aboutsummaryrefslogtreecommitdiffstats
path: root/libavcodec/vc2enc_dwt.c
blob: 441af040ecf5a9819156234a9db2352c11194ee1 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
/*
 * Copyright (C) 2007 Marco Gerards <marco@gnu.org>
 * Copyright (C) 2016 Open Broadcast Systems Ltd.
 * Author        2016 Rostislav Pehlivanov <atomnuker@gmail.com>
 *
 * 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
 */

#include "libavutil/attributes.h"
#include "libavutil/mem.h"
#include "vc2enc_dwt.h"

/* Since the transforms spit out interleaved coefficients, this function
 * rearranges the coefficients into the more traditional subdivision,
 * making it easier to encode and perform another level. */
static av_always_inline void deinterleave(dwtcoef *linell, ptrdiff_t stride,
                                          int width, int height, dwtcoef *synthl)
{
    int x, y;
    ptrdiff_t synthw = width << 1;
    dwtcoef *linehl = linell + width;
    dwtcoef *linelh = linell + height*stride;
    dwtcoef *linehh = linelh + width;

    /* Deinterleave the coefficients. */
    for (y = 0; y < height; y++) {
        for (x = 0; x < width; x++) {
            linell[x] = synthl[(x << 1)];
            linehl[x] = synthl[(x << 1) + 1];
            linelh[x] = synthl[(x << 1) + synthw];
            linehh[x] = synthl[(x << 1) + synthw + 1];
        }
        synthl += synthw << 1;
        linell += stride;
        linelh += stride;
        linehl += stride;
        linehh += stride;
    }
}

static void vc2_subband_dwt_97(VC2TransformContext *t, dwtcoef *data,
                               ptrdiff_t stride, int width, int height)
{
    int x, y;
    dwtcoef *datal = data, *synth = t->buffer, *synthl = synth;
    const ptrdiff_t synth_width  = width  << 1;
    const ptrdiff_t synth_height = height << 1;

    /*
     * Shift in one bit that is used for additional precision and copy
     * the data to the buffer.
     */
    for (y = 0; y < synth_height; y++) {
        for (x = 0; x < synth_width; x++)
            synthl[x] = datal[x] * 2;
        synthl += synth_width;
        datal += stride;
    }

    /* Horizontal synthesis. */
    synthl = synth;
    for (y = 0; y < synth_height; y++) {
        /* Lifting stage 2. */
        synthl[1] -= (8*synthl[0] + 9*synthl[2] - synthl[4] + 8) >> 4;
        for (x = 1; x < width - 2; x++)
            synthl[2*x + 1] -= (9*synthl[2*x] + 9*synthl[2*x + 2] - synthl[2*x + 4] -
                                synthl[2 * x - 2] + 8) >> 4;
        synthl[synth_width - 1] -= (17*synthl[synth_width - 2] -
                                    synthl[synth_width - 4] + 8) >> 4;
        synthl[synth_width - 3] -= (8*synthl[synth_width - 2] +
                                    9*synthl[synth_width - 4] -
                                    synthl[synth_width - 6] + 8) >> 4;
        /* Lifting stage 1. */
        synthl[0] += (synthl[1] + synthl[1] + 2) >> 2;
        for (x = 1; x < width - 1; x++)
            synthl[2*x] += (synthl[2*x - 1] + synthl[2*x + 1] + 2) >> 2;

        synthl[synth_width - 2] += (synthl[synth_width - 3] +
                                    synthl[synth_width - 1] + 2) >> 2;
        synthl += synth_width;
    }

    /* Vertical synthesis: Lifting stage 2. */
    synthl = synth + synth_width;
    for (x = 0; x < synth_width; x++)
        synthl[x] -= (8*synthl[x - synth_width] + 9*synthl[x + synth_width] -
                      synthl[x + 3 * synth_width] + 8) >> 4;

    synthl = synth + (synth_width << 1);
    for (y = 1; y < height - 2; y++) {
        for (x = 0; x < synth_width; x++)
            synthl[x + synth_width] -= (9*synthl[x] +
                                        9*synthl[x + 2 * synth_width] -
                                        synthl[x - 2 * synth_width] -
                                        synthl[x + 4 * synth_width] + 8) >> 4;
        synthl += synth_width << 1;
    }

    synthl = synth + (synth_height - 1) * synth_width;
    for (x = 0; x < synth_width; x++) {
        synthl[x] -= (17*synthl[x - synth_width] -
                      synthl[x - 3*synth_width] + 8) >> 4;
                      synthl[x - 2*synth_width] -= (9*synthl[x - 3*synth_width] +
                      8*synthl[x - 1*synth_width] - synthl[x - 5*synth_width] + 8) >> 4;
    }

    /* Vertical synthesis: Lifting stage 1. */
    synthl = synth;
    for (x = 0; x < synth_width; x++)
        synthl[x] += (synthl[x + synth_width] + synthl[x + synth_width] + 2) >> 2;

    synthl = synth + (synth_width << 1);
    for (y = 1; y < height - 1; y++) {
        for (x = 0; x < synth_width; x++)
            synthl[x] += (synthl[x - synth_width] + synthl[x + synth_width] + 2) >> 2;
        synthl += synth_width << 1;
    }

    synthl = synth + (synth_height - 2) * synth_width;
    for (x = 0; x < synth_width; x++)
        synthl[x] += (synthl[x - synth_width] + synthl[x + synth_width] + 2) >> 2;

    deinterleave(data, stride, width, height, synth);
}

static void vc2_subband_dwt_53(VC2TransformContext *t, dwtcoef *data,
                               ptrdiff_t stride, int width, int height)
{
    int x, y;
    dwtcoef *synth = t->buffer, *synthl = synth, *datal = data;
    const ptrdiff_t synth_width  = width  << 1;
    const ptrdiff_t synth_height = height << 1;

    /*
     * Shift in one bit that is used for additional precision and copy
     * the data to the buffer.
     */
    for (y = 0; y < synth_height; y++) {
        for (x = 0; x < synth_width; x++)
            synthl[x] = datal[x] << 1;
        synthl += synth_width;
        datal  += stride;
    }

    /* Horizontal synthesis. */
    synthl = synth;
    for (y = 0; y < synth_height; y++) {
        /* Lifting stage 2. */
        for (x = 0; x < width - 1; x++)
            synthl[2 * x + 1] -= (synthl[2 * x] + synthl[2 * x + 2] + 1) >> 1;

        synthl[synth_width - 1] -= (2*synthl[synth_width - 2] + 1) >> 1;

        /* Lifting stage 1. */
        synthl[0] += (2*synthl[1] + 2) >> 2;
        for (x = 1; x < width - 1; x++)
            synthl[2 * x] += (synthl[2 * x - 1] + synthl[2 * x + 1] + 2) >> 2;

        synthl[synth_width - 2] += (synthl[synth_width - 3] + synthl[synth_width - 1] + 2) >> 2;

        synthl += synth_width;
    }

    /* Vertical synthesis: Lifting stage 2. */
    synthl = synth + synth_width;
    for (x = 0; x < synth_width; x++)
        synthl[x] -= (synthl[x - synth_width] + synthl[x + synth_width] + 1) >> 1;

    synthl = synth + (synth_width << 1);
    for (y = 1; y < height - 1; y++) {
        for (x = 0; x < synth_width; x++)
            synthl[x + synth_width] -= (synthl[x] + synthl[x + synth_width * 2] + 1) >> 1;
        synthl += (synth_width << 1);
    }

    synthl = synth + (synth_height - 1) * synth_width;
    for (x = 0; x < synth_width; x++)
        synthl[x] -= (2*synthl[x - synth_width] + 1) >> 1;

    /* Vertical synthesis: Lifting stage 1. */
    synthl = synth;
    for (x = 0; x < synth_width; x++)
        synthl[x] += (2*synthl[synth_width + x] + 2) >> 2;

    synthl = synth + (synth_width << 1);
    for (y = 1; y < height - 1; y++) {
        for (x = 0; x < synth_width; x++)
            synthl[x] += (synthl[x + synth_width] + synthl[x - synth_width] + 2) >> 2;
        synthl += (synth_width << 1);
    }

    synthl = synth + (synth_height - 2)*synth_width;
    for (x = 0; x < synth_width; x++)
        synthl[x] += (synthl[x - synth_width] + synthl[x + synth_width] + 2) >> 2;


    deinterleave(data, stride, width, height, synth);
}

static av_always_inline void dwt_haar(VC2TransformContext *t, dwtcoef *data,
                                      ptrdiff_t stride, int width, int height,
                                      const int s)
{
    int x, y;
    dwtcoef *synth = t->buffer, *synthl = synth, *datal = data;
    const ptrdiff_t synth_width  = width  << 1;
    const ptrdiff_t synth_height = height << 1;

    /* Horizontal synthesis. */
    for (y = 0; y < synth_height; y++) {
        for (x = 0; x < synth_width; x += 2) {
            synthl[y*synth_width + x + 1] = (datal[y*stride + x + 1] << s) -
                                            (datal[y*stride + x] << s);
            synthl[y*synth_width + x] = (datal[y*stride + x + 0] << s) +
                                        ((synthl[y*synth_width + x + 1] + 1) >> 1);
        }
    }

    /* Vertical synthesis. */
    for (x = 0; x < synth_width; x++) {
        for (y = 0; y < synth_height; y += 2) {
            synthl[(y + 1)*synth_width + x] = synthl[(y + 1)*synth_width + x] -
                                              synthl[y*synth_width + x];
            synthl[y*synth_width + x] = synthl[y*synth_width + x] +
                                        ((synthl[(y + 1)*synth_width + x] + 1) >> 1);
        }
    }

    deinterleave(data, stride, width, height, synth);
}

static void vc2_subband_dwt_haar(VC2TransformContext *t, dwtcoef *data,
                                 ptrdiff_t stride, int width, int height)
{
    dwt_haar(t, data, stride, width, height, 0);
}

static void vc2_subband_dwt_haar_shift(VC2TransformContext *t, dwtcoef *data,
                                       ptrdiff_t stride, int width, int height)
{
    dwt_haar(t, data, stride, width, height, 1);
}

av_cold int ff_vc2enc_init_transforms(VC2TransformContext *s, int p_stride,
                                      int p_height, int slice_w, int slice_h)
{
    s->vc2_subband_dwt[VC2_TRANSFORM_9_7]    = vc2_subband_dwt_97;
    s->vc2_subband_dwt[VC2_TRANSFORM_5_3]    = vc2_subband_dwt_53;
    s->vc2_subband_dwt[VC2_TRANSFORM_HAAR]   = vc2_subband_dwt_haar;
    s->vc2_subband_dwt[VC2_TRANSFORM_HAAR_S] = vc2_subband_dwt_haar_shift;

    /* Pad by the slice size, only matters for non-Haar wavelets */
    s->buffer = av_calloc((p_stride + slice_w)*(p_height + slice_h), sizeof(dwtcoef));
    if (!s->buffer)
        return 1;

    s->padding = (slice_h >> 1)*p_stride + (slice_w >> 1);
    s->buffer += s->padding;

    return 0;
}

av_cold void ff_vc2enc_free_transforms(VC2TransformContext *s)
{
    if (s->buffer) {
        av_free(s->buffer - s->padding);
        s->buffer = NULL;
    }
}