -1)) + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((src[x] << (14 - BIT_DEPTH)) + src2[x] + offset) >> shift); + src += srcstride; + dst += dststride; + src2 += MAX_PB_SIZE; + } +} - stride /= sizeof(pixel); +static void FUNC(put_hevc_pel_uni_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + int shift = denom + 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif - init_x = -(8 >> chroma) - 2; - width = (8 >> chroma) + 2; - if (!borders[3]) - height -= (4 >> chroma) + 2; + ox = ox * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel((((src[x] << (14 - BIT_DEPTH)) * wx + offset) >> shift) + ox); + src += srcstride; + dst += dststride; + } +} - dst = dst + (init_y * stride + init_x); - src = src + (init_y * stride + init_x); - init_y = init_x = 0; - if (sao_eo_class != SAO_EO_HORIZ) { - if (borders[1]) { - int offset_val = sao_offset_val[0]; - for (x = init_x; x < width; x++) - dst[x] = av_clip_pixel(src[x] + offset_val); - init_y = 1; - } - if (borders[3]) { - int offset_val = sao_offset_val[0]; - int y_stride = stride * (height - 1); - for (x = init_x; x < width; x++) - dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + offset_val); - height--; +static void FUNC(put_hevc_pel_bi_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, int denom, int wx0, int wx1, + int ox0, int ox1, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + + int shift = 14 + 1 - BIT_DEPTH; + int log2Wd = denom + shift - 1; + + ox0 = ox0 * (1 << (BIT_DEPTH - 8)); + ox1 = ox1 * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) { + dst[x] = av_clip_pixel(( (src[x] << (14 - BIT_DEPTH)) * wx1 + src2[x] * wx0 + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1)); } + src += srcstride; + dst += dststride; + src2 += MAX_PB_SIZE; } - { - int y_stride = init_y * stride; - int pos_0_0 = pos[sao_eo_class][0][0]; - int pos_0_1 = pos[sao_eo_class][0][1]; - int pos_1_0 = pos[sao_eo_class][1][0]; - int pos_1_1 = pos[sao_eo_class][1][1]; - - int y_stride_0_1 = (init_y + pos_0_1) * stride; - int y_stride_1_1 = (init_y + pos_1_1) * stride; - for (y = init_y; y < height; y++) { - for (x = init_x; x < width; x++) { - int diff0 = CMP(src[x + y_stride], src[x + pos_0_0 + y_stride_0_1]); - int diff1 = CMP(src[x + y_stride], src[x + pos_1_0 + y_stride_1_1]); - int offset_val = edge_idx[2 + diff0 + diff1]; - dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + sao_offset_val[offset_val]); - } - y_stride += stride; - y_stride_0_1 += stride; - y_stride_1_1 += stride; - } +} + +//////////////////////////////////////////////////////////////////////////////// +// +//////////////////////////////////////////////////////////////////////////////// +#define QPEL_FILTER(src, stride) \ + (filter[0] * src[x - 3 * stride] + \ + filter[1] * src[x - 2 * stride] + \ + filter[2] * src[x - stride] + \ + filter[3] * src[x ] + \ + filter[4] * src[x + stride] + \ + filter[5] * src[x + 2 * stride] + \ + filter[6] * src[x + 3 * stride] + \ + filter[7] * src[x + 4 * stride]) + +static void FUNC(put_hevc_qpel_h)(int16_t *dst, + uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + dst += MAX_PB_SIZE; } +} - { - // Restore pixels that can't be modified - int save_upper_right = !diag_edge && sao_eo_class == SAO_EO_45D && !borders[1]; - if(vert_edge && sao_eo_class != SAO_EO_VERT) - for(y = init_y+save_upper_right; y< height; y++) - dst[y*stride+width-1] = src[y*stride+width-1]; - if(horiz_edge && sao_eo_class != SAO_EO_HORIZ) - for(x = init_x; x<width-save_upper_right; x++) - dst[x] = src[x]; - if(diag_edge && sao_eo_class == SAO_EO_45D) - dst[width-1] = src[width-1]; +static void FUNC(put_hevc_qpel_v)(int16_t *dst, + uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + const int8_t *filter = ff_hevc_qpel_filters[my - 1]; + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8); + src += srcstride; + dst += MAX_PB_SIZE; } -#undef CMP } -static void FUNC(sao_edge_filter_3)(uint8_t *_dst, uint8_t *_src, - ptrdiff_t stride, SAOParams *sao, - int *borders, int _width, int _height, - int c_idx, uint8_t vert_edge, - uint8_t horiz_edge, uint8_t diag_edge) +static void FUNC(put_hevc_qpel_hv)(int16_t *dst, + uint8_t *_src, + ptrdiff_t _srcstride, + int height, intptr_t mx, + intptr_t my, int width) { int x, y; - pixel *dst = (pixel *)_dst; - pixel *src = (pixel *)_src; - int chroma = !!c_idx; - int *sao_offset_val = sao->offset_val[c_idx]; - int sao_eo_class = sao->eo_class[c_idx]; - int init_x = 0, init_y = 0, width = _width, height = _height; + const int8_t *filter; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; - static const int8_t pos[4][2][2] = { - { { -1, 0 }, { 1, 0 } }, // horizontal - { { 0, -1 }, { 0, 1 } }, // vertical - { { -1, -1 }, { 1, 1 } }, // 45 degree - { { 1, -1 }, { -1, 1 } }, // 135 degree - }; - static const uint8_t edge_idx[] = { 1, 2, 0, 3, 4 }; + src -= QPEL_EXTRA_BEFORE * srcstride; + filter = ff_hevc_qpel_filters[mx - 1]; + for (y = 0; y < height + QPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } -#define CMP(a, b) ((a) > (b) ? 1 : ((a) == (b) ? 0 : -1)) + tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_qpel_filters[my - 1]; + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6; + tmp += MAX_PB_SIZE; + dst += MAX_PB_SIZE; + } +} - stride /= sizeof(pixel); +static void FUNC(put_hevc_qpel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride, + uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; + int shift = 14 - BIT_DEPTH; + +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift); + src += srcstride; + dst += dststride; + } +} - init_y = -(4 >> chroma) - 2; - init_x = -(8 >> chroma) - 2; - width = (8 >> chroma) + 2; - height = (4 >> chroma) + 2; +static void FUNC(put_hevc_qpel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; - dst = dst + (init_y * stride + init_x); - src = src + (init_y * stride + init_x); - init_y = init_x = 0; + int shift = 14 + 1 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif - { - int y_stride = init_y * stride; - int pos_0_0 = pos[sao_eo_class][0][0]; - int pos_0_1 = pos[sao_eo_class][0][1]; - int pos_1_0 = pos[sao_eo_class][1][0]; - int pos_1_1 = pos[sao_eo_class][1][1]; - - int y_stride_0_1 = (init_y + pos_0_1) * stride; - int y_stride_1_1 = (init_y + pos_1_1) * stride; - - for (y = init_y; y < height; y++) { - for (x = init_x; x < width; x++) { - int diff0 = CMP(src[x + y_stride], src[x + pos_0_0 + y_stride_0_1]); - int diff1 = CMP(src[x + y_stride], src[x + pos_1_0 + y_stride_1_1]); - int offset_val = edge_idx[2 + diff0 + diff1]; - dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + sao_offset_val[offset_val]); - } - y_stride += stride; - y_stride_0_1 += stride; - y_stride_1_1 += stride; - } + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift); + src += srcstride; + dst += dststride; + src2 += MAX_PB_SIZE; } +} - { - // Restore pixels that can't be modified - int save_lower_right = !diag_edge && sao_eo_class == SAO_EO_135D; - if(vert_edge && sao_eo_class != SAO_EO_VERT) - for(y = init_y; y< height-save_lower_right; y++) - dst[y*stride+width-1] = src[y*stride+width-1]; - if(horiz_edge && sao_eo_class != SAO_EO_HORIZ) - for(x = init_x; x<width-save_lower_right; x++) - dst[(height-1)*stride+x] = src[(height-1)*stride+x]; - if(diag_edge && sao_eo_class == SAO_EO_135D) - dst[stride*(height-1)+width-1] = src[stride*(height-1)+width-1]; +static void FUNC(put_hevc_qpel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride, + uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_qpel_filters[my - 1]; + int shift = 14 - BIT_DEPTH; + +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift); + src += srcstride; + dst += dststride; } -#undef CMP } -#undef SET -#undef SCALE -#undef ADD_AND_SCALE -#undef TR_4 -#undef TR_8 -#undef TR_16 -#undef TR_32 -static void FUNC(put_hevc_qpel_pixels)(int16_t *dst, ptrdiff_t dststride, +static void FUNC(put_hevc_qpel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + + const int8_t *filter = ff_hevc_qpel_filters[my - 1]; + + int shift = 14 + 1 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift); + src += srcstride; + dst += dststride; + src2 += MAX_PB_SIZE; + } +} + +static void FUNC(put_hevc_qpel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, - int width, int height, int16_t* mcbuffer) + int height, intptr_t mx, intptr_t my, int width) { int x, y; - pixel *src = (pixel *)_src; + const int8_t *filter; + pixel *src = (pixel*)_src; ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = 14 - BIT_DEPTH; + +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + src -= QPEL_EXTRA_BEFORE * srcstride; + filter = ff_hevc_qpel_filters[mx - 1]; + for (y = 0; y < height + QPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } + + tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_qpel_filters[my - 1]; for (y = 0; y < height; y++) { for (x = 0; x < width; x++) - dst[x] = src[x] << (14 - BIT_DEPTH); + dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift); + tmp += MAX_PB_SIZE; + dst += dststride; + } +} + +static void FUNC(put_hevc_qpel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + const int8_t *filter; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = 14 + 1 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + src -= QPEL_EXTRA_BEFORE * srcstride; + filter = ff_hevc_qpel_filters[mx - 1]; + for (y = 0; y < height + QPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } + + tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_qpel_filters[my - 1]; + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + src2[x] + offset) >> shift); + tmp += MAX_PB_SIZE; + dst += dststride; + src2 += MAX_PB_SIZE; + } +} + +static void FUNC(put_hevc_qpel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride, + uint8_t *_src, ptrdiff_t _srcstride, + int height, int denom, int wx, int ox, + intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; + int shift = denom + 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + ox = ox * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel((((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); src += srcstride; dst += dststride; } } -#define QPEL_FILTER_1(src, stride) \ - (1 * -src[x - 3 * stride] + \ - 4 * src[x - 2 * stride] - \ - 10 * src[x - stride] + \ - 58 * src[x] + \ - 17 * src[x + stride] - \ - 5 * src[x + 2 * stride] + \ - 1 * src[x + 3 * stride]) - -#define QPEL_FILTER_2(src, stride) \ - (1 * -src[x - 3 * stride] + \ - 4 * src[x - 2 * stride] - \ - 11 * src[x - stride] + \ - 40 * src[x] + \ - 40 * src[x + stride] - \ - 11 * src[x + 2 * stride] + \ - 4 * src[x + 3 * stride] - \ - 1 * src[x + 4 * stride]) - -#define QPEL_FILTER_3(src, stride) \ - (1 * src[x - 2 * stride] - \ - 5 * src[x - stride] + \ - 17 * src[x] + \ - 58 * src[x + stride] - \ - 10 * src[x + 2 * stride] + \ - 4 * src[x + 3 * stride] - \ - 1 * src[x + 4 * stride]) - - -#define PUT_HEVC_QPEL_H(H) \ -static void FUNC(put_hevc_qpel_h ## H)(int16_t *dst, ptrdiff_t dststride, \ - uint8_t *_src, ptrdiff_t _srcstride, \ - int width, int height, \ - int16_t* mcbuffer) \ -{ \ - int x, y; \ - pixel *src = (pixel*)_src; \ - ptrdiff_t srcstride = _srcstride / sizeof(pixel); \ - \ - for (y = 0; y < height; y++) { \ - for (x = 0; x < width; x++) \ - dst[x] = QPEL_FILTER_ ## H(src, 1) >> (BIT_DEPTH - 8); \ - src += srcstride; \ - dst += dststride; \ - } \ +static void FUNC(put_hevc_qpel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, int denom, int wx0, int wx1, + int ox0, int ox1, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + + const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; + + int shift = 14 + 1 - BIT_DEPTH; + int log2Wd = denom + shift - 1; + + ox0 = ox0 * (1 << (BIT_DEPTH - 8)); + ox1 = ox1 * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 + + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1)); + src += srcstride; + dst += dststride; + src2 += MAX_PB_SIZE; + } } -#define PUT_HEVC_QPEL_V(V) \ -static void FUNC(put_hevc_qpel_v ## V)(int16_t *dst, ptrdiff_t dststride, \ - uint8_t *_src, ptrdiff_t _srcstride, \ - int width, int height, \ - int16_t* mcbuffer) \ -{ \ - int x, y; \ - pixel *src = (pixel*)_src; \ - ptrdiff_t srcstride = _srcstride / sizeof(pixel); \ - \ - for (y = 0; y < height; y++) { \ - for (x = 0; x < width; x++) \ - dst[x] = QPEL_FILTER_ ## V(src, srcstride) >> (BIT_DEPTH - 8); \ - src += srcstride; \ - dst += dststride; \ - } \ +static void FUNC(put_hevc_qpel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride, + uint8_t *_src, ptrdiff_t _srcstride, + int height, int denom, int wx, int ox, + intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_qpel_filters[my - 1]; + int shift = denom + 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + ox = ox * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel((((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); + src += srcstride; + dst += dststride; + } } -#define PUT_HEVC_QPEL_HV(H, V) \ -static void FUNC(put_hevc_qpel_h ## H ## v ## V)(int16_t *dst, \ - ptrdiff_t dststride, \ - uint8_t *_src, \ - ptrdiff_t _srcstride, \ - int width, int height, \ - int16_t* mcbuffer) \ -{ \ - int x, y; \ - pixel *src = (pixel*)_src; \ - ptrdiff_t srcstride = _srcstride / sizeof(pixel); \ - \ - int16_t tmp_array[(MAX_PB_SIZE + 7) * MAX_PB_SIZE]; \ - int16_t *tmp = tmp_array; \ - \ - src -= ff_hevc_qpel_extra_before[V] * srcstride; \ - \ - for (y = 0; y < height + ff_hevc_qpel_extra[V]; y++) { \ - for (x = 0; x < width; x++) \ - tmp[x] = QPEL_FILTER_ ## H(src, 1) >> (BIT_DEPTH - 8); \ - src += srcstride; \ - tmp += MAX_PB_SIZE; \ - } \ - \ - tmp = tmp_array + ff_hevc_qpel_extra_before[V] * MAX_PB_SIZE; \ - \ - for (y = 0; y < height; y++) { \ - for (x = 0; x < width; x++) \ - dst[x] = QPEL_FILTER_ ## V(tmp, MAX_PB_SIZE) >> 6; \ - tmp += MAX_PB_SIZE; \ - dst += dststride; \ - } \ +static void FUNC(put_hevc_qpel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, int denom, int wx0, int wx1, + int ox0, int ox1, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + + const int8_t *filter = ff_hevc_qpel_filters[my - 1]; + + int shift = 14 + 1 - BIT_DEPTH; + int log2Wd = denom + shift - 1; + + ox0 = ox0 * (1 << (BIT_DEPTH - 8)); + ox1 = ox1 * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 + + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1)); + src += srcstride; + dst += dststride; + src2 += MAX_PB_SIZE; + } } -PUT_HEVC_QPEL_H(1) -PUT_HEVC_QPEL_H(2) -PUT_HEVC_QPEL_H(3) -PUT_HEVC_QPEL_V(1) -PUT_HEVC_QPEL_V(2) -PUT_HEVC_QPEL_V(3) -PUT_HEVC_QPEL_HV(1, 1) -PUT_HEVC_QPEL_HV(1, 2) -PUT_HEVC_QPEL_HV(1, 3) -PUT_HEVC_QPEL_HV(2, 1) -PUT_HEVC_QPEL_HV(2, 2) -PUT_HEVC_QPEL_HV(2, 3) -PUT_HEVC_QPEL_HV(3, 1) -PUT_HEVC_QPEL_HV(3, 2) -PUT_HEVC_QPEL_HV(3, 3) - -static void FUNC(put_hevc_epel_pixels)(int16_t *dst, ptrdiff_t dststride, - uint8_t *_src, ptrdiff_t _srcstride, - int width, int height, int mx, int my, - int16_t* mcbuffer) +static void FUNC(put_hevc_qpel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, + uint8_t *_src, ptrdiff_t _srcstride, + int height, int denom, int wx, int ox, + intptr_t mx, intptr_t my, int width) { int x, y; - pixel *src = (pixel *)_src; + const int8_t *filter; + pixel *src = (pixel*)_src; ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = denom + 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif - for (y = 0; y < height; y++) { + src -= QPEL_EXTRA_BEFORE * srcstride; + filter = ff_hevc_qpel_filters[mx - 1]; + for (y = 0; y < height + QPEL_EXTRA; y++) { for (x = 0; x < width; x++) - dst[x] = src[x] << (14 - BIT_DEPTH); + tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); src += srcstride; + tmp += MAX_PB_SIZE; + } + + tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_qpel_filters[my - 1]; + + ox = ox * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel((((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox); + tmp += MAX_PB_SIZE; dst += dststride; } } -#define EPEL_FILTER(src, stride) \ - (filter_0 * src[x - stride] + \ - filter_1 * src[x] + \ - filter_2 * src[x + stride] + \ - filter_3 * src[x + 2 * stride]) +static void FUNC(put_hevc_qpel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, int denom, int wx0, int wx1, + int ox0, int ox1, intptr_t mx, intptr_t my, int width) +{ + int x, y; + const int8_t *filter; + pixel *src = (pixel*)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = 14 + 1 - BIT_DEPTH; + int log2Wd = denom + shift - 1; + + src -= QPEL_EXTRA_BEFORE * srcstride; + filter = ff_hevc_qpel_filters[mx - 1]; + for (y = 0; y < height + QPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } + + tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_qpel_filters[my - 1]; + + ox0 = ox0 * (1 << (BIT_DEPTH - 8)); + ox1 = ox1 * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx1 + src2[x] * wx0 + + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1)); + tmp += MAX_PB_SIZE; + dst += dststride; + src2 += MAX_PB_SIZE; + } +} + +//////////////////////////////////////////////////////////////////////////////// +// +//////////////////////////////////////////////////////////////////////////////// +#define EPEL_FILTER(src, stride) \ + (filter[0] * src[x - stride] + \ + filter[1] * src[x] + \ + filter[2] * src[x + stride] + \ + filter[3] * src[x + 2 * stride]) -static void FUNC(put_hevc_epel_h)(int16_t *dst, ptrdiff_t dststride, +static void FUNC(put_hevc_epel_h)(int16_t *dst, uint8_t *_src, ptrdiff_t _srcstride, - int width, int height, int mx, int my, - int16_t* mcbuffer) + int height, intptr_t mx, intptr_t my, int width) { int x, y; pixel *src = (pixel *)_src; ptrdiff_t srcstride = _srcstride / sizeof(pixel); const int8_t *filter = ff_hevc_epel_filters[mx - 1]; - int8_t filter_0 = filter[0]; - int8_t filter_1 = filter[1]; - int8_t filter_2 = filter[2]; - int8_t filter_3 = filter[3]; for (y = 0; y < height; y++) { for (x = 0; x < width; x++) dst[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); src += srcstride; - dst += dststride; + dst += MAX_PB_SIZE; } } -static void FUNC(put_hevc_epel_v)(int16_t *dst, ptrdiff_t dststride, +static void FUNC(put_hevc_epel_v)(int16_t *dst, uint8_t *_src, ptrdiff_t _srcstride, - int width, int height, int mx, int my, - int16_t* mcbuffer) + int height, intptr_t mx, intptr_t my, int width) { int x, y; pixel *src = (pixel *)_src; ptrdiff_t srcstride = _srcstride / sizeof(pixel); const int8_t *filter = ff_hevc_epel_filters[my - 1]; - int8_t filter_0 = filter[0]; - int8_t filter_1 = filter[1]; - int8_t filter_2 = filter[2]; - int8_t filter_3 = filter[3]; for (y = 0; y < height; y++) { for (x = 0; x < width; x++) dst[x] = EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8); src += srcstride; - dst += dststride; + dst += MAX_PB_SIZE; } } -static void FUNC(put_hevc_epel_hv)(int16_t *dst, ptrdiff_t dststride, +static void FUNC(put_hevc_epel_hv)(int16_t *dst, uint8_t *_src, ptrdiff_t _srcstride, - int width, int height, int mx, int my, - int16_t* mcbuffer) + int height, intptr_t mx, intptr_t my, int width) { int x, y; pixel *src = (pixel *)_src; ptrdiff_t srcstride = _srcstride / sizeof(pixel); - const int8_t *filter_h = ff_hevc_epel_filters[mx - 1]; - const int8_t *filter_v = ff_hevc_epel_filters[my - 1]; - int8_t filter_0 = filter_h[0]; - int8_t filter_1 = filter_h[1]; - int8_t filter_2 = filter_h[2]; - int8_t filter_3 = filter_h[3]; - int16_t tmp_array[(MAX_PB_SIZE + 3) * MAX_PB_SIZE]; + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; int16_t *tmp = tmp_array; src -= EPEL_EXTRA_BEFORE * srcstride; @@ -1026,49 +1171,101 @@ static void FUNC(put_hevc_epel_hv)(int16_t *dst, ptrdiff_t dststride, } tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; - filter_0 = filter_v[0]; - filter_1 = filter_v[1]; - filter_2 = filter_v[2]; - filter_3 = filter_v[3]; + filter = ff_hevc_epel_filters[my - 1]; + for (y = 0; y < height; y++) { for (x = 0; x < width; x++) dst[x] = EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6; tmp += MAX_PB_SIZE; + dst += MAX_PB_SIZE; + } +} + +static void FUNC(put_hevc_epel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int shift = 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift); + src += srcstride; dst += dststride; } } -static void FUNC(put_unweighted_pred)(uint8_t *_dst, ptrdiff_t _dststride, - int16_t *src, ptrdiff_t srcstride, - int width, int height) +static void FUNC(put_hevc_epel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, intptr_t mx, intptr_t my, int width) { int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); pixel *dst = (pixel *)_dst; ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int shift = 14 + 1 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) { + dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift); + } + dst += dststride; + src += srcstride; + src2 += MAX_PB_SIZE; + } +} +static void FUNC(put_hevc_epel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[my - 1]; int shift = 14 - BIT_DEPTH; #if BIT_DEPTH < 14 int offset = 1 << (shift - 1); #else int offset = 0; #endif + for (y = 0; y < height; y++) { for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((src[x] + offset) >> shift); - dst += dststride; + dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift); src += srcstride; + dst += dststride; } } -static void FUNC(put_weighted_pred_avg)(uint8_t *_dst, ptrdiff_t _dststride, - int16_t *src1, int16_t *src2, - ptrdiff_t srcstride, - int width, int height) +static void FUNC(put_hevc_epel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, intptr_t mx, intptr_t my, int width) { int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[my - 1]; pixel *dst = (pixel *)_dst; ptrdiff_t dststride = _dststride / sizeof(pixel); - int shift = 14 + 1 - BIT_DEPTH; #if BIT_DEPTH < 14 int offset = 1 << (shift - 1); @@ -1078,71 +1275,273 @@ static void FUNC(put_weighted_pred_avg)(uint8_t *_dst, ptrdiff_t _dststride, for (y = 0; y < height; y++) { for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((src1[x] + src2[x] + offset) >> shift); + dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift); dst += dststride; - src1 += srcstride; - src2 += srcstride; + src += srcstride; + src2 += MAX_PB_SIZE; + } +} + +static void FUNC(put_hevc_epel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + src -= EPEL_EXTRA_BEFORE * srcstride; + + for (y = 0; y < height + EPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } + + tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_epel_filters[my - 1]; + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift); + tmp += MAX_PB_SIZE; + dst += dststride; } } -static void FUNC(weighted_pred)(uint8_t denom, int16_t wlxFlag, int16_t olxFlag, - uint8_t *_dst, ptrdiff_t _dststride, - int16_t *src, ptrdiff_t srcstride, - int width, int height) +static void FUNC(put_hevc_epel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, intptr_t mx, intptr_t my, int width) { - int shift, log2Wd, wx, ox, x, y, offset; + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); pixel *dst = (pixel *)_dst; ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = 14 + 1 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + src -= EPEL_EXTRA_BEFORE * srcstride; + + for (y = 0; y < height + EPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } - shift = 14 - BIT_DEPTH; - log2Wd = denom + shift; - offset = 1 << (log2Wd - 1); - wx = wlxFlag; - ox = olxFlag * (1 << (BIT_DEPTH - 8)); + tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_epel_filters[my - 1]; + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + src2[x] + offset) >> shift); + tmp += MAX_PB_SIZE; + dst += dststride; + src2 += MAX_PB_SIZE; + } +} + +static void FUNC(put_hevc_epel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int shift = denom + 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + ox = ox * (1 << (BIT_DEPTH - 8)); for (y = 0; y < height; y++) { for (x = 0; x < width; x++) { - if (log2Wd >= 1) { - dst[x] = av_clip_pixel(((src[x] * wx + offset) >> log2Wd) + ox); - } else { - dst[x] = av_clip_pixel(src[x] * wx + ox); - } + dst[x] = av_clip_pixel((((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); } dst += dststride; src += srcstride; } } -static void FUNC(weighted_pred_avg)(uint8_t denom, - int16_t wl0Flag, int16_t wl1Flag, - int16_t ol0Flag, int16_t ol1Flag, - uint8_t *_dst, ptrdiff_t _dststride, - int16_t *src1, int16_t *src2, - ptrdiff_t srcstride, - int width, int height) +static void FUNC(put_hevc_epel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, int denom, int wx0, int wx1, + int ox0, int ox1, intptr_t mx, intptr_t my, int width) { - int shift, log2Wd, w0, w1, o0, o1, x, y; - pixel *dst = (pixel *)_dst; + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int shift = 14 + 1 - BIT_DEPTH; + int log2Wd = denom + shift - 1; + + ox0 = ox0 * (1 << (BIT_DEPTH - 8)); + ox1 = ox1 * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 + + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1)); + src += srcstride; + dst += dststride; + src2 += MAX_PB_SIZE; + } +} + +static void FUNC(put_hevc_epel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[my - 1]; + int shift = denom + 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + ox = ox * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) { + dst[x] = av_clip_pixel((((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); + } + dst += dststride; + src += srcstride; + } +} - shift = 14 - BIT_DEPTH; - log2Wd = denom + shift; - w0 = wl0Flag; - w1 = wl1Flag; - o0 = ol0Flag * (1 << (BIT_DEPTH - 8)); - o1 = ol1Flag * (1 << (BIT_DEPTH - 8)); +static void FUNC(put_hevc_epel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, int denom, int wx0, int wx1, + int ox0, int ox1, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[my - 1]; + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + int shift = 14 + 1 - BIT_DEPTH; + int log2Wd = denom + shift - 1; + ox0 = ox0 * (1 << (BIT_DEPTH - 8)); + ox1 = ox1 * (1 << (BIT_DEPTH - 8)); for (y = 0; y < height; y++) { for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((src1[x] * w0 + src2[x] * w1 + - ((o0 + o1 + 1) << log2Wd)) >> (log2Wd + 1)); + dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 + + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1)); + src += srcstride; dst += dststride; - src1 += srcstride; - src2 += srcstride; + src2 += MAX_PB_SIZE; + } +} + +static void FUNC(put_hevc_epel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = denom + 14 - BIT_DEPTH; +#if BIT_DEPTH < 14 + int offset = 1 << (shift - 1); +#else + int offset = 0; +#endif + + src -= EPEL_EXTRA_BEFORE * srcstride; + + for (y = 0; y < height + EPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } + + tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_epel_filters[my - 1]; + + ox = ox * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel((((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox); + tmp += MAX_PB_SIZE; + dst += dststride; } } -// line zero +static void FUNC(put_hevc_epel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride, + int16_t *src2, + int height, int denom, int wx0, int wx1, + int ox0, int ox1, intptr_t mx, intptr_t my, int width) +{ + int x, y; + pixel *src = (pixel *)_src; + ptrdiff_t srcstride = _srcstride / sizeof(pixel); + pixel *dst = (pixel *)_dst; + ptrdiff_t dststride = _dststride / sizeof(pixel); + const int8_t *filter = ff_hevc_epel_filters[mx - 1]; + int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; + int16_t *tmp = tmp_array; + int shift = 14 + 1 - BIT_DEPTH; + int log2Wd = denom + shift - 1; + + src -= EPEL_EXTRA_BEFORE * srcstride; + + for (y = 0; y < height + EPEL_EXTRA; y++) { + for (x = 0; x < width; x++) + tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); + src += srcstride; + tmp += MAX_PB_SIZE; + } + + tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; + filter = ff_hevc_epel_filters[my - 1]; + + ox0 = ox0 * (1 << (BIT_DEPTH - 8)); + ox1 = ox1 * (1 << (BIT_DEPTH - 8)); + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) + dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx1 + src2[x] * wx0 + + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1)); + tmp += MAX_PB_SIZE; + dst += dststride; + src2 += MAX_PB_SIZE; + } +}// line zero #define P3 pix[-4 * xstride] #define P2 pix[-3 * xstride] #define P1 pix[-2 * xstride] @@ -1293,21 +1692,21 @@ static void FUNC(hevc_loop_filter_chroma)(uint8_t *_pix, ptrdiff_t _xstride, } static void FUNC(hevc_h_loop_filter_chroma)(uint8_t *pix, ptrdiff_t stride, - int *tc, uint8_t *no_p, + int32_t *tc, uint8_t *no_p, uint8_t *no_q) { FUNC(hevc_loop_filter_chroma)(pix, stride, sizeof(pixel), tc, no_p, no_q); } static void FUNC(hevc_v_loop_filter_chroma)(uint8_t *pix, ptrdiff_t stride, - int *tc, uint8_t *no_p, + int32_t *tc, uint8_t *no_p, uint8_t *no_q) { FUNC(hevc_loop_filter_chroma)(pix, sizeof(pixel), stride, tc, no_p, no_q); } static void FUNC(hevc_h_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride, - int beta, int *tc, uint8_t *no_p, + int beta, int32_t *tc, uint8_t *no_p, uint8_t *no_q) { FUNC(hevc_loop_filter_luma)(pix, stride, sizeof(pixel), @@ -1315,7 +1714,7 @@ static void FUNC(hevc_h_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride, } static void FUNC(hevc_v_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride, - int beta, int *tc, uint8_t *no_p, + int beta, int32_t *tc, uint8_t *no_p, uint8_t *no_q) { FUNC(hevc_loop_filter_luma)(pix, sizeof(pixel), stride, |