xref: /third_party/ffmpeg/libavutil/tx.h (revision cabdff1a) 
1/*
2 * This file is part of FFmpeg.
3 *
4 * FFmpeg is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
8 *
9 * FFmpeg is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
12 * Lesser General Public License for more details.
13 *
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with FFmpeg; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18
19#ifndef AVUTIL_TX_H
20#define AVUTIL_TX_H
21
22#include <stdint.h>
23#include <stddef.h>
24
25typedef struct AVTXContext AVTXContext;
26
27typedef struct AVComplexFloat {
28    float re, im;
29} AVComplexFloat;
30
31typedef struct AVComplexDouble {
32    double re, im;
33} AVComplexDouble;
34
35typedef struct AVComplexInt32 {
36    int32_t re, im;
37} AVComplexInt32;
38
39enum AVTXType {
40    /**
41     * Standard complex to complex FFT with sample data type of AVComplexFloat,
42     * AVComplexDouble or AVComplexInt32, for each respective variant.
43     *
44     * Output is not 1/len normalized. Scaling currently unsupported.
45     * The stride parameter must be set to the size of a single sample in bytes.
46     */
47    AV_TX_FLOAT_FFT  = 0,
48    AV_TX_DOUBLE_FFT = 2,
49    AV_TX_INT32_FFT  = 4,
50
51    /**
52     * Standard MDCT with a sample data type of float, double or int32_t,
53     * respecively. For the float and int32 variants, the scale type is
54     * 'float', while for the double variant, it's 'double'.
55     * If scale is NULL, 1.0 will be used as a default.
56     *
57     * Length is the frame size, not the window size (which is 2x frame).
58     * For forward transforms, the stride specifies the spacing between each
59     * sample in the output array in bytes. The input must be a flat array.
60     *
61     * For inverse transforms, the stride specifies the spacing between each
62     * sample in the input array in bytes. The output must be a flat array.
63     *
64     * NOTE: the inverse transform is half-length, meaning the output will not
65     * contain redundant data. This is what most codecs work with. To do a full
66     * inverse transform, set the AV_TX_FULL_IMDCT flag on init.
67     */
68    AV_TX_FLOAT_MDCT  = 1,
69    AV_TX_DOUBLE_MDCT = 3,
70    AV_TX_INT32_MDCT  = 5,
71
72    /**
73     * Real to complex and complex to real DFTs.
74     * For the float and int32 variants, the scale type is 'float', while for
75     * the double variant, it's a 'double'. If scale is NULL, 1.0 will be used
76     * as a default.
77     *
78     * The stride parameter must be set to the size of a single sample in bytes.
79     *
80     * The forward transform performs a real-to-complex DFT of N samples to
81     * N/2+1 complex values.
82     *
83     * The inverse transform performs a complex-to-real DFT of N/2+1 complex
84     * values to N real samples. The output is not normalized, but can be
85     * made so by setting the scale value to 1.0/len.
86     * NOTE: the inverse transform always overwrites the input.
87     */
88    AV_TX_FLOAT_RDFT  = 6,
89    AV_TX_DOUBLE_RDFT = 7,
90    AV_TX_INT32_RDFT  = 8,
91
92    /* Not part of the API, do not use */
93    AV_TX_NB,
94};
95
96/**
97 * Function pointer to a function to perform the transform.
98 *
99 * @note Using a different context than the one allocated during av_tx_init()
100 * is not allowed.
101 *
102 * @param s the transform context
103 * @param out the output array
104 * @param in the input array
105 * @param stride the input or output stride in bytes
106 *
107 * The out and in arrays must be aligned to the maximum required by the CPU
108 * architecture unless the AV_TX_UNALIGNED flag was set in av_tx_init().
109 * The stride must follow the constraints the transform type has specified.
110 */
111typedef void (*av_tx_fn)(AVTXContext *s, void *out, void *in, ptrdiff_t stride);
112
113/**
114 * Flags for av_tx_init()
115 */
116enum AVTXFlags {
117    /**
118     * Performs an in-place transformation on the input. The output argument
119     * of av_tn_fn() MUST match the input. May be unsupported or slower for some
120     * transform types.
121     */
122    AV_TX_INPLACE = 1ULL << 0,
123
124    /**
125     * Relaxes alignment requirement for the in and out arrays of av_tx_fn().
126     * May be slower with certain transform types.
127     */
128    AV_TX_UNALIGNED = 1ULL << 1,
129
130    /**
131     * Performs a full inverse MDCT rather than leaving out samples that can be
132     * derived through symmetry. Requires an output array of 'len' floats,
133     * rather than the usual 'len/2' floats.
134     * Ignored for all transforms but inverse MDCTs.
135     */
136    AV_TX_FULL_IMDCT = 1ULL << 2,
137};
138
139/**
140 * Initialize a transform context with the given configuration
141 * (i)MDCTs with an odd length are currently not supported.
142 *
143 * @param ctx the context to allocate, will be NULL on error
144 * @param tx pointer to the transform function pointer to set
145 * @param type type the type of transform
146 * @param inv whether to do an inverse or a forward transform
147 * @param len the size of the transform in samples
148 * @param scale pointer to the value to scale the output if supported by type
149 * @param flags a bitmask of AVTXFlags or 0
150 *
151 * @return 0 on success, negative error code on failure
152 */
153int av_tx_init(AVTXContext **ctx, av_tx_fn *tx, enum AVTXType type,
154               int inv, int len, const void *scale, uint64_t flags);
155
156/**
157 * Frees a context and sets *ctx to NULL, does nothing when *ctx == NULL.
158 */
159void av_tx_uninit(AVTXContext **ctx);
160
161#endif /* AVUTIL_TX_H */
162