diff options
Diffstat (limited to 'libavcodec/vc1_block.c')
-rw-r--r-- | libavcodec/vc1_block.c | 1157 |
1 files changed, 542 insertions, 615 deletions
diff --git a/libavcodec/vc1_block.c b/libavcodec/vc1_block.c index d8f45f9..86320db 100644 --- a/libavcodec/vc1_block.c +++ b/libavcodec/vc1_block.c @@ -4,20 +4,20 @@ * Copyright (c) 2006-2007 Konstantin Shishkov * Partly based on vc9.c (c) 2005 Anonymous, Alex Beregszaszi, Michael Niedermayer * - * 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 */ @@ -30,7 +30,7 @@ #include "mpegutils.h" #include "mpegvideo.h" #include "msmpeg4data.h" -#include "unary_legacy.h" +#include "unary.h" #include "vc1.h" #include "vc1_pred.h" #include "vc1acdata.h" @@ -40,8 +40,13 @@ #define DC_VLC_BITS 9 // offset tables for interlaced picture MVDATA decoding -static const int offset_table1[9] = { 0, 1, 2, 4, 8, 16, 32, 64, 128 }; -static const int offset_table2[9] = { 0, 1, 3, 7, 15, 31, 63, 127, 255 }; +static const uint8_t offset_table[2][9] = { + { 0, 1, 2, 4, 8, 16, 32, 64, 128 }, + { 0, 1, 3, 7, 15, 31, 63, 127, 255 }, +}; + +// mapping table for internal block representation +static const int block_map[6] = {0, 2, 1, 3, 4, 5}; /***********************************************************************/ /** @@ -50,22 +55,8 @@ static const int offset_table2[9] = { 0, 1, 3, 7, 15, 31, 63, 127, 255 }; * @{ */ -/** - * Imode types - * @{ - */ -enum Imode { - IMODE_RAW, - IMODE_NORM2, - IMODE_DIFF2, - IMODE_NORM6, - IMODE_DIFF6, - IMODE_ROWSKIP, - IMODE_COLSKIP -}; -/** @} */ //imode defines -static void init_block_index(VC1Context *v) +static inline void init_block_index(VC1Context *v) { MpegEncContext *s = &v->s; ff_init_block_index(s); @@ -78,72 +69,97 @@ static void init_block_index(VC1Context *v) /** @} */ //Bitplane group -static void vc1_put_signed_blocks_clamped(VC1Context *v) +static void vc1_put_blocks_clamped(VC1Context *v, int put_signed) { MpegEncContext *s = &v->s; - int topleft_mb_pos, top_mb_pos; - int stride_y, fieldtx = 0; - int v_dist; - - /* The put pixels loop is always one MB row behind the decoding loop, - * because we can only put pixels when overlap filtering is done, and - * for filtering of the bottom edge of a MB, we need the next MB row - * present as well. - * Within the row, the put pixels loop is also one MB col behind the - * decoding loop. The reason for this is again, because for filtering - * of the right MB edge, we need the next MB present. */ - if (!s->first_slice_line) { + uint8_t *dest; + int block_count = CONFIG_GRAY && (s->avctx->flags & AV_CODEC_FLAG_GRAY) ? 4 : 6; + int fieldtx = 0; + int i; + + /* The put pixels loop is one MB row and one MB column behind the decoding + * loop because we can only put pixels when overlap filtering is done. For + * interlaced frame pictures, however, the put pixels loop is only one + * column behind the decoding loop as interlaced frame pictures only need + * horizontal overlap filtering. */ + if (!s->first_slice_line && v->fcm != ILACE_FRAME) { + if (s->mb_x) { + for (i = 0; i < block_count; i++) { + if (i > 3 ? v->mb_type[0][s->block_index[i] - s->block_wrap[i] - 1] : + v->mb_type[0][s->block_index[i] - 2 * s->block_wrap[i] - 2]) { + dest = s->dest[0] + ((i & 2) - 4) * 4 * s->linesize + ((i & 1) - 2) * 8; + if (put_signed) + s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] - 8 * s->uvlinesize - 8 : dest, + i > 3 ? s->uvlinesize : s->linesize); + else + s->idsp.put_pixels_clamped(v->block[v->topleft_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] - 8 * s->uvlinesize - 8 : dest, + i > 3 ? s->uvlinesize : s->linesize); + } + } + } + if (s->mb_x == v->end_mb_x - 1) { + for (i = 0; i < block_count; i++) { + if (i > 3 ? v->mb_type[0][s->block_index[i] - s->block_wrap[i]] : + v->mb_type[0][s->block_index[i] - 2 * s->block_wrap[i]]) { + dest = s->dest[0] + ((i & 2) - 4) * 4 * s->linesize + (i & 1) * 8; + if (put_signed) + s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] - 8 * s->uvlinesize : dest, + i > 3 ? s->uvlinesize : s->linesize); + else + s->idsp.put_pixels_clamped(v->block[v->top_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] - 8 * s->uvlinesize : dest, + i > 3 ? s->uvlinesize : s->linesize); + } + } + } + } + if (s->mb_y == s->end_mb_y - 1 || v->fcm == ILACE_FRAME) { if (s->mb_x) { - topleft_mb_pos = (s->mb_y - 1) * s->mb_stride + s->mb_x - 1; if (v->fcm == ILACE_FRAME) - fieldtx = v->fieldtx_plane[topleft_mb_pos]; - stride_y = s->linesize << fieldtx; - v_dist = (16 - fieldtx) >> (fieldtx == 0); - s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][0], - s->dest[0] - 16 * s->linesize - 16, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][1], - s->dest[0] - 16 * s->linesize - 8, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][2], - s->dest[0] - v_dist * s->linesize - 16, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][3], - s->dest[0] - v_dist * s->linesize - 8, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][4], - s->dest[1] - 8 * s->uvlinesize - 8, - s->uvlinesize); - s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][5], - s->dest[2] - 8 * s->uvlinesize - 8, - s->uvlinesize); + fieldtx = v->fieldtx_plane[s->mb_y * s->mb_stride + s->mb_x - 1]; + for (i = 0; i < block_count; i++) { + if (i > 3 ? v->mb_type[0][s->block_index[i] - 1] : + v->mb_type[0][s->block_index[i] - 2]) { + if (fieldtx) + dest = s->dest[0] + ((i & 2) >> 1) * s->linesize + ((i & 1) - 2) * 8; + else + dest = s->dest[0] + (i & 2) * 4 * s->linesize + ((i & 1) - 2) * 8; + if (put_signed) + s->idsp.put_signed_pixels_clamped(v->block[v->left_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] - 8 : dest, + i > 3 ? s->uvlinesize : s->linesize << fieldtx); + else + s->idsp.put_pixels_clamped(v->block[v->left_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] - 8 : dest, + i > 3 ? s->uvlinesize : s->linesize << fieldtx); + } + } } - if (s->mb_x == s->mb_width - 1) { - top_mb_pos = (s->mb_y - 1) * s->mb_stride + s->mb_x; + if (s->mb_x == v->end_mb_x - 1) { if (v->fcm == ILACE_FRAME) - fieldtx = v->fieldtx_plane[top_mb_pos]; - stride_y = s->linesize << fieldtx; - v_dist = fieldtx ? 15 : 8; - s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][0], - s->dest[0] - 16 * s->linesize, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][1], - s->dest[0] - 16 * s->linesize + 8, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][2], - s->dest[0] - v_dist * s->linesize, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][3], - s->dest[0] - v_dist * s->linesize + 8, - stride_y); - s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][4], - s->dest[1] - 8 * s->uvlinesize, - s->uvlinesize); - s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][5], - s->dest[2] - 8 * s->uvlinesize, - s->uvlinesize); + fieldtx = v->fieldtx_plane[s->mb_y * s->mb_stride + s->mb_x]; + for (i = 0; i < block_count; i++) { + if (v->mb_type[0][s->block_index[i]]) { + if (fieldtx) + dest = s->dest[0] + ((i & 2) >> 1) * s->linesize + (i & 1) * 8; + else + dest = s->dest[0] + (i & 2) * 4 * s->linesize + (i & 1) * 8; + if (put_signed) + s->idsp.put_signed_pixels_clamped(v->block[v->cur_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] : dest, + i > 3 ? s->uvlinesize : s->linesize << fieldtx); + else + s->idsp.put_pixels_clamped(v->block[v->cur_blk_idx][block_map[i]], + i > 3 ? s->dest[i - 3] : dest, + i > 3 ? s->uvlinesize : s->linesize << fieldtx); + } + } } } +} #define inc_blk_idx(idx) do { \ idx++; \ @@ -151,12 +167,6 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v) idx = 0; \ } while (0) - inc_blk_idx(v->topleft_blk_idx); - inc_blk_idx(v->top_blk_idx); - inc_blk_idx(v->left_blk_idx); - inc_blk_idx(v->cur_blk_idx); -} - /***********************************************************************/ /** * @name VC-1 Block-level functions @@ -173,13 +183,13 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v) int edges = 0; \ if (v->dqprofile == DQPROFILE_ALL_MBS) { \ if (v->dqbilevel) { \ - mquant = (get_bits1(gb)) ? v->altpq : v->pq; \ + mquant = (get_bits1(gb)) ? -v->altpq : v->pq; \ } else { \ mqdiff = get_bits(gb, 3); \ if (mqdiff != 7) \ - mquant = v->pq + mqdiff; \ + mquant = -v->pq - mqdiff; \ else \ - mquant = get_bits(gb, 5); \ + mquant = -get_bits(gb, 5); \ } \ } \ if (v->dqprofile == DQPROFILE_SINGLE_EDGE) \ @@ -189,14 +199,15 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v) else if (v->dqprofile == DQPROFILE_FOUR_EDGES) \ edges = 15; \ if ((edges&1) && !s->mb_x) \ - mquant = v->altpq; \ - if ((edges&2) && s->first_slice_line) \ - mquant = v->altpq; \ + mquant = -v->altpq; \ + if ((edges&2) && !s->mb_y) \ + mquant = -v->altpq; \ if ((edges&4) && s->mb_x == (s->mb_width - 1)) \ - mquant = v->altpq; \ - if ((edges&8) && s->mb_y == (s->mb_height - 1)) \ - mquant = v->altpq; \ - if (!mquant || mquant > 31) { \ + mquant = -v->altpq; \ + if ((edges&8) && \ + s->mb_y == ((s->mb_height >> v->field_mode) - 1)) \ + mquant = -v->altpq; \ + if (!mquant || mquant > 31 || mquant < -31) { \ av_log(v->s.avctx, AV_LOG_ERROR, \ "Overriding invalid mquant %d\n", mquant); \ mquant = 1; \ @@ -230,33 +241,32 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v) s->mb_intra = 1; \ } else { \ index1 = index % 6; \ - if (!s->quarter_sample && index1 == 5) val = 1; \ - else val = 0; \ - if (size_table[index1] - val > 0) \ - val = get_bits(gb, size_table[index1] - val); \ - else val = 0; \ - sign = 0 - (val&1); \ - _dmv_x = (sign ^ ((val>>1) + offset_table[index1])) - sign; \ + _dmv_x = offset_table[1][index1]; \ + val = size_table[index1] - (!s->quarter_sample && index1 == 5); \ + if (val > 0) { \ + val = get_bits(gb, val); \ + sign = 0 - (val & 1); \ + _dmv_x = (sign ^ ((val >> 1) + _dmv_x)) - sign; \ + } \ \ index1 = index / 6; \ - if (!s->quarter_sample && index1 == 5) val = 1; \ - else val = 0; \ - if (size_table[index1] - val > 0) \ - val = get_bits(gb, size_table[index1] - val); \ - else val = 0; \ - sign = 0 - (val & 1); \ - _dmv_y = (sign ^ ((val >> 1) + offset_table[index1])) - sign; \ + _dmv_y = offset_table[1][index1]; \ + val = size_table[index1] - (!s->quarter_sample && index1 == 5); \ + if (val > 0) { \ + val = get_bits(gb, val); \ + sign = 0 - (val & 1); \ + _dmv_y = (sign ^ ((val >> 1) + _dmv_y)) - sign; \ + } \ } static av_always_inline void get_mvdata_interlaced(VC1Context *v, int *dmv_x, int *dmv_y, int *pred_flag) { int index, index1; - int extend_x = 0, extend_y = 0; + int extend_x, extend_y; GetBitContext *gb = &v->s.gb; int bits, esc; int val, sign; - const int* offs_tab; if (v->numref) { bits = VC1_2REF_MVDATA_VLC_BITS; @@ -265,51 +275,32 @@ static av_always_inline void get_mvdata_interlaced(VC1Context *v, int *dmv_x, bits = VC1_1REF_MVDATA_VLC_BITS; esc = 71; } - switch (v->dmvrange) { - case 1: - extend_x = 1; - break; - case 2: - extend_y = 1; - break; - case 3: - extend_x = extend_y = 1; - break; - } + extend_x = v->dmvrange & 1; + extend_y = (v->dmvrange >> 1) & 1; index = get_vlc2(gb, v->imv_vlc->table, bits, 3); if (index == esc) { *dmv_x = get_bits(gb, v->k_x); *dmv_y = get_bits(gb, v->k_y); if (v->numref) { - if (pred_flag) { + if (pred_flag) *pred_flag = *dmv_y & 1; - *dmv_y = (*dmv_y + *pred_flag) >> 1; - } else { - *dmv_y = (*dmv_y + (*dmv_y & 1)) >> 1; - } + *dmv_y = (*dmv_y + (*dmv_y & 1)) >> 1; } } else { - if (extend_x) - offs_tab = offset_table2; - else - offs_tab = offset_table1; + av_assert0(index < esc); index1 = (index + 1) % 9; if (index1 != 0) { val = get_bits(gb, index1 + extend_x); - sign = 0 -(val & 1); - *dmv_x = (sign ^ ((val >> 1) + offs_tab[index1])) - sign; + sign = 0 - (val & 1); + *dmv_x = (sign ^ ((val >> 1) + offset_table[extend_x][index1])) - sign; } else *dmv_x = 0; - if (extend_y) - offs_tab = offset_table2; - else - offs_tab = offset_table1; index1 = (index + 1) / 9; if (index1 > v->numref) { - val = get_bits(gb, (index1 + (extend_y << v->numref)) >> v->numref); + val = get_bits(gb, (index1 >> v->numref) + extend_y); sign = 0 - (val & 1); - *dmv_y = (sign ^ ((val >> 1) + offs_tab[index1 >> v->numref])) - sign; + *dmv_y = (sign ^ ((val >> 1) + offset_table[extend_y][index1 >> v->numref])) - sign; } else *dmv_y = 0; if (v->numref && pred_flag) @@ -420,6 +411,12 @@ static inline int ff_vc1_pred_dc(MpegEncContext *s, int overlap, int pq, int n, int q1, q2 = 0; int dqscale_index; + /* scale predictors if needed */ + q1 = FFABS(s->current_picture.qscale_table[mb_pos]); + dqscale_index = s->y_dc_scale_table[q1] - 1; + if (dqscale_index < 0) + return 0; + wrap = s->block_wrap[n]; dc_val = s->dc_val[0] + s->block_index[n]; @@ -429,18 +426,14 @@ static inline int ff_vc1_pred_dc(MpegEncContext *s, int overlap, int pq, int n, c = dc_val[ - 1]; b = dc_val[ - 1 - wrap]; a = dc_val[ - wrap]; - /* scale predictors if needed */ - q1 = s->current_picture.qscale_table[mb_pos]; - dqscale_index = s->y_dc_scale_table[q1] - 1; - if (dqscale_index < 0) - return 0; + if (c_avail && (n != 1 && n != 3)) { - q2 = s->current_picture.qscale_table[mb_pos - 1]; + q2 = FFABS(s->current_picture.qscale_table[mb_pos - 1]); if (q2 && q2 != q1) c = (c * s->y_dc_scale_table[q2] * ff_vc1_dqscale[dqscale_index] + 0x20000) >> 18; } if (a_avail && (n != 2 && n != 3)) { - q2 = s->current_picture.qscale_table[mb_pos - s->mb_stride]; + q2 = FFABS(s->current_picture.qscale_table[mb_pos - s->mb_stride]); if (q2 && q2 != q1) a = (a * s->y_dc_scale_table[q2] * ff_vc1_dqscale[dqscale_index] + 0x20000) >> 18; } @@ -450,25 +443,17 @@ static inline int ff_vc1_pred_dc(MpegEncContext *s, int overlap, int pq, int n, off--; if (n != 2) off -= s->mb_stride; - q2 = s->current_picture.qscale_table[off]; + q2 = FFABS(s->current_picture.qscale_table[off]); if (q2 && q2 != q1) b = (b * s->y_dc_scale_table[q2] * ff_vc1_dqscale[dqscale_index] + 0x20000) >> 18; } - if (a_avail && c_avail) { - if (abs(a - b) <= abs(b - c)) { - pred = c; - *dir_ptr = 1; // left - } else { - pred = a; - *dir_ptr = 0; // top - } + if (c_avail && (!a_avail || abs(a - b) <= abs(b - c))) { + pred = c; + *dir_ptr = 1; // left } else if (a_avail) { pred = a; *dir_ptr = 0; // top - } else if (c_avail) { - pred = c; - *dir_ptr = 1; // left } else { pred = 0; *dir_ptr = 1; // left @@ -527,17 +512,16 @@ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip, int *value, int codingset) { GetBitContext *gb = &v->s.gb; - int index, escape, run = 0, level = 0, lst = 0; + int index, run, level, lst, sign; index = get_vlc2(gb, ff_vc1_ac_coeff_table[codingset].table, AC_VLC_BITS, 3); if (index != ff_vc1_ac_sizes[codingset] - 1) { run = vc1_index_decode_table[codingset][index][0]; level = vc1_index_decode_table[codingset][index][1]; lst = index >= vc1_last_decode_table[codingset] || get_bits_left(gb) < 0; - if (get_bits1(gb)) - level = -level; + sign = get_bits1(gb); } else { - escape = decode210(gb); + int escape = decode210(gb); if (escape != 2) { index = get_vlc2(gb, ff_vc1_ac_coeff_table[codingset].table, AC_VLC_BITS, 3); run = vc1_index_decode_table[codingset][index][0]; @@ -554,10 +538,8 @@ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip, else run += vc1_delta_run_table[codingset][level] + 1; } - if (get_bits1(gb)) - level = -level; + sign = get_bits1(gb); } else { - int sign; lst = get_bits1(gb); if (v->s.esc3_level_length == 0) { if (v->pq < 8 || v->dquantfrm) { // table 59 @@ -572,14 +554,12 @@ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip, run = get_bits(gb, v->s.esc3_run_length); sign = get_bits1(gb); level = get_bits(gb, v->s.esc3_level_length); - if (sign) - level = -level; } } *last = lst; *skip = run; - *value = level; + *value = (level ^ -sign) + sign; } /** Decode intra block in intra frames - should be faster than decode_intra_block @@ -598,7 +578,7 @@ static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n, int i; int16_t *dc_val; int16_t *ac_val, *ac_val2; - int dcdiff; + int dcdiff, scale; /* Get DC differential */ if (n < 4) { @@ -611,16 +591,12 @@ static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n, return -1; } if (dcdiff) { + const int m = (v->pq == 1 || v->pq == 2) ? 3 - v->pq : 0; if (dcdiff == 119 /* ESC index value */) { - /* TODO: Optimize */ - if (v->pq == 1) dcdiff = get_bits(gb, 10); - else if (v->pq == 2) dcdiff = get_bits(gb, 9); - else dcdiff = get_bits(gb, 8); + dcdiff = get_bits(gb, 8 + m); } else { - if (v->pq == 1) - dcdiff = (dcdiff << 2) + get_bits(gb, 2) - 3; - else if (v->pq == 2) - dcdiff = (dcdiff << 1) + get_bits1(gb) - 1; + if (m) + dcdiff = (dcdiff << m) + get_bits(gb, m) - ((1 << m) - 1); } if (get_bits1(gb)) dcdiff = -dcdiff; @@ -631,27 +607,29 @@ static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n, *dc_val = dcdiff; /* Store the quantized DC coeff, used for prediction */ - if (n < 4) { - block[0] = dcdiff * s->y_dc_scale; - } else { - block[0] = dcdiff * s->c_dc_scale; - } - /* Skip ? */ - if (!coded) { - goto not_coded; - } + if (n < 4) + scale = s->y_dc_scale; + else + scale = s->c_dc_scale; + block[0] = dcdiff * scale; - // AC Decoding - i = 1; + ac_val = s->ac_val[0][s->block_index[n]]; + ac_val2 = ac_val; + if (dc_pred_dir) // left + ac_val -= 16; + else // top + ac_val -= 16 * s->block_wrap[n]; + + scale = v->pq * 2 + v->halfpq; - { + //AC Decoding + i = !!coded; + + if (coded) { int last = 0, skip, value; const uint8_t *zz_table; - int scale; int k; - scale = v->pq * 2 + v->halfpq; - if (v->s.ac_pred) { if (!dc_pred_dir) zz_table = v->zz_8x8[2]; @@ -660,13 +638,6 @@ static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n, } else zz_table = v->zz_8x8[1]; - ac_val = s->ac_val[0][0] + s->block_index[n] * 16; - ac_val2 = ac_val; - if (dc_pred_dir) // left - ac_val -= 16; - else // top - ac_val -= 16 * s->block_wrap[n]; - while (!last) { vc1_decode_ac_coeff(v, &last, &skip, &value, codingset); i += skip; @@ -677,13 +648,15 @@ static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n, /* apply AC prediction if needed */ if (s->ac_pred) { + int sh; if (dc_pred_dir) { // left - for (k = 1; k < 8; k++) - block[k << v->left_blk_sh] += ac_val[k]; + sh = v->left_blk_sh; } else { // top - for (k = 1; k < 8; k++) - block[k << v->top_blk_sh] += ac_val[k + 8]; + sh = v->top_blk_sh; + ac_val += 8; } + for (k = 1; k < 8; k++) + block[k << sh] += ac_val[k]; } /* save AC coeffs for further prediction */ for (k = 1; k < 8; k++) { @@ -699,46 +672,30 @@ static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n, block[k] += (block[k] < 0) ? -v->pq : v->pq; } - if (s->ac_pred) i = 63; - } - -not_coded: - if (!coded) { - int k, scale; - ac_val = s->ac_val[0][0] + s->block_index[n] * 16; - ac_val2 = ac_val; + } else { + int k; - i = 0; - scale = v->pq * 2 + v->halfpq; memset(ac_val2, 0, 16 * 2); - if (dc_pred_dir) { // left - ac_val -= 16; - if (s->ac_pred) - memcpy(ac_val2, ac_val, 8 * 2); - } else { // top - ac_val -= 16 * s->block_wrap[n]; - if (s->ac_pred) - memcpy(ac_val2 + 8, ac_val + 8, 8 * 2); - } /* apply AC prediction if needed */ if (s->ac_pred) { + int sh; if (dc_pred_dir) { //left - for (k = 1; k < 8; k++) { - block[k << v->left_blk_sh] = ac_val[k] * scale; - if (!v->pquantizer && block[k << v->left_blk_sh]) - block[k << v->left_blk_sh] += (block[k << v->left_blk_sh] < 0) ? -v->pq : v->pq; - } + sh = v->left_blk_sh; } else { // top - for (k = 1; k < 8; k++) { - block[k << v->top_blk_sh] = ac_val[k + 8] * scale; - if (!v->pquantizer && block[k << v->top_blk_sh]) - block[k << v->top_blk_sh] += (block[k << v->top_blk_sh] < 0) ? -v->pq : v->pq; - } + sh = v->top_blk_sh; + ac_val += 8; + ac_val2 += 8; + } + memcpy(ac_val2, ac_val, 8 * 2); + for (k = 1; k < 8; k++) { + block[k << sh] = ac_val[k] * scale; + if (!v->pquantizer && block[k << sh]) + block[k << sh] += (block[k << sh] < 0) ? -v->pq : v->pq; } - i = 63; } } + if (s->ac_pred) i = 63; s->block_last_index[n] = i; return 0; @@ -759,7 +716,7 @@ static int vc1_decode_i_block_adv(VC1Context *v, int16_t block[64], int n, MpegEncContext *s = &v->s; int dc_pred_dir = 0; /* Direction of the DC prediction used */ int i; - int16_t *dc_val; + int16_t *dc_val = NULL; int16_t *ac_val, *ac_val2; int dcdiff; int a_avail = v->a_avail, c_avail = v->c_avail; @@ -767,6 +724,7 @@ static int vc1_decode_i_block_adv(VC1Context *v, int16_t block[64], int n, int scale; int q1, q2 = 0; int mb_pos = s->mb_x + s->mb_y * s->mb_stride; + int quant = FFABS(mquant); /* Get DC differential */ if (n < 4) { @@ -779,59 +737,58 @@ static int vc1_decode_i_block_adv(VC1Context *v, int16_t block[64], int n, return -1; } if (dcdiff) { + const int m = (quant == 1 || quant == 2) ? 3 - quant : 0; if (dcdiff == 119 /* ESC index value */) { - /* TODO: Optimize */ - if (mquant == 1) dcdiff = get_bits(gb, 10); - else if (mquant == 2) dcdiff = get_bits(gb, 9); - else dcdiff = get_bits(gb, 8); + dcdiff = get_bits(gb, 8 + m); } else { - if (mquant == 1) - dcdiff = (dcdiff << 2) + get_bits(gb, 2) - 3; - else if (mquant == 2) - dcdiff = (dcdiff << 1) + get_bits1(gb) - 1; + if (m) + dcdiff = (dcdiff << m) + get_bits(gb, m) - ((1 << m) - 1); } if (get_bits1(gb)) dcdiff = -dcdiff; } /* Prediction */ - dcdiff += ff_vc1_pred_dc(&v->s, v->overlap, mquant, n, v->a_avail, v->c_avail, &dc_val, &dc_pred_dir); + dcdiff += ff_vc1_pred_dc(&v->s, v->overlap, quant, n, v->a_avail, v->c_avail, &dc_val, &dc_pred_dir); *dc_val = dcdiff; /* Store the quantized DC coeff, used for prediction */ - if (n < 4) { - block[0] = dcdiff * s->y_dc_scale; - } else { - block[0] = dcdiff * s->c_dc_scale; - } - - //AC Decoding - i = 1; + if (n < 4) + scale = s->y_dc_scale; + else + scale = s->c_dc_scale; + block[0] = dcdiff * scale; /* check if AC is needed at all */ if (!a_avail && !c_avail) use_pred = 0; - ac_val = s->ac_val[0][0] + s->block_index[n] * 16; - ac_val2 = ac_val; - scale = mquant * 2 + ((mquant == v->pq) ? v->halfpq : 0); + scale = quant * 2 + ((mquant < 0) ? 0 : v->halfpq); + ac_val = s->ac_val[0][s->block_index[n]]; + ac_val2 = ac_val; if (dc_pred_dir) // left ac_val -= 16; else // top ac_val -= 16 * s->block_wrap[n]; q1 = s->current_picture.qscale_table[mb_pos]; - if (dc_pred_dir && c_avail && mb_pos) - q2 = s->current_picture.qscale_table[mb_pos - 1]; - if (!dc_pred_dir && a_avail && mb_pos >= s->mb_stride) - q2 = s->current_picture.qscale_table[mb_pos - s->mb_stride]; - if (dc_pred_dir && n == 1) - q2 = q1; - if (!dc_pred_dir && n == 2) - q2 = q1; if (n == 3) q2 = q1; + else if (dc_pred_dir) { + if (n == 1) + q2 = q1; + else if (c_avail && mb_pos) + q2 = s->current_picture.qscale_table[mb_pos - 1]; + } else { + if (n == 2) + q2 = q1; + else if (a_avail && mb_pos >= s->mb_stride) + q2 = s->current_picture.qscale_table[mb_pos - s->mb_stride]; + } + + //AC Decoding + i = 1; if (coded) { int last = 0, skip, value; @@ -864,28 +821,25 @@ static int vc1_decode_i_block_adv(VC1Context *v, int16_t block[64], int n, /* apply AC prediction if needed */ if (use_pred) { + int sh; + if (dc_pred_dir) { // left + sh = v->left_blk_sh; + } else { // top + sh = v->top_blk_sh; + ac_val += 8; + } /* scale predictors if needed*/ + q1 = FFABS(q1) * 2 + ((q1 < 0) ? 0 : v->halfpq) - 1; + if (q1 < 1) + return AVERROR_INVALIDDATA; + if (q2) + q2 = FFABS(q2) * 2 + ((q2 < 0) ? 0 : v->halfpq) - 1; if (q2 && q1 != q2) { - q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; - q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; - - if (q1 < 1) - return AVERROR_INVALIDDATA; - if (dc_pred_dir) { // left - for (k = 1; k < 8; k++) - block[k << v->left_blk_sh] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; - } else { // top - for (k = 1; k < 8; k++) - block[k << v->top_blk_sh] += (ac_val[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; - } + for (k = 1; k < 8; k++) + block[k << sh] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } else { - if (dc_pred_dir) { //left - for (k = 1; k < 8; k++) - block[k << v->left_blk_sh] += ac_val[k]; - } else { //top - for (k = 1; k < 8; k++) - block[k << v->top_blk_sh] += ac_val[k + 8]; - } + for (k = 1; k < 8; k++) + block[k << sh] += ac_val[k]; } } /* save AC coeffs for further prediction */ @@ -899,58 +853,42 @@ static int vc1_decode_i_block_adv(VC1Context *v, int16_t block[64], int n, if (block[k]) { block[k] *= scale; if (!v->pquantizer) - block[k] += (block[k] < 0) ? -mquant : mquant; + block[k] += (block[k] < 0) ? -quant : quant; } - if (use_pred) i = 63; } else { // no AC coeffs int k; memset(ac_val2, 0, 16 * 2); - if (dc_pred_dir) { // left - if (use_pred) { - memcpy(ac_val2, ac_val, 8 * 2); - if (q2 && q1 != q2) { - q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; - q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; - if (q1 < 1) - return AVERROR_INVALIDDATA; - for (k = 1; k < 8; k++) - ac_val2[k] = (ac_val2[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; - } - } - } else { // top - if (use_pred) { - memcpy(ac_val2 + 8, ac_val + 8, 8 * 2); - if (q2 && q1 != q2) { - q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; - q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; - if (q1 < 1) - return AVERROR_INVALIDDATA; - for (k = 1; k < 8; k++) - ac_val2[k + 8] = (ac_val2[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; - } - } - } /* apply AC prediction if needed */ if (use_pred) { + int sh; if (dc_pred_dir) { // left - for (k = 1; k < 8; k++) { - block[k << v->left_blk_sh] = ac_val2[k] * scale; - if (!v->pquantizer && block[k << v->left_blk_sh]) - block[k << v->left_blk_sh] += (block[k << v->left_blk_sh] < 0) ? -mquant : mquant; - } + sh = v->left_blk_sh; } else { // top - for (k = 1; k < 8; k++) { - block[k << v->top_blk_sh] = ac_val2[k + 8] * scale; - if (!v->pquantizer && block[k << v->top_blk_sh]) - block[k << v->top_blk_sh] += (block[k << v->top_blk_sh] < 0) ? -mquant : mquant; - } + sh = v->top_blk_sh; + ac_val += 8; + ac_val2 += 8; + } + memcpy(ac_val2, ac_val, 8 * 2); + q1 = FFABS(q1) * 2 + ((q1 < 0) ? 0 : v->halfpq) - 1; + if (q1 < 1) + return AVERROR_INVALIDDATA; + if (q2) + q2 = FFABS(q2) * 2 + ((q2 < 0) ? 0 : v->halfpq) - 1; + if (q2 && q1 != q2) { + for (k = 1; k < 8; k++) + ac_val2[k] = (ac_val2[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; + } + for (k = 1; k < 8; k++) { + block[k << sh] = ac_val2[k] * scale; + if (!v->pquantizer && block[k << sh]) + block[k << sh] += (block[k << sh] < 0) ? -quant : quant; } - i = 63; } } + if (use_pred) i = 63; s->block_last_index[n] = i; return 0; @@ -971,7 +909,7 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, MpegEncContext *s = &v->s; int dc_pred_dir = 0; /* Direction of the DC prediction used */ int i; - int16_t *dc_val; + int16_t *dc_val = NULL; int16_t *ac_val, *ac_val2; int dcdiff; int mb_pos = s->mb_x + s->mb_y * s->mb_stride; @@ -979,15 +917,16 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, int use_pred = s->ac_pred; int scale; int q1, q2 = 0; + int quant = FFABS(mquant); s->bdsp.clear_block(block); /* XXX: Guard against dumb values of mquant */ - mquant = (mquant < 1) ? 0 : ((mquant > 31) ? 31 : mquant); + quant = av_clip_uintp2(quant, 5); /* Set DC scale - y and c use the same */ - s->y_dc_scale = s->y_dc_scale_table[mquant]; - s->c_dc_scale = s->c_dc_scale_table[mquant]; + s->y_dc_scale = s->y_dc_scale_table[quant]; + s->c_dc_scale = s->c_dc_scale_table[quant]; /* Get DC differential */ if (n < 4) { @@ -1000,23 +939,19 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, return -1; } if (dcdiff) { + const int m = (quant == 1 || quant == 2) ? 3 - quant : 0; if (dcdiff == 119 /* ESC index value */) { - /* TODO: Optimize */ - if (mquant == 1) dcdiff = get_bits(gb, 10); - else if (mquant == 2) dcdiff = get_bits(gb, 9); - else dcdiff = get_bits(gb, 8); + dcdiff = get_bits(gb, 8 + m); } else { - if (mquant == 1) - dcdiff = (dcdiff << 2) + get_bits(gb, 2) - 3; - else if (mquant == 2) - dcdiff = (dcdiff << 1) + get_bits1(gb) - 1; + if (m) + dcdiff = (dcdiff << m) + get_bits(gb, m) - ((1 << m) - 1); } if (get_bits1(gb)) dcdiff = -dcdiff; } /* Prediction */ - dcdiff += ff_vc1_pred_dc(&v->s, v->overlap, mquant, n, a_avail, c_avail, &dc_val, &dc_pred_dir); + dcdiff += ff_vc1_pred_dc(&v->s, v->overlap, quant, n, a_avail, c_avail, &dc_val, &dc_pred_dir); *dc_val = dcdiff; /* Store the quantized DC coeff, used for prediction */ @@ -1034,10 +969,10 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, if (!a_avail) dc_pred_dir = 1; if (!c_avail) dc_pred_dir = 0; if (!a_avail && !c_avail) use_pred = 0; - ac_val = s->ac_val[0][0] + s->block_index[n] * 16; + ac_val = s->ac_val[0][s->block_index[n]]; ac_val2 = ac_val; - scale = mquant * 2 + v->halfpq; + scale = quant * 2 + ((mquant < 0) ? 0 : v->halfpq); if (dc_pred_dir) //left ac_val -= 16; @@ -1081,12 +1016,12 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, /* apply AC prediction if needed */ if (use_pred) { /* scale predictors if needed*/ + q1 = FFABS(q1) * 2 + ((q1 < 0) ? 0 : v->halfpq) - 1; + if (q1 < 1) + return AVERROR_INVALIDDATA; + if (q2) + q2 = FFABS(q2) * 2 + ((q2 < 0) ? 0 : v->halfpq) - 1; if (q2 && q1 != q2) { - q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; - q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; - - if (q1 < 1) - return AVERROR_INVALIDDATA; if (dc_pred_dir) { // left for (k = 1; k < 8; k++) block[k << v->left_blk_sh] += (ac_val[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; @@ -1115,7 +1050,7 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, if (block[k]) { block[k] *= scale; if (!v->pquantizer) - block[k] += (block[k] < 0) ? -mquant : mquant; + block[k] += (block[k] < 0) ? -quant : quant; } if (use_pred) i = 63; @@ -1126,11 +1061,12 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, if (dc_pred_dir) { // left if (use_pred) { memcpy(ac_val2, ac_val, 8 * 2); + q1 = FFABS(q1) * 2 + ((q1 < 0) ? 0 : v->halfpq) - 1; + if (q1 < 1) + return AVERROR_INVALIDDATA; + if (q2) + q2 = FFABS(q2) * 2 + ((q2 < 0) ? 0 : v->halfpq) - 1; if (q2 && q1 != q2) { - q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; - q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; - if (q1 < 1) - return AVERROR_INVALIDDATA; for (k = 1; k < 8; k++) ac_val2[k] = (ac_val2[k] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } @@ -1138,11 +1074,12 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, } else { // top if (use_pred) { memcpy(ac_val2 + 8, ac_val + 8, 8 * 2); + q1 = FFABS(q1) * 2 + ((q1 < 0) ? 0 : v->halfpq) - 1; + if (q1 < 1) + return AVERROR_INVALIDDATA; + if (q2) + q2 = FFABS(q2) * 2 + ((q2 < 0) ? 0 : v->halfpq) - 1; if (q2 && q1 != q2) { - q1 = q1 * 2 + ((q1 == v->pq) ? v->halfpq : 0) - 1; - q2 = q2 * 2 + ((q2 == v->pq) ? v->halfpq : 0) - 1; - if (q1 < 1) - return AVERROR_INVALIDDATA; for (k = 1; k < 8; k++) ac_val2[k + 8] = (ac_val2[k + 8] * q2 * ff_vc1_dqscale[q1 - 1] + 0x20000) >> 18; } @@ -1155,13 +1092,13 @@ static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n, for (k = 1; k < 8; k++) { block[k << v->left_blk_sh] = ac_val2[k] * scale; if (!v->pquantizer && block[k << v->left_blk_sh]) - block[k << v->left_blk_sh] += (block[k << v->left_blk_sh] < 0) ? -mquant : mquant; + block[k << v->left_blk_sh] += (block[k << v->left_blk_sh] < 0) ? -quant : quant; } } else { // top for (k = 1; k < 8; k++) { block[k << v->top_blk_sh] = ac_val2[k + 8] * scale; if (!v->pquantizer && block[k << v->top_blk_sh]) - block[k << v->top_blk_sh] += (block[k << v->top_blk_sh] < 0) ? -mquant : mquant; + block[k << v->top_blk_sh] += (block[k << v->top_blk_sh] < 0) ? -quant : quant; } } i = 63; @@ -1186,6 +1123,7 @@ static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n, int scale, off, idx, last, skip, value; int ttblk = ttmb & 7; int pat = 0; + int quant = FFABS(mquant); s->bdsp.clear_block(block); @@ -1206,7 +1144,7 @@ static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n, if (ttblk == TT_4X8_RIGHT || ttblk == TT_4X8_LEFT) ttblk = TT_4X8; } - scale = 2 * mquant + ((v->pq == mquant) ? v->halfpq : 0); + scale = quant * 2 + ((mquant < 0) ? 0 : v->halfpq); // convert transforms like 8X4_TOP to generic TT and SUBBLKPAT if (ttblk == TT_8X4_TOP || ttblk == TT_8X4_BOTTOM) { @@ -1233,7 +1171,7 @@ static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n, idx = v->zzi_8x8[i++]; block[idx] = value * scale; if (!v->pquantizer) - block[idx] += (block[idx] < 0) ? -mquant : mquant; + block[idx] += (block[idx] < 0) ? -quant : quant; } if (!skip_block) { if (i == 1) @@ -1261,7 +1199,7 @@ static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n, idx = ff_vc1_adv_interlaced_4x4_zz[i++]; block[idx + off] = value * scale; if (!v->pquantizer) - block[idx + off] += (block[idx + off] < 0) ? -mquant : mquant; + block[idx + off] += (block[idx + off] < 0) ? -quant : quant; } if (!(subblkpat & (1 << (3 - j))) && !skip_block) { if (i == 1) @@ -1288,7 +1226,7 @@ static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n, idx = ff_vc1_adv_interlaced_8x4_zz[i++] + off; block[idx] = value * scale; if (!v->pquantizer) - block[idx] += (block[idx] < 0) ? -mquant : mquant; + block[idx] += (block[idx] < 0) ? -quant : quant; } if (!(subblkpat & (1 << (1 - j))) && !skip_block) { if (i == 1) @@ -1315,7 +1253,7 @@ static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n, idx = ff_vc1_adv_interlaced_4x8_zz[i++] + off; block[idx] = value * scale; if (!v->pquantizer) - block[idx] += (block[idx] < 0) ? -mquant : mquant; + block[idx] += (block[idx] < 0) ? -quant : quant; } if (!(subblkpat & (1 << (1 - j))) && !skip_block) { if (i == 1) @@ -1333,8 +1271,7 @@ static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n, /** @} */ // Macroblock group -static const int size_table [6] = { 0, 2, 3, 4, 5, 8 }; -static const int offset_table[6] = { 0, 1, 3, 7, 15, 31 }; +static const uint8_t size_table[6] = { 0, 2, 3, 4, 5, 8 }; /** Decode one P-frame MB */ @@ -1414,30 +1351,20 @@ static int vc1_decode_p_mb(VC1Context *v) if (i == 1 || i == 3 || s->mb_x) v->c_avail = v->mb_type[0][s->block_index[i] - 1]; - vc1_decode_intra_block(v, s->block[i], i, val, mquant, + vc1_decode_intra_block(v, v->block[v->cur_blk_idx][block_map[i]], i, val, mquant, (i & 4) ? v->codingset2 : v->codingset); - if ((i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && (i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; - v->vc1dsp.vc1_inv_trans_8x8(s->block[i]); + v->vc1dsp.vc1_inv_trans_8x8(v->block[v->cur_blk_idx][block_map[i]]); if (v->rangeredfrm) for (j = 0; j < 64; j++) - s->block[i][j] <<= 1; - s->idsp.put_signed_pixels_clamped(s->block[i], - s->dest[dst_idx] + off, - i & 4 ? s->uvlinesize - : s->linesize); - if (v->pq >= 9 && v->overlap) { - if (v->c_avail) - v->vc1dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize); - if (v->a_avail) - v->vc1dsp.vc1_v_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize); - } + v->block[v->cur_blk_idx][block_map[i]][j] <<= 1; block_cbp |= 0xF << (i << 2); block_intra |= 1 << i; } else if (val) { - pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, + pat = vc1_decode_p_block(v, v->block[v->cur_blk_idx][block_map[i]], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i & 4) ? s->uvlinesize : s->linesize, - (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); + CONFIG_GRAY && (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); block_cbp |= pat << (i << 2); if (!v->ttmbf && ttmb < 8) ttmb = -1; @@ -1525,31 +1452,21 @@ static int vc1_decode_p_mb(VC1Context *v) if (i == 1 || i == 3 || s->mb_x) v->c_avail = v->mb_type[0][s->block_index[i] - 1]; - vc1_decode_intra_block(v, s->block[i], i, is_coded[i], mquant, + vc1_decode_intra_block(v, v->block[v->cur_blk_idx][block_map[i]], i, is_coded[i], mquant, (i & 4) ? v->codingset2 : v->codingset); - if ((i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && (i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; - v->vc1dsp.vc1_inv_trans_8x8(s->block[i]); + v->vc1dsp.vc1_inv_trans_8x8(v->block[v->cur_blk_idx][block_map[i]]); if (v->rangeredfrm) for (j = 0; j < 64; j++) - s->block[i][j] <<= 1; - s->idsp.put_signed_pixels_clamped(s->block[i], - s->dest[dst_idx] + off, - (i & 4) ? s->uvlinesize - : s->linesize); - if (v->pq >= 9 && v->overlap) { - if (v->c_avail) - v->vc1dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize); - if (v->a_avail) - v->vc1dsp.vc1_v_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize); - } + v->block[v->cur_blk_idx][block_map[i]][j] <<= 1; block_cbp |= 0xF << (i << 2); block_intra |= 1 << i; } else if (is_coded[i]) { - pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, + pat = vc1_decode_p_block(v, v->block[v->cur_blk_idx][block_map[i]], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i & 4) ? s->uvlinesize : s->linesize, - (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), + CONFIG_GRAY && (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); block_cbp |= pat << (i << 2); if (!v->ttmbf && ttmb < 8) @@ -1573,6 +1490,10 @@ static int vc1_decode_p_mb(VC1Context *v) } } end: + if (v->overlap && v->pq >= 9) + ff_vc1_p_overlap_filter(v); + vc1_put_blocks_clamped(v, 1); + v->cbp[s->mb_x] = block_cbp; v->ttblk[s->mb_x] = block_tt; v->is_intra[s->mb_x] = block_intra; @@ -1600,7 +1521,7 @@ static int vc1_decode_p_mb_intfr(VC1Context *v) int skipped, fourmv = 0, twomv = 0; int block_cbp = 0, pat, block_tt = 0; int idx_mbmode = 0, mvbp; - int stride_y, fieldtx; + int fieldtx; mquant = v->pq; /* Lossy initialization */ @@ -1659,8 +1580,8 @@ static int vc1_decode_p_mb_intfr(VC1Context *v) GET_MQUANT(); s->current_picture.qscale_table[mb_pos] = mquant; /* Set DC scale - y and c use the same (not sure if necessary here) */ - s->y_dc_scale = s->y_dc_scale_table[mquant]; - s->c_dc_scale = s->c_dc_scale_table[mquant]; + s->y_dc_scale = s->y_dc_scale_table[FFABS(mquant)]; + s->c_dc_scale = s->c_dc_scale_table[FFABS(mquant)]; dst_idx = 0; for (i = 0; i < 6; i++) { v->a_avail = v->c_avail = 0; @@ -1673,22 +1594,16 @@ static int vc1_decode_p_mb_intfr(VC1Context *v) if (i == 1 || i == 3 || s->mb_x) v->c_avail = v->mb_type[0][s->block_index[i] - 1]; - vc1_decode_intra_block(v, s->block[i], i, val, mquant, + vc1_decode_intra_block(v, v->block[v->cur_blk_idx][block_map[i]], i, val, mquant, (i & 4) ? v->codingset2 : v->codingset); - if ((i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && (i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; - v->vc1dsp.vc1_inv_trans_8x8(s->block[i]); - if (i < 4) { - stride_y = s->linesize << fieldtx; + v->vc1dsp.vc1_inv_trans_8x8(v->block[v->cur_blk_idx][block_map[i]]); + if (i < 4) off = (fieldtx) ? ((i & 1) * 8) + ((i & 2) >> 1) * s->linesize : (i & 1) * 8 + 4 * (i & 2) * s->linesize; - } else { - stride_y = s->uvlinesize; + else off = 0; - } - s->idsp.put_signed_pixels_clamped(s->block[i], - s->dest[dst_idx] + off, - stride_y); - //TODO: loop filter + block_cbp |= 0xf << (i << 2); } } else { // inter MB @@ -1711,19 +1626,14 @@ static int vc1_decode_p_mb_intfr(VC1Context *v) dst_idx = 0; if (fourmv) { mvbp = v->fourmvbp; - for (i = 0; i < 6; i++) { - if (i < 4) { - dmv_x = dmv_y = 0; - val = ((mvbp >> (3 - i)) & 1); - if (val) { - get_mvdata_interlaced(v, &dmv_x, &dmv_y, 0); - } - ff_vc1_pred_mv_intfr(v, i, dmv_x, dmv_y, 0, v->range_x, v->range_y, v->mb_type[0], 0); - ff_vc1_mc_4mv_luma(v, i, 0, 0); - } else if (i == 4) { - ff_vc1_mc_4mv_chroma4(v, 0, 0, 0); - } + for (i = 0; i < 4; i++) { + dmv_x = dmv_y = 0; + if (mvbp & (8 >> i)) + get_mvdata_interlaced(v, &dmv_x, &dmv_y, 0); + ff_vc1_pred_mv_intfr(v, i, dmv_x, dmv_y, 0, v->range_x, v->range_y, v->mb_type[0], 0); + ff_vc1_mc_4mv_luma(v, i, 0, 0); } + ff_vc1_mc_4mv_chroma4(v, 0, 0, 0); } else if (twomv) { mvbp = v->twomvbp; dmv_x = dmv_y = 0; @@ -1764,10 +1674,10 @@ static int vc1_decode_p_mb_intfr(VC1Context *v) else off = (i & 4) ? 0 : ((i & 1) * 8 + ((i > 1) * s->linesize)); if (val) { - pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, + pat = vc1_decode_p_block(v, v->block[v->cur_blk_idx][block_map[i]], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i & 4) ? s->uvlinesize : (s->linesize << fieldtx), - (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); + CONFIG_GRAY && (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); block_cbp |= pat << (i << 2); if (!v->ttmbf && ttmb < 8) ttmb = -1; @@ -1789,9 +1699,15 @@ static int vc1_decode_p_mb_intfr(VC1Context *v) v->blk_mv_type[s->block_index[3]] = 0; ff_vc1_pred_mv_intfr(v, 0, 0, 0, 1, v->range_x, v->range_y, v->mb_type[0], 0); ff_vc1_mc_1mv(v, 0); + v->fieldtx_plane[mb_pos] = 0; } - if (s->mb_x == s->mb_width - 1) - memmove(v->is_intra_base, v->is_intra, sizeof(v->is_intra_base[0])*s->mb_stride); + if (v->overlap && v->pq >= 9) + ff_vc1_p_overlap_filter(v); + vc1_put_blocks_clamped(v, 1); + + v->cbp[s->mb_x] = block_cbp; + v->ttblk[s->mb_x] = block_tt; + return 0; } @@ -1810,7 +1726,7 @@ static int vc1_decode_p_mb_intfi(VC1Context *v) int val; /* temp values */ int first_block = 1; int dst_idx, off; - int pred_flag; + int pred_flag = 0; int block_cbp = 0, pat, block_tt = 0; int idx_mbmode = 0; @@ -1826,8 +1742,8 @@ static int vc1_decode_p_mb_intfi(VC1Context *v) GET_MQUANT(); s->current_picture.qscale_table[mb_pos] = mquant; /* Set DC scale - y and c use the same (not sure if necessary here) */ - s->y_dc_scale = s->y_dc_scale_table[mquant]; - s->c_dc_scale = s->c_dc_scale_table[mquant]; + s->y_dc_scale = s->y_dc_scale_table[FFABS(mquant)]; + s->c_dc_scale = s->c_dc_scale_table[FFABS(mquant)]; v->s.ac_pred = v->acpred_plane[mb_pos] = get_bits1(gb); mb_has_coeffs = idx_mbmode & 1; if (mb_has_coeffs) @@ -1844,22 +1760,19 @@ static int vc1_decode_p_mb_intfi(VC1Context *v) if (i == 1 || i == 3 || s->mb_x) v->c_avail = v->mb_type[0][s->block_index[i] - 1]; - vc1_decode_intra_block(v, s->block[i], i, val, mquant, + vc1_decode_intra_block(v, v->block[v->cur_blk_idx][block_map[i]], i, val, mquant, (i & 4) ? v->codingset2 : v->codingset); - if ((i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && (i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; - v->vc1dsp.vc1_inv_trans_8x8(s->block[i]); + v->vc1dsp.vc1_inv_trans_8x8(v->block[v->cur_blk_idx][block_map[i]]); off = (i & 4) ? 0 : ((i & 1) * 8 + (i & 2) * 4 * s->linesize); - s->idsp.put_signed_pixels_clamped(s->block[i], - s->dest[dst_idx] + off, - (i & 4) ? s->uvlinesize - : s->linesize); - // TODO: loop filter + block_cbp |= 0xf << (i << 2); } } else { s->mb_intra = v->is_intra[s->mb_x] = 0; s->current_picture.mb_type[mb_pos + v->mb_off] = MB_TYPE_16x16; - for (i = 0; i < 6; i++) v->mb_type[0][s->block_index[i]] = 0; + for (i = 0; i < 6; i++) + v->mb_type[0][s->block_index[i]] = 0; if (idx_mbmode <= 5) { // 1-MV dmv_x = dmv_y = pred_flag = 0; if (idx_mbmode & 1) { @@ -1870,18 +1783,14 @@ static int vc1_decode_p_mb_intfi(VC1Context *v) mb_has_coeffs = !(idx_mbmode & 2); } else { // 4-MV v->fourmvbp = get_vlc2(gb, v->fourmvbp_vlc->table, VC1_4MV_BLOCK_PATTERN_VLC_BITS, 1); - for (i = 0; i < 6; i++) { - if (i < 4) { - dmv_x = dmv_y = pred_flag = 0; - val = ((v->fourmvbp >> (3 - i)) & 1); - if (val) { - get_mvdata_interlaced(v, &dmv_x, &dmv_y, &pred_flag); - } - ff_vc1_pred_mv(v, i, dmv_x, dmv_y, 0, v->range_x, v->range_y, v->mb_type[0], pred_flag, 0); - ff_vc1_mc_4mv_luma(v, i, 0, 0); - } else if (i == 4) - ff_vc1_mc_4mv_chroma(v, 0); + for (i = 0; i < 4; i++) { + dmv_x = dmv_y = pred_flag = 0; + if (v->fourmvbp & (8 >> i)) + get_mvdata_interlaced(v, &dmv_x, &dmv_y, &pred_flag); + ff_vc1_pred_mv(v, i, dmv_x, dmv_y, 0, v->range_x, v->range_y, v->mb_type[0], pred_flag, 0); + ff_vc1_mc_4mv_luma(v, i, 0, 0); } + ff_vc1_mc_4mv_chroma(v, 0); mb_has_coeffs = idx_mbmode & 1; } if (mb_has_coeffs) @@ -1900,19 +1809,25 @@ static int vc1_decode_p_mb_intfi(VC1Context *v) val = ((cbp >> (5 - i)) & 1); off = (i & 4) ? 0 : (i & 1) * 8 + (i & 2) * 4 * s->linesize; if (val) { - pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, + pat = vc1_decode_p_block(v, v->block[v->cur_blk_idx][block_map[i]], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i & 4) ? s->uvlinesize : s->linesize, - (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), + CONFIG_GRAY && (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); block_cbp |= pat << (i << 2); - if (!v->ttmbf && ttmb < 8) ttmb = -1; + if (!v->ttmbf && ttmb < 8) + ttmb = -1; first_block = 0; } } } - if (s->mb_x == s->mb_width - 1) - memmove(v->is_intra_base, v->is_intra, sizeof(v->is_intra_base[0]) * s->mb_stride); + if (v->overlap && v->pq >= 9) + ff_vc1_p_overlap_filter(v); + vc1_put_blocks_clamped(v, 1); + + v->cbp[s->mb_x] = block_cbp; + v->ttblk[s->mb_x] = block_tt; + return 0; } @@ -2049,7 +1964,7 @@ static void vc1_decode_b_mb(VC1Context *v) vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i & 4) ? v->codingset2 : v->codingset); - if ((i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && (i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; v->vc1dsp.vc1_inv_trans_8x8(s->block[i]); if (v->rangeredfrm) @@ -2063,7 +1978,7 @@ static void vc1_decode_b_mb(VC1Context *v) vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i & 4) ? s->uvlinesize : s->linesize, - (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), NULL); + CONFIG_GRAY && (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), NULL); if (!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; @@ -2089,7 +2004,8 @@ static void vc1_decode_b_mb_intfi(VC1Context *v) int fwd; int dmv_x[2], dmv_y[2], pred_flag[2]; int bmvtype = BMV_TYPE_BACKWARD; - int idx_mbmode, interpmvp; + int block_cbp = 0, pat, block_tt = 0; + int idx_mbmode; mquant = v->pq; /* Lossy initialization */ s->mb_intra = 0; @@ -2104,8 +2020,8 @@ static void vc1_decode_b_mb_intfi(VC1Context *v) GET_MQUANT(); s->current_picture.qscale_table[mb_pos] = mquant; /* Set DC scale - y and c use the same (not sure if necessary here) */ - s->y_dc_scale = s->y_dc_scale_table[mquant]; - s->c_dc_scale = s->c_dc_scale_table[mquant]; + s->y_dc_scale = s->y_dc_scale_table[FFABS(mquant)]; + s->c_dc_scale = s->c_dc_scale_table[FFABS(mquant)]; v->s.ac_pred = v->acpred_plane[mb_pos] = get_bits1(gb); mb_has_coeffs = idx_mbmode & 1; if (mb_has_coeffs) @@ -2124,7 +2040,7 @@ static void vc1_decode_b_mb_intfi(VC1Context *v) vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i & 4) ? v->codingset2 : v->codingset); - if ((i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && (i > 3) && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; v->vc1dsp.vc1_inv_trans_8x8(s->block[i]); if (v->rangeredfrm) @@ -2135,17 +2051,18 @@ static void vc1_decode_b_mb_intfi(VC1Context *v) s->dest[dst_idx] + off, (i & 4) ? s->uvlinesize : s->linesize); - // TODO: yet to perform loop filter } } else { s->mb_intra = v->is_intra[s->mb_x] = 0; s->current_picture.mb_type[mb_pos + v->mb_off] = MB_TYPE_16x16; - for (i = 0; i < 6; i++) v->mb_type[0][s->block_index[i]] = 0; + for (i = 0; i < 6; i++) + v->mb_type[0][s->block_index[i]] = 0; if (v->fmb_is_raw) fwd = v->forward_mb_plane[mb_pos] = get_bits1(gb); else fwd = v->forward_mb_plane[mb_pos]; if (idx_mbmode <= 5) { // 1-MV + int interpmvp = 0; dmv_x[0] = dmv_x[1] = dmv_y[0] = dmv_y[1] = 0; pred_flag[0] = pred_flag[1] = 0; if (fwd) @@ -2168,12 +2085,16 @@ static void vc1_decode_b_mb_intfi(VC1Context *v) if (bmvtype != BMV_TYPE_DIRECT && idx_mbmode & 1) { get_mvdata_interlaced(v, &dmv_x[bmvtype == BMV_TYPE_BACKWARD], &dmv_y[bmvtype == BMV_TYPE_BACKWARD], &pred_flag[bmvtype == BMV_TYPE_BACKWARD]); } - if (bmvtype == BMV_TYPE_INTERPOLATED && interpmvp) { + if (interpmvp) { get_mvdata_interlaced(v, &dmv_x[1], &dmv_y[1], &pred_flag[1]); } if (bmvtype == BMV_TYPE_DIRECT) { dmv_x[0] = dmv_y[0] = pred_flag[0] = 0; dmv_x[1] = dmv_y[1] = pred_flag[0] = 0; + if (!s->next_picture_ptr->field_picture) { + av_log(s->avctx, AV_LOG_ERROR, "Mixed field/frame direct mode not supported\n"); + return; + } } ff_vc1_pred_b_mv_intfi(v, 0, dmv_x, dmv_y, 1, pred_flag); vc1_b_mc(v, dmv_x, dmv_y, (bmvtype == BMV_TYPE_DIRECT), bmvtype); @@ -2183,21 +2104,18 @@ static void vc1_decode_b_mb_intfi(VC1Context *v) bmvtype = BMV_TYPE_FORWARD; v->bmvtype = bmvtype; v->fourmvbp = get_vlc2(gb, v->fourmvbp_vlc->table, VC1_4MV_BLOCK_PATTERN_VLC_BITS, 1); - for (i = 0; i < 6; i++) { - if (i < 4) { - dmv_x[0] = dmv_y[0] = pred_flag[0] = 0; - dmv_x[1] = dmv_y[1] = pred_flag[1] = 0; - val = ((v->fourmvbp >> (3 - i)) & 1); - if (val) { - get_mvdata_interlaced(v, &dmv_x[bmvtype == BMV_TYPE_BACKWARD], - &dmv_y[bmvtype == BMV_TYPE_BACKWARD], - &pred_flag[bmvtype == BMV_TYPE_BACKWARD]); - } - ff_vc1_pred_b_mv_intfi(v, i, dmv_x, dmv_y, 0, pred_flag); - ff_vc1_mc_4mv_luma(v, i, bmvtype == BMV_TYPE_BACKWARD, 0); - } else if (i == 4) - ff_vc1_mc_4mv_chroma(v, bmvtype == BMV_TYPE_BACKWARD); + for (i = 0; i < 4; i++) { + dmv_x[0] = dmv_y[0] = pred_flag[0] = 0; + dmv_x[1] = dmv_y[1] = pred_flag[1] = 0; + if (v->fourmvbp & (8 >> i)) { + get_mvdata_interlaced(v, &dmv_x[bmvtype == BMV_TYPE_BACKWARD], + &dmv_y[bmvtype == BMV_TYPE_BACKWARD], + &pred_flag[bmvtype == BMV_TYPE_BACKWARD]); + } + ff_vc1_pred_b_mv_intfi(v, i, dmv_x, dmv_y, 0, pred_flag); + ff_vc1_mc_4mv_luma(v, i, bmvtype == BMV_TYPE_BACKWARD, 0); } + ff_vc1_mc_4mv_chroma(v, bmvtype == BMV_TYPE_BACKWARD); mb_has_coeffs = idx_mbmode & 1; } if (mb_has_coeffs) @@ -2216,16 +2134,19 @@ static void vc1_decode_b_mb_intfi(VC1Context *v) val = ((cbp >> (5 - i)) & 1); off = (i & 4) ? 0 : (i & 1) * 8 + (i & 2) * 4 * s->linesize; if (val) { - vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, - first_block, s->dest[dst_idx] + off, - (i & 4) ? s->uvlinesize : s->linesize, - (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), NULL); + pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, + first_block, s->dest[dst_idx] + off, + (i & 4) ? s->uvlinesize : s->linesize, + CONFIG_GRAY && (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); + block_cbp |= pat << (i << 2); if (!v->ttmbf && ttmb < 8) ttmb = -1; first_block = 0; } } } + v->cbp[s->mb_x] = block_cbp; + v->ttblk[s->mb_x] = block_tt; } /** Decode one B-frame MB (in interlaced frame B picture) @@ -2275,39 +2196,6 @@ static int vc1_decode_b_mb_intfr(VC1Context *v) } } - if (v->dmb_is_raw) - direct = get_bits1(gb); - else - direct = v->direct_mb_plane[mb_pos]; - - if (direct) { - s->mv[0][0][0] = s->current_picture.motion_val[0][s->block_index[0]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][0], v->bfraction, 0, s->quarter_sample); - s->mv[0][0][1] = s->current_picture.motion_val[0][s->block_index[0]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][1], v->bfraction, 0, s->quarter_sample); - s->mv[1][0][0] = s->current_picture.motion_val[1][s->block_index[0]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][0], v->bfraction, 1, s->quarter_sample); - s->mv[1][0][1] = s->current_picture.motion_val[1][s->block_index[0]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][1], v->bfraction, 1, s->quarter_sample); - - if (twomv) { - s->mv[0][2][0] = s->current_picture.motion_val[0][s->block_index[2]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][0], v->bfraction, 0, s->quarter_sample); - s->mv[0][2][1] = s->current_picture.motion_val[0][s->block_index[2]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][1], v->bfraction, 0, s->quarter_sample); - s->mv[1][2][0] = s->current_picture.motion_val[1][s->block_index[2]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][0], v->bfraction, 1, s->quarter_sample); - s->mv[1][2][1] = s->current_picture.motion_val[1][s->block_index[2]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][1], v->bfraction, 1, s->quarter_sample); - - for (i = 1; i < 4; i += 2) { - s->mv[0][i][0] = s->current_picture.motion_val[0][s->block_index[i]][0] = s->mv[0][i-1][0]; - s->mv[0][i][1] = s->current_picture.motion_val[0][s->block_index[i]][1] = s->mv[0][i-1][1]; - s->mv[1][i][0] = s->current_picture.motion_val[1][s->block_index[i]][0] = s->mv[1][i-1][0]; - s->mv[1][i][1] = s->current_picture.motion_val[1][s->block_index[i]][1] = s->mv[1][i-1][1]; - } - } else { - for (i = 1; i < 4; i++) { - s->mv[0][i][0] = s->current_picture.motion_val[0][s->block_index[i]][0] = s->mv[0][0][0]; - s->mv[0][i][1] = s->current_picture.motion_val[0][s->block_index[i]][1] = s->mv[0][0][1]; - s->mv[1][i][0] = s->current_picture.motion_val[1][s->block_index[i]][0] = s->mv[1][0][0]; - s->mv[1][i][1] = s->current_picture.motion_val[1][s->block_index[i]][1] = s->mv[1][0][1]; - } - } - } - if (ff_vc1_mbmode_intfrp[0][idx_mbmode][0] == MV_PMODE_INTFR_INTRA) { // intra MB for (i = 0; i < 4; i++) { s->mv[0][i][0] = s->current_picture.motion_val[0][s->block_index[i]][0] = 0; @@ -2326,8 +2214,8 @@ static int vc1_decode_b_mb_intfr(VC1Context *v) GET_MQUANT(); s->current_picture.qscale_table[mb_pos] = mquant; /* Set DC scale - y and c use the same (not sure if necessary here) */ - s->y_dc_scale = s->y_dc_scale_table[mquant]; - s->c_dc_scale = s->c_dc_scale_table[mquant]; + s->y_dc_scale = s->y_dc_scale_table[FFABS(mquant)]; + s->c_dc_scale = s->c_dc_scale_table[FFABS(mquant)]; dst_idx = 0; for (i = 0; i < 6; i++) { v->a_avail = v->c_avail = 0; @@ -2342,7 +2230,7 @@ static int vc1_decode_b_mb_intfr(VC1Context *v) vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i & 4) ? v->codingset2 : v->codingset); - if (i > 3 && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && i > 3 && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; v->vc1dsp.vc1_inv_trans_8x8(s->block[i]); if (i < 4) { @@ -2358,6 +2246,42 @@ static int vc1_decode_b_mb_intfr(VC1Context *v) } } else { s->mb_intra = v->is_intra[s->mb_x] = 0; + + if (v->dmb_is_raw) + direct = get_bits1(gb); + else + direct = v->direct_mb_plane[mb_pos]; + + if (direct) { + if (s->next_picture_ptr->field_picture) + av_log(s->avctx, AV_LOG_WARNING, "Mixed frame/field direct mode not supported\n"); + s->mv[0][0][0] = s->current_picture.motion_val[0][s->block_index[0]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][0], v->bfraction, 0, s->quarter_sample); + s->mv[0][0][1] = s->current_picture.motion_val[0][s->block_index[0]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][1], v->bfraction, 0, s->quarter_sample); + s->mv[1][0][0] = s->current_picture.motion_val[1][s->block_index[0]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][0], v->bfraction, 1, s->quarter_sample); + s->mv[1][0][1] = s->current_picture.motion_val[1][s->block_index[0]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[0]][1], v->bfraction, 1, s->quarter_sample); + + if (twomv) { + s->mv[0][2][0] = s->current_picture.motion_val[0][s->block_index[2]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][0], v->bfraction, 0, s->quarter_sample); + s->mv[0][2][1] = s->current_picture.motion_val[0][s->block_index[2]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][1], v->bfraction, 0, s->quarter_sample); + s->mv[1][2][0] = s->current_picture.motion_val[1][s->block_index[2]][0] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][0], v->bfraction, 1, s->quarter_sample); + s->mv[1][2][1] = s->current_picture.motion_val[1][s->block_index[2]][1] = scale_mv(s->next_picture.motion_val[1][s->block_index[2]][1], v->bfraction, 1, s->quarter_sample); + + for (i = 1; i < 4; i += 2) { + s->mv[0][i][0] = s->current_picture.motion_val[0][s->block_index[i]][0] = s->mv[0][i-1][0]; + s->mv[0][i][1] = s->current_picture.motion_val[0][s->block_index[i]][1] = s->mv[0][i-1][1]; + s->mv[1][i][0] = s->current_picture.motion_val[1][s->block_index[i]][0] = s->mv[1][i-1][0]; + s->mv[1][i][1] = s->current_picture.motion_val[1][s->block_index[i]][1] = s->mv[1][i-1][1]; + } + } else { + for (i = 1; i < 4; i++) { + s->mv[0][i][0] = s->current_picture.motion_val[0][s->block_index[i]][0] = s->mv[0][0][0]; + s->mv[0][i][1] = s->current_picture.motion_val[0][s->block_index[i]][1] = s->mv[0][0][1]; + s->mv[1][i][0] = s->current_picture.motion_val[1][s->block_index[i]][0] = s->mv[1][0][0]; + s->mv[1][i][1] = s->current_picture.motion_val[1][s->block_index[i]][1] = s->mv[1][0][1]; + } + } + } + if (!direct) { if (skipped || !s->mb_intra) { bmvtype = decode012(gb); @@ -2508,7 +2432,7 @@ static int vc1_decode_b_mb_intfr(VC1Context *v) pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i & 4) ? s->uvlinesize : (s->linesize << fieldtx), - (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); + CONFIG_GRAY && (i & 4) && (s->avctx->flags & AV_CODEC_FLAG_GRAY), &block_tt); block_cbp |= pat << (i << 2); if (!v->ttmbf && ttmb < 8) ttmb = -1; @@ -2564,12 +2488,12 @@ static int vc1_decode_b_mb_intfr(VC1Context *v) if (direct || bmvtype == BMV_TYPE_INTERPOLATED) { ff_vc1_interp_mc(v); } + v->fieldtx_plane[mb_pos] = 0; } } - if (s->mb_x == s->mb_width - 1) - memmove(v->is_intra_base, v->is_intra, sizeof(v->is_intra_base[0]) * s->mb_stride); v->cbp[s->mb_x] = block_cbp; v->ttblk[s->mb_x] = block_tt; + return 0; } @@ -2616,30 +2540,27 @@ static void vc1_decode_i_blocks(VC1Context *v) s->mb_x = s->mb_y = 0; s->mb_intra = 1; s->first_slice_line = 1; - for (s->mb_y = 0; s->mb_y < s->end_mb_y; s->mb_y++) { + for (s->mb_y = s->start_mb_y; s->mb_y < s->end_mb_y; s->mb_y++) { s->mb_x = 0; init_block_index(v); for (; s->mb_x < v->end_mb_x; s->mb_x++) { - uint8_t *dst[6]; ff_update_block_index(s); - dst[0] = s->dest[0]; - dst[1] = dst[0] + 8; - dst[2] = s->dest[0] + s->linesize * 8; - dst[3] = dst[2] + 8; - dst[4] = s->dest[1]; - dst[5] = s->dest[2]; - s->bdsp.clear_blocks(s->block[0]); + s->bdsp.clear_blocks(v->block[v->cur_blk_idx][0]); mb_pos = s->mb_x + s->mb_y * s->mb_width; s->current_picture.mb_type[mb_pos] = MB_TYPE_INTRA; s->current_picture.qscale_table[mb_pos] = v->pq; - s->current_picture.motion_val[1][s->block_index[0]][0] = 0; - s->current_picture.motion_val[1][s->block_index[0]][1] = 0; + for (int i = 0; i < 4; i++) { + s->current_picture.motion_val[1][s->block_index[i]][0] = 0; + s->current_picture.motion_val[1][s->block_index[i]][1] = 0; + } // do actual MB decoding and displaying cbp = get_vlc2(&v->s.gb, ff_msmp4_mb_i_vlc.table, MB_INTRA_VLC_BITS, 2); v->s.ac_pred = get_bits1(&v->s.gb); for (k = 0; k < 6; k++) { + v->mb_type[0][s->block_index[k]] = 1; + val = ((cbp >> (5 - k)) & 1); if (k < 4) { @@ -2649,52 +2570,30 @@ static void vc1_decode_i_blocks(VC1Context *v) } cbp |= val << (5 - k); - vc1_decode_i_block(v, s->block[k], k, val, (k < 4) ? v->codingset : v->codingset2); + vc1_decode_i_block(v, v->block[v->cur_blk_idx][block_map[k]], k, val, (k < 4) ? v->codingset : v->codingset2); - if (k > 3 && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && k > 3 && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; - v->vc1dsp.vc1_inv_trans_8x8(s->block[k]); - if (v->pq >= 9 && v->overlap) { - if (v->rangeredfrm) + v->vc1dsp.vc1_inv_trans_8x8(v->block[v->cur_blk_idx][block_map[k]]); + } + + if (v->overlap && v->pq >= 9) { + ff_vc1_i_overlap_filter(v); + if (v->rangeredfrm) + for (k = 0; k < 6; k++) for (j = 0; j < 64; j++) - s->block[k][j] <<= 1; - s->idsp.put_signed_pixels_clamped(s->block[k], dst[k], - k & 4 ? s->uvlinesize - : s->linesize); - } else { - if (v->rangeredfrm) + v->block[v->cur_blk_idx][block_map[k]][j] <<= 1; + vc1_put_blocks_clamped(v, 1); + } else { + if (v->rangeredfrm) + for (k = 0; k < 6; k++) for (j = 0; j < 64; j++) - s->block[k][j] = (s->block[k][j] - 64) << 1; - s->idsp.put_pixels_clamped(s->block[k], dst[k], - k & 4 ? s->uvlinesize - : s->linesize); - } + v->block[v->cur_blk_idx][block_map[k]][j] = (v->block[v->cur_blk_idx][block_map[k]][j] - 64) << 1; + vc1_put_blocks_clamped(v, 0); } - if (v->pq >= 9 && v->overlap) { - if (s->mb_x) { - v->vc1dsp.vc1_h_overlap(s->dest[0], s->linesize); - v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize, s->linesize); - if (!(s->avctx->flags & AV_CODEC_FLAG_GRAY)) { - v->vc1dsp.vc1_h_overlap(s->dest[1], s->uvlinesize); - v->vc1dsp.vc1_h_overlap(s->dest[2], s->uvlinesize); - } - } - v->vc1dsp.vc1_h_overlap(s->dest[0] + 8, s->linesize); - v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize); - if (!s->first_slice_line) { - v->vc1dsp.vc1_v_overlap(s->dest[0], s->linesize); - v->vc1dsp.vc1_v_overlap(s->dest[0] + 8, s->linesize); - if (!(s->avctx->flags & AV_CODEC_FLAG_GRAY)) { - v->vc1dsp.vc1_v_overlap(s->dest[1], s->uvlinesize); - v->vc1dsp.vc1_v_overlap(s->dest[2], s->uvlinesize); - } - } - v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize); - v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize); - } if (v->s.loop_filter) - ff_vc1_loop_filter_iblk(v, v->pq); + ff_vc1_i_loop_filter(v); if (get_bits_count(&s->gb) > v->bits) { ff_er_add_slice(&s->er, 0, 0, s->mb_x, s->mb_y, ER_MB_ERROR); @@ -2702,6 +2601,11 @@ static void vc1_decode_i_blocks(VC1Context *v) get_bits_count(&s->gb), v->bits); return; } + + v->topleft_blk_idx = (v->topleft_blk_idx + 1) % (v->end_mb_x + 2); + v->top_blk_idx = (v->top_blk_idx + 1) % (v->end_mb_x + 2); + v->left_blk_idx = (v->left_blk_idx + 1) % (v->end_mb_x + 2); + v->cur_blk_idx = (v->cur_blk_idx + 1) % (v->end_mb_x + 2); } if (!v->s.loop_filter) ff_mpeg_draw_horiz_band(s, s->mb_y * 16, 16); @@ -2727,7 +2631,7 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) int cbp, val; uint8_t *coded_val; int mb_pos; - int mquant = v->pq; + int mquant; int mqdiff; GetBitContext *gb = &s->gb; @@ -2771,13 +2675,15 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) s->mb_x = 0; init_block_index(v); for (;s->mb_x < s->mb_width; s->mb_x++) { - int16_t (*block)[64] = v->block[v->cur_blk_idx]; + mquant = v->pq; ff_update_block_index(s); - s->bdsp.clear_blocks(block[0]); + s->bdsp.clear_blocks(v->block[v->cur_blk_idx][0]); mb_pos = s->mb_x + s->mb_y * s->mb_stride; s->current_picture.mb_type[mb_pos + v->mb_off] = MB_TYPE_INTRA; - s->current_picture.motion_val[1][s->block_index[0] + v->blocks_off][0] = 0; - s->current_picture.motion_val[1][s->block_index[0] + v->blocks_off][1] = 0; + for (int i = 0; i < 4; i++) { + s->current_picture.motion_val[1][s->block_index[i] + v->blocks_off][0] = 0; + s->current_picture.motion_val[1][s->block_index[i] + v->blocks_off][1] = 0; + } // do actual MB decoding and displaying if (v->fieldtx_is_raw) @@ -2795,10 +2701,12 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) s->current_picture.qscale_table[mb_pos] = mquant; /* Set DC scale - y and c use the same */ - s->y_dc_scale = s->y_dc_scale_table[mquant]; - s->c_dc_scale = s->c_dc_scale_table[mquant]; + s->y_dc_scale = s->y_dc_scale_table[FFABS(mquant)]; + s->c_dc_scale = s->c_dc_scale_table[FFABS(mquant)]; for (k = 0; k < 6; k++) { + v->mb_type[0][s->block_index[k]] = 1; + val = ((cbp >> (5 - k)) & 1); if (k < 4) { @@ -2811,18 +2719,19 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) v->a_avail = !s->first_slice_line || (k == 2 || k == 3); v->c_avail = !!s->mb_x || (k == 1 || k == 3); - vc1_decode_i_block_adv(v, block[k], k, val, + vc1_decode_i_block_adv(v, v->block[v->cur_blk_idx][block_map[k]], k, val, (k < 4) ? v->codingset : v->codingset2, mquant); - if (k > 3 && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) + if (CONFIG_GRAY && k > 3 && (s->avctx->flags & AV_CODEC_FLAG_GRAY)) continue; - v->vc1dsp.vc1_inv_trans_8x8(block[k]); + v->vc1dsp.vc1_inv_trans_8x8(v->block[v->cur_blk_idx][block_map[k]]); } - ff_vc1_smooth_overlap_filter_iblk(v); - vc1_put_signed_blocks_clamped(v); + if (v->overlap && (v->pq >= 9 || v->condover != CONDOVER_NONE)) + ff_vc1_i_overlap_filter(v); + vc1_put_blocks_clamped(v, 1); if (v->s.loop_filter) - ff_vc1_loop_filter_iblk_delayed(v, v->pq); + ff_vc1_i_loop_filter(v); if (get_bits_count(&s->gb) > v->bits) { // TODO: may need modification to handle slice coding @@ -2831,6 +2740,10 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) get_bits_count(&s->gb), v->bits); return; } + inc_blk_idx(v->topleft_blk_idx); + inc_blk_idx(v->top_blk_idx); + inc_blk_idx(v->left_blk_idx); + inc_blk_idx(v->cur_blk_idx); } if (!v->s.loop_filter) ff_mpeg_draw_horiz_band(s, s->mb_y * 16, 16); @@ -2839,18 +2752,8 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) s->first_slice_line = 0; } - /* raw bottom MB row */ - s->mb_x = 0; - init_block_index(v); - - for (;s->mb_x < s->mb_width; s->mb_x++) { - ff_update_block_index(s); - vc1_put_signed_blocks_clamped(v); - if (v->s.loop_filter) - ff_vc1_loop_filter_iblk_delayed(v, v->pq); - } if (v->s.loop_filter) - ff_mpeg_draw_horiz_band(s, (s->end_mb_y-1)*16, 16); + ff_mpeg_draw_horiz_band(s, (s->end_mb_y - 1) * 16, 16); ff_er_add_slice(&s->er, 0, s->start_mb_y << v->field_mode, s->mb_width - 1, (s->end_mb_y << v->field_mode) - 1, ER_MB_END); } @@ -2885,23 +2788,28 @@ static void vc1_decode_p_blocks(VC1Context *v) break; } - apply_loop_filter = s->loop_filter && !(s->avctx->skip_loop_filter >= AVDISCARD_NONKEY) && - v->fcm == PROGRESSIVE; + apply_loop_filter = s->loop_filter && !(s->avctx->skip_loop_filter >= AVDISCARD_NONKEY); s->first_slice_line = 1; - memset(v->cbp_base, 0, sizeof(v->cbp_base[0])*2*s->mb_stride); + memset(v->cbp_base, 0, sizeof(v->cbp_base[0]) * 3 * s->mb_stride); for (s->mb_y = s->start_mb_y; s->mb_y < s->end_mb_y; s->mb_y++) { s->mb_x = 0; init_block_index(v); for (; s->mb_x < s->mb_width; s->mb_x++) { ff_update_block_index(s); - if (v->fcm == ILACE_FIELD) + if (v->fcm == ILACE_FIELD) { vc1_decode_p_mb_intfi(v); - else if (v->fcm == ILACE_FRAME) + if (apply_loop_filter) + ff_vc1_p_loop_filter(v); + } else if (v->fcm == ILACE_FRAME) { vc1_decode_p_mb_intfr(v); - else vc1_decode_p_mb(v); - if (s->mb_y != s->start_mb_y && apply_loop_filter) - ff_vc1_apply_p_loop_filter(v); + if (apply_loop_filter) + ff_vc1_p_intfr_loop_filter(v); + } else { + vc1_decode_p_mb(v); + if (apply_loop_filter) + ff_vc1_p_loop_filter(v); + } if (get_bits_count(&s->gb) > v->bits || get_bits_count(&s->gb) < 0) { // TODO: may need modification to handle slice coding ff_er_add_slice(&s->er, 0, s->start_mb_y, s->mb_x, s->mb_y, ER_MB_ERROR); @@ -2909,22 +2817,27 @@ static void vc1_decode_p_blocks(VC1Context *v) get_bits_count(&s->gb), v->bits, s->mb_x, s->mb_y); return; } + inc_blk_idx(v->topleft_blk_idx); + inc_blk_idx(v->top_blk_idx); + inc_blk_idx(v->left_blk_idx); + inc_blk_idx(v->cur_blk_idx); } - memmove(v->cbp_base, v->cbp, sizeof(v->cbp_base[0]) * s->mb_stride); - memmove(v->ttblk_base, v->ttblk, sizeof(v->ttblk_base[0]) * s->mb_stride); - memmove(v->is_intra_base, v->is_intra, sizeof(v->is_intra_base[0]) * s->mb_stride); - memmove(v->luma_mv_base, v->luma_mv, sizeof(v->luma_mv_base[0]) * s->mb_stride); - if (s->mb_y != s->start_mb_y) ff_mpeg_draw_horiz_band(s, (s->mb_y - 1) * 16, 16); + memmove(v->cbp_base, + v->cbp - s->mb_stride, + sizeof(v->cbp_base[0]) * 2 * s->mb_stride); + memmove(v->ttblk_base, + v->ttblk - s->mb_stride, + sizeof(v->ttblk_base[0]) * 2 * s->mb_stride); + memmove(v->is_intra_base, + v->is_intra - s->mb_stride, + sizeof(v->is_intra_base[0]) * 2 * s->mb_stride); + memmove(v->luma_mv_base, + v->luma_mv - s->mb_stride, + sizeof(v->luma_mv_base[0]) * 2 * s->mb_stride); + if (s->mb_y != s->start_mb_y) + ff_mpeg_draw_horiz_band(s, (s->mb_y - 1) * 16, 16); s->first_slice_line = 0; } - if (apply_loop_filter) { - s->mb_x = 0; - init_block_index(v); - for (; s->mb_x < s->mb_width; s->mb_x++) { - ff_update_block_index(s); - ff_vc1_apply_p_loop_filter(v); - } - } if (s->end_mb_y >= s->start_mb_y) ff_mpeg_draw_horiz_band(s, (s->end_mb_y - 1) * 16, 16); ff_er_add_slice(&s->er, 0, s->start_mb_y << v->field_mode, s->mb_width - 1, @@ -2967,12 +2880,19 @@ static void vc1_decode_b_blocks(VC1Context *v) for (; s->mb_x < s->mb_width; s->mb_x++) { ff_update_block_index(s); - if (v->fcm == ILACE_FIELD) + if (v->fcm == ILACE_FIELD) { vc1_decode_b_mb_intfi(v); - else if (v->fcm == ILACE_FRAME) + if (v->s.loop_filter) + ff_vc1_b_intfi_loop_filter(v); + } else if (v->fcm == ILACE_FRAME) { vc1_decode_b_mb_intfr(v); - else + if (v->s.loop_filter) + ff_vc1_p_intfr_loop_filter(v); + } else { vc1_decode_b_mb(v); + if (v->s.loop_filter) + ff_vc1_i_loop_filter(v); + } if (get_bits_count(&s->gb) > v->bits || get_bits_count(&s->gb) < 0) { // TODO: may need modification to handle slice coding ff_er_add_slice(&s->er, 0, s->start_mb_y, s->mb_x, s->mb_y, ER_MB_ERROR); @@ -2980,9 +2900,16 @@ static void vc1_decode_b_blocks(VC1Context *v) get_bits_count(&s->gb), v->bits, s->mb_x, s->mb_y); return; } - if (v->s.loop_filter) - ff_vc1_loop_filter_iblk(v, v->pq); } + memmove(v->cbp_base, + v->cbp - s->mb_stride, + sizeof(v->cbp_base[0]) * 2 * s->mb_stride); + memmove(v->ttblk_base, + v->ttblk - s->mb_stride, + sizeof(v->ttblk_base[0]) * 2 * s->mb_stride); + memmove(v->is_intra_base, + v->is_intra - s->mb_stride, + sizeof(v->is_intra_base[0]) * 2 * s->mb_stride); if (!v->s.loop_filter) ff_mpeg_draw_horiz_band(s, s->mb_y * 16, 16); else if (s->mb_y) |