1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
FFmpeg/libavcodec/dv.c
Christophe Gisquet 4fa772acbb dv: increase VLC reading bits to 10
This also requires a bump in the table size of bit patterns to 1664.
From 356 to 348 cycles.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-10-28 12:22:03 +01:00

273 lines
8.7 KiB
C

/*
* DV decoder
* Copyright (c) 2002 Fabrice Bellard
* Copyright (c) 2004 Roman Shaposhnik
*
* DV encoder
* Copyright (c) 2003 Roman Shaposhnik
*
* 50 Mbps (DVCPRO50) support
* Copyright (c) 2006 Daniel Maas <dmaas@maasdigital.com>
*
* 100 Mbps (DVCPRO HD) support
* Initial code by Daniel Maas <dmaas@maasdigital.com> (funded by BBC R&D)
* Final code by Roman Shaposhnik
*
* Many thanks to Dan Dennedy <dan@dennedy.org> for providing wealth
* of DV technical info.
*
* This file is part of FFmpeg.
*
* 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.
*
* 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file
* DV codec.
*/
#include "libavutil/internal.h"
#include "libavutil/pixdesc.h"
#include "avcodec.h"
#include "dv.h"
#include "dvdata.h"
#include "get_bits.h"
#include "internal.h"
#include "put_bits.h"
#include "simple_idct.h"
/* XXX: also include quantization */
RL_VLC_ELEM ff_dv_rl_vlc[1664];
static inline void dv_calc_mb_coordinates(const AVDVProfile *d, int chan,
int seq, int slot, uint16_t *tbl)
{
static const uint8_t off[] = { 2, 6, 8, 0, 4 };
static const uint8_t shuf1[] = { 36, 18, 54, 0, 72 };
static const uint8_t shuf2[] = { 24, 12, 36, 0, 48 };
static const uint8_t shuf3[] = { 18, 9, 27, 0, 36 };
static const uint8_t l_start[] = { 0, 4, 9, 13, 18, 22, 27, 31, 36, 40 };
static const uint8_t l_start_shuffled[] = { 9, 4, 13, 0, 18 };
static const uint8_t serpent1[] = {
0, 1, 2, 2, 1, 0,
0, 1, 2, 2, 1, 0,
0, 1, 2, 2, 1, 0,
0, 1, 2, 2, 1, 0,
0, 1, 2
};
static const uint8_t serpent2[] = {
0, 1, 2, 3, 4, 5, 5, 4, 3, 2, 1, 0,
0, 1, 2, 3, 4, 5, 5, 4, 3, 2, 1, 0,
0, 1, 2, 3, 4, 5
};
static const uint8_t remap[][2] = {
{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, /* dummy */
{ 0, 0 }, { 0, 1 }, { 0, 2 }, { 0, 3 }, { 10, 0 },
{ 10, 1 }, { 10, 2 }, { 10, 3 }, { 20, 0 }, { 20, 1 },
{ 20, 2 }, { 20, 3 }, { 30, 0 }, { 30, 1 }, { 30, 2 },
{ 30, 3 }, { 40, 0 }, { 40, 1 }, { 40, 2 }, { 40, 3 },
{ 50, 0 }, { 50, 1 }, { 50, 2 }, { 50, 3 }, { 60, 0 },
{ 60, 1 }, { 60, 2 }, { 60, 3 }, { 70, 0 }, { 70, 1 },
{ 70, 2 }, { 70, 3 }, { 0, 64 }, { 0, 65 }, { 0, 66 },
{ 10, 64 }, { 10, 65 }, { 10, 66 }, { 20, 64 }, { 20, 65 },
{ 20, 66 }, { 30, 64 }, { 30, 65 }, { 30, 66 }, { 40, 64 },
{ 40, 65 }, { 40, 66 }, { 50, 64 }, { 50, 65 }, { 50, 66 },
{ 60, 64 }, { 60, 65 }, { 60, 66 }, { 70, 64 }, { 70, 65 },
{ 70, 66 }, { 0, 67 }, { 20, 67 }, { 40, 67 }, { 60, 67 }
};
int i, k, m;
int x, y, blk;
for (m = 0; m < 5; m++) {
switch (d->width) {
case 1440:
blk = (chan * 11 + seq) * 27 + slot;
if (chan == 0 && seq == 11) {
x = m * 27 + slot;
if (x < 90) {
y = 0;
} else {
x = (x - 90) * 2;
y = 67;
}
} else {
i = (4 * chan + blk + off[m]) % 11;
k = (blk / 11) % 27;
x = shuf1[m] + (chan & 1) * 9 + k % 9;
y = (i * 3 + k / 9) * 2 + (chan >> 1) + 1;
}
tbl[m] = (x << 1) | (y << 9);
break;
case 1280:
blk = (chan * 10 + seq) * 27 + slot;
i = (4 * chan + (seq / 5) + 2 * blk + off[m]) % 10;
k = (blk / 5) % 27;
x = shuf1[m] + (chan & 1) * 9 + k % 9;
y = (i * 3 + k / 9) * 2 + (chan >> 1) + 4;
if (x >= 80) {
x = remap[y][0] + ((x - 80) << (y > 59));
y = remap[y][1];
}
tbl[m] = (x << 1) | (y << 9);
break;
case 960:
blk = (chan * 10 + seq) * 27 + slot;
i = (4 * chan + (seq / 5) + 2 * blk + off[m]) % 10;
k = (blk / 5) % 27 + (i & 1) * 3;
x = shuf2[m] + k % 6 + 6 * (chan & 1);
y = l_start[i] + k / 6 + 45 * (chan >> 1);
tbl[m] = (x << 1) | (y << 9);
break;
case 720:
switch (d->pix_fmt) {
case AV_PIX_FMT_YUV422P:
x = shuf3[m] + slot / 3;
y = serpent1[slot] +
((((seq + off[m]) % d->difseg_size) << 1) + chan) * 3;
tbl[m] = (x << 1) | (y << 8);
break;
case AV_PIX_FMT_YUV420P:
x = shuf3[m] + slot / 3;
y = serpent1[slot] +
((seq + off[m]) % d->difseg_size) * 3;
tbl[m] = (x << 1) | (y << 9);
break;
case AV_PIX_FMT_YUV411P:
i = (seq + off[m]) % d->difseg_size;
k = slot + ((m == 1 || m == 2) ? 3 : 0);
x = l_start_shuffled[m] + k / 6;
y = serpent2[k] + i * 6;
if (x > 21)
y = y * 2 - i * 6;
tbl[m] = (x << 2) | (y << 8);
break;
}
default:
break;
}
}
}
/* quantization quanta by QNO for DV100 */
static const uint8_t dv100_qstep[16] = {
1, /* QNO = 0 and 1 both have no quantization */
1,
2, 3, 4, 5, 6, 7, 8, 16, 18, 20, 22, 24, 28, 52
};
static const uint8_t dv_quant_areas[4] = { 6, 21, 43, 64 };
int ff_dv_init_dynamic_tables(DVVideoContext *ctx, const AVDVProfile *d)
{
int j, i, c, s, p;
p = i = 0;
for (c = 0; c < d->n_difchan; c++) {
for (s = 0; s < d->difseg_size; s++) {
p += 6;
for (j = 0; j < 27; j++) {
p += !(j % 3);
if (!(DV_PROFILE_IS_1080i50(d) && c != 0 && s == 11) &&
!(DV_PROFILE_IS_720p50(d) && s > 9)) {
dv_calc_mb_coordinates(d, c, s, j, &ctx->work_chunks[i].mb_coordinates[0]);
ctx->work_chunks[i++].buf_offset = p;
}
p += 5;
}
}
}
return 0;
}
av_cold int ff_dvvideo_init(AVCodecContext *avctx)
{
DVVideoContext *s = avctx->priv_data;
static int done = 0;
int i, j;
if (!done) {
VLC dv_vlc;
uint16_t new_dv_vlc_bits[NB_DV_VLC * 2];
uint8_t new_dv_vlc_len[NB_DV_VLC * 2];
uint8_t new_dv_vlc_run[NB_DV_VLC * 2];
int16_t new_dv_vlc_level[NB_DV_VLC * 2];
done = 1;
/* it's faster to include sign bit in a generic VLC parsing scheme */
for (i = 0, j = 0; i < NB_DV_VLC; i++, j++) {
new_dv_vlc_bits[j] = ff_dv_vlc_bits[i];
new_dv_vlc_len[j] = ff_dv_vlc_len[i];
new_dv_vlc_run[j] = ff_dv_vlc_run[i];
new_dv_vlc_level[j] = ff_dv_vlc_level[i];
if (ff_dv_vlc_level[i]) {
new_dv_vlc_bits[j] <<= 1;
new_dv_vlc_len[j]++;
j++;
new_dv_vlc_bits[j] = (ff_dv_vlc_bits[i] << 1) | 1;
new_dv_vlc_len[j] = ff_dv_vlc_len[i] + 1;
new_dv_vlc_run[j] = ff_dv_vlc_run[i];
new_dv_vlc_level[j] = -ff_dv_vlc_level[i];
}
}
/* NOTE: as a trick, we use the fact the no codes are unused
* to accelerate the parsing of partial codes */
init_vlc(&dv_vlc, TEX_VLC_BITS, j, new_dv_vlc_len,
1, 1, new_dv_vlc_bits, 2, 2, 0);
av_assert1(dv_vlc.table_size == 1664);
for (i = 0; i < dv_vlc.table_size; i++) {
int code = dv_vlc.table[i][0];
int len = dv_vlc.table[i][1];
int level, run;
if (len < 0) { // more bits needed
run = 0;
level = code;
} else {
run = new_dv_vlc_run[code] + 1;
level = new_dv_vlc_level[code];
}
ff_dv_rl_vlc[i].len = len;
ff_dv_rl_vlc[i].level = level;
ff_dv_rl_vlc[i].run = run;
}
ff_free_vlc(&dv_vlc);
}
s->avctx = avctx;
avctx->chroma_sample_location = AVCHROMA_LOC_TOPLEFT;
return 0;
}