virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
Code Issues Releases Activity
d4d09c8e42
FFmpeg/libavfilter/avf_showfreqs.c
Andreas Rheinhardt 790f793844 avutil/common: Don't auto-include mem.h 
There are lots of files that don't need it: The number of object
files that actually need it went down from 2011 to 884 here.

Keep it for external users in order to not cause breakages.

Also improve the other headers a bit while just at it.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2024-03-31 00:08:43 +01:00

571 lines
19 KiB
C
Raw Blame History

/*
* Copyright (c) 2015 Paul B Mahol
*
* 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 <float.h>
#include <math.h>
#include "libavutil/mem.h"
#include "libavutil/tx.h"
#include "libavutil/avassert.h"
#include "libavutil/avstring.h"
#include "libavutil/channel_layout.h"
#include "libavutil/intreadwrite.h"
#include "libavutil/opt.h"
#include "libavutil/parseutils.h"
#include "audio.h"
#include "filters.h"
#include "formats.h"
#include "video.h"
#include "avfilter.h"
#include "internal.h"
#include "window_func.h"
enum DataMode { MAGNITUDE, PHASE, DELAY, NB_DATA };
enum DisplayMode { LINE, BAR, DOT, NB_MODES };
enum ChannelMode { COMBINED, SEPARATE, NB_CMODES };
enum FrequencyScale { FS_LINEAR, FS_LOG, FS_RLOG, NB_FSCALES };
enum AmplitudeScale { AS_LINEAR, AS_SQRT, AS_CBRT, AS_LOG, NB_ASCALES };
typedef struct ShowFreqsContext {
const AVClass *class;
int w, h;
int mode;
int data_mode;
int cmode;
int fft_size;
int ascale, fscale;
int avg;
int win_func;
char *ch_layout_str;
uint8_t *bypass;
AVChannelLayout ch_layout;
AVTXContext *fft;
av_tx_fn tx_fn;
AVComplexFloat **fft_input;
AVComplexFloat **fft_data;
AVFrame *window;
float **avg_data;
float *window_func_lut;
float overlap;
float minamp;
int hop_size;
int nb_channels;
int nb_draw_channels;
int nb_freq;
int win_size;
float scale;
char *colors;
int64_t pts;
int64_t old_pts;
AVRational frame_rate;
} ShowFreqsContext;
#define OFFSET(x) offsetof(ShowFreqsContext, x)
#define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
static const AVOption showfreqs_options[] = {
{ "size", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "1024x512"}, 0, 0, FLAGS },
{ "s", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "1024x512"}, 0, 0, FLAGS },
{ "rate", "set video rate", OFFSET(frame_rate), AV_OPT_TYPE_VIDEO_RATE, {.str = "25"}, 0, INT_MAX, FLAGS },
{ "r", "set video rate", OFFSET(frame_rate), AV_OPT_TYPE_VIDEO_RATE, {.str = "25"}, 0, INT_MAX, FLAGS },
{ "mode", "set display mode", OFFSET(mode), AV_OPT_TYPE_INT, {.i64=BAR}, 0, NB_MODES-1, FLAGS, .unit = "mode" },
{ "line", "show lines", 0, AV_OPT_TYPE_CONST, {.i64=LINE}, 0, 0, FLAGS, .unit = "mode" },
{ "bar", "show bars", 0, AV_OPT_TYPE_CONST, {.i64=BAR}, 0, 0, FLAGS, .unit = "mode" },
{ "dot", "show dots", 0, AV_OPT_TYPE_CONST, {.i64=DOT}, 0, 0, FLAGS, .unit = "mode" },
{ "ascale", "set amplitude scale", OFFSET(ascale), AV_OPT_TYPE_INT, {.i64=AS_LOG}, 0, NB_ASCALES-1, FLAGS, .unit = "ascale" },
{ "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=AS_LINEAR}, 0, 0, FLAGS, .unit = "ascale" },
{ "sqrt", "square root", 0, AV_OPT_TYPE_CONST, {.i64=AS_SQRT}, 0, 0, FLAGS, .unit = "ascale" },
{ "cbrt", "cubic root", 0, AV_OPT_TYPE_CONST, {.i64=AS_CBRT}, 0, 0, FLAGS, .unit = "ascale" },
{ "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=AS_LOG}, 0, 0, FLAGS, .unit = "ascale" },
{ "fscale", "set frequency scale", OFFSET(fscale), AV_OPT_TYPE_INT, {.i64=FS_LINEAR}, 0, NB_FSCALES-1, FLAGS, .unit = "fscale" },
{ "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=FS_LINEAR}, 0, 0, FLAGS, .unit = "fscale" },
{ "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=FS_LOG}, 0, 0, FLAGS, .unit = "fscale" },
{ "rlog", "reverse logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=FS_RLOG}, 0, 0, FLAGS, .unit = "fscale" },
{ "win_size", "set window size", OFFSET(fft_size), AV_OPT_TYPE_INT, {.i64=2048}, 16, 65536, FLAGS },
WIN_FUNC_OPTION("win_func", OFFSET(win_func), FLAGS, WFUNC_HANNING),
{ "overlap", "set window overlap", OFFSET(overlap), AV_OPT_TYPE_FLOAT, {.dbl=1.}, 0., 1., FLAGS },
{ "averaging", "set time averaging", OFFSET(avg), AV_OPT_TYPE_INT, {.i64=1}, 0, INT32_MAX, FLAGS },
{ "colors", "set channels colors", OFFSET(colors), AV_OPT_TYPE_STRING, {.str = "red|green|blue|yellow|orange|lime|pink|magenta|brown" }, 0, 0, FLAGS },
{ "cmode", "set channel mode", OFFSET(cmode), AV_OPT_TYPE_INT, {.i64=COMBINED}, 0, NB_CMODES-1, FLAGS, .unit = "cmode" },
{ "combined", "show all channels in same window", 0, AV_OPT_TYPE_CONST, {.i64=COMBINED}, 0, 0, FLAGS, .unit = "cmode" },
{ "separate", "show each channel in own window", 0, AV_OPT_TYPE_CONST, {.i64=SEPARATE}, 0, 0, FLAGS, .unit = "cmode" },
{ "minamp", "set minimum amplitude", OFFSET(minamp), AV_OPT_TYPE_FLOAT, {.dbl=1e-6}, FLT_MIN, 1e-6, FLAGS },
{ "data", "set data mode", OFFSET(data_mode), AV_OPT_TYPE_INT, {.i64=MAGNITUDE}, 0, NB_DATA-1, FLAGS, .unit = "data" },
{ "magnitude", "show magnitude", 0, AV_OPT_TYPE_CONST, {.i64=MAGNITUDE}, 0, 0, FLAGS, .unit = "data" },
{ "phase", "show phase", 0, AV_OPT_TYPE_CONST, {.i64=PHASE}, 0, 0, FLAGS, .unit = "data" },
{ "delay", "show group delay",0, AV_OPT_TYPE_CONST, {.i64=DELAY}, 0, 0, FLAGS, .unit = "data" },
{ "channels", "set channels to draw", OFFSET(ch_layout_str), AV_OPT_TYPE_STRING, {.str="all"}, 0, 0, FLAGS },
{ NULL }
};
AVFILTER_DEFINE_CLASS(showfreqs);
static int query_formats(AVFilterContext *ctx)
{
AVFilterFormats *formats = NULL;
AVFilterChannelLayouts *layouts = NULL;
AVFilterLink *inlink = ctx->inputs[0];
AVFilterLink *outlink = ctx->outputs[0];
static const enum AVSampleFormat sample_fmts[] = { AV_SAMPLE_FMT_FLTP, AV_SAMPLE_FMT_NONE };
static const enum AVPixelFormat pix_fmts[] = { AV_PIX_FMT_RGBA, AV_PIX_FMT_NONE };
int ret;
/* set input audio formats */
formats = ff_make_format_list(sample_fmts);
if ((ret = ff_formats_ref(formats, &inlink->outcfg.formats)) < 0)
return ret;
layouts = ff_all_channel_counts();
if ((ret = ff_channel_layouts_ref(layouts, &inlink->outcfg.channel_layouts)) < 0)
return ret;
formats = ff_all_samplerates();
if ((ret = ff_formats_ref(formats, &inlink->outcfg.samplerates)) < 0)
return ret;
/* set output video format */
formats = ff_make_format_list(pix_fmts);
if ((ret = ff_formats_ref(formats, &outlink->incfg.formats)) < 0)
return ret;
return 0;
}
static int config_output(AVFilterLink *outlink)
{
AVFilterContext *ctx = outlink->src;
AVFilterLink *inlink = ctx->inputs[0];
ShowFreqsContext *s = ctx->priv;
float overlap, scale = 1.f;
int i, ret;
s->old_pts = AV_NOPTS_VALUE;
s->nb_freq = s->fft_size / 2;
s->win_size = s->fft_size;
av_tx_uninit(&s->fft);
ret = av_tx_init(&s->fft, &s->tx_fn, AV_TX_FLOAT_FFT, 0, s->fft_size, &scale, 0);
if (ret < 0) {
av_log(ctx, AV_LOG_ERROR, "Unable to create FFT context. "
"The window size might be too high.\n");
return ret;
}
/* FFT buffers: x2 for each (display) channel buffer.
* Note: we use free and malloc instead of a realloc-like function to
* make sure the buffer is aligned in memory for the FFT functions. */
for (i = 0; i < s->nb_channels; i++) {
av_freep(&s->fft_input[i]);
av_freep(&s->fft_data[i]);
av_freep(&s->avg_data[i]);
}
av_freep(&s->bypass);
av_freep(&s->fft_input);
av_freep(&s->fft_data);
av_freep(&s->avg_data);
s->nb_channels = inlink->ch_layout.nb_channels;
s->bypass = av_calloc(s->nb_channels, sizeof(*s->bypass));
if (!s->bypass)
return AVERROR(ENOMEM);
s->fft_input = av_calloc(s->nb_channels, sizeof(*s->fft_input));
if (!s->fft_input)
return AVERROR(ENOMEM);
s->fft_data = av_calloc(s->nb_channels, sizeof(*s->fft_data));
if (!s->fft_data)
return AVERROR(ENOMEM);
s->avg_data = av_calloc(s->nb_channels, sizeof(*s->avg_data));
if (!s->avg_data)
return AVERROR(ENOMEM);
for (i = 0; i < s->nb_channels; i++) {
s->fft_input[i] = av_calloc(FFALIGN(s->win_size, 512), sizeof(**s->fft_input));
s->fft_data[i] = av_calloc(FFALIGN(s->win_size, 512), sizeof(**s->fft_data));
s->avg_data[i] = av_calloc(s->nb_freq, sizeof(**s->avg_data));
if (!s->fft_data[i] || !s->avg_data[i] || !s->fft_input[i])
return AVERROR(ENOMEM);
}
/* pre-calc windowing function */
s->window_func_lut = av_realloc_f(s->window_func_lut, s->win_size,
sizeof(*s->window_func_lut));
if (!s->window_func_lut)
return AVERROR(ENOMEM);
generate_window_func(s->window_func_lut, s->win_size, s->win_func, &overlap);
if (s->overlap == 1.)
s->overlap = overlap;
s->hop_size = (1. - s->overlap) * s->win_size;
if (s->hop_size < 1) {
av_log(ctx, AV_LOG_ERROR, "overlap %f too big\n", s->overlap);
return AVERROR(EINVAL);
}
for (s->scale = 0, i = 0; i < s->win_size; i++) {
s->scale += s->window_func_lut[i] * s->window_func_lut[i];
}
s->window = ff_get_audio_buffer(inlink, s->win_size * 2);
if (!s->window)
return AVERROR(ENOMEM);
outlink->frame_rate = s->frame_rate;
outlink->time_base = av_inv_q(outlink->frame_rate);
outlink->sample_aspect_ratio = (AVRational){1,1};
outlink->w = s->w;
outlink->h = s->h;
ret = av_channel_layout_copy(&s->ch_layout, &inlink->ch_layout);
if (ret < 0)
return ret;
s->nb_draw_channels = s->nb_channels;
if (strcmp(s->ch_layout_str, "all")) {
int nb_draw_channels = 0;
av_channel_layout_from_string(&s->ch_layout,
s->ch_layout_str);
for (int ch = 0; ch < s->nb_channels; ch++) {
const enum AVChannel channel = av_channel_layout_channel_from_index(&inlink->ch_layout, ch);
s->bypass[ch] = av_channel_layout_index_from_channel(&s->ch_layout, channel) < 0;
nb_draw_channels += s->bypass[ch] == 0;
}
s->nb_draw_channels = nb_draw_channels;
}
return 0;
}
static inline void draw_dot(AVFrame *out, int x, int y, uint8_t fg[4])
{
uint32_t color = AV_RL32(out->data[0] + y * out->linesize[0] + x * 4);
if ((color & 0xffffff) != 0)
AV_WL32(out->data[0] + y * out->linesize[0] + x * 4, AV_RL32(fg) | color);
else
AV_WL32(out->data[0] + y * out->linesize[0] + x * 4, AV_RL32(fg));
}
static int get_sx(ShowFreqsContext *s, int f)
{
switch (s->fscale) {
case FS_LINEAR:
return (s->w/(float)s->nb_freq)*f;
case FS_LOG:
return s->w-pow(s->w, (s->nb_freq-f-1)/(s->nb_freq-1.));
case FS_RLOG:
return pow(s->w, f/(s->nb_freq-1.));
}
return 0;
}
static float get_bsize(ShowFreqsContext *s, int f)
{
switch (s->fscale) {
case FS_LINEAR:
return s->w/(float)s->nb_freq;
case FS_LOG:
return pow(s->w, (s->nb_freq-f-1)/(s->nb_freq-1.))-
pow(s->w, (s->nb_freq-f-2)/(s->nb_freq-1.));
case FS_RLOG:
return pow(s->w, (f+1)/(s->nb_freq-1.))-
pow(s->w, f /(s->nb_freq-1.));
}
return 1.;
}
static inline void plot_freq(ShowFreqsContext *s, int ch,
double a, int f, uint8_t fg[4], int *prev_y,
AVFrame *out, AVFilterLink *outlink)
{
const int w = s->w;
const float min = s->minamp;
const float avg = s->avg_data[ch][f];
const float bsize = get_bsize(s, f);
const int sx = get_sx(s, f);
int end = outlink->h;
int x, y, i;
switch(s->ascale) {
case AS_SQRT:
a = 1.0 - sqrt(a);
break;
case AS_CBRT:
a = 1.0 - cbrt(a);
break;
case AS_LOG:
a = log(av_clipd(a, min, 1)) / log(min);
break;
case AS_LINEAR:
a = 1.0 - a;
break;
}
switch (s->cmode) {
case COMBINED:
y = a * outlink->h - 1;
break;
case SEPARATE:
end = (outlink->h / s->nb_draw_channels) * (ch + 1);
y = (outlink->h / s->nb_draw_channels) * ch + a * (outlink->h / s->nb_draw_channels) - 1;
break;
default:
av_assert0(0);
}
if (y < 0)
return;
switch (s->avg) {
case 0:
y = s->avg_data[ch][f] = !outlink->frame_count_in ? y : FFMIN(0, y);
break;
case 1:
break;
default:
s->avg_data[ch][f] = avg + y * (y - avg) / (FFMIN(outlink->frame_count_in + 1, s->avg) * (float)y);
y = av_clip(s->avg_data[ch][f], 0, outlink->h - 1);
break;
}
switch(s->mode) {
case LINE:
if (*prev_y == -1) {
*prev_y = y;
}
if (y <= *prev_y) {
for (x = sx + 1; x < sx + bsize && x < w; x++)
draw_dot(out, x, y, fg);
for (i = y; i <= *prev_y; i++)
draw_dot(out, sx, i, fg);
} else {
for (i = *prev_y; i <= y; i++)
draw_dot(out, sx, i, fg);
for (x = sx + 1; x < sx + bsize && x < w; x++)
draw_dot(out, x, i - 1, fg);
}
*prev_y = y;
break;
case BAR:
for (x = sx; x < sx + bsize && x < w; x++)
for (i = y; i < end; i++)
draw_dot(out, x, i, fg);
break;
case DOT:
for (x = sx; x < sx + bsize && x < w; x++)
draw_dot(out, x, y, fg);
break;
}
}
static int plot_freqs(AVFilterLink *inlink, int64_t pts)
{
AVFilterContext *ctx = inlink->dst;
AVFilterLink *outlink = ctx->outputs[0];
ShowFreqsContext *s = ctx->priv;
AVFrame *in = s->window;
const int win_size = s->win_size;
char *colors, *color, *saveptr = NULL;
AVFrame *out;
int ch, n;
/* fill FFT input with the number of samples available */
for (ch = 0; ch < s->nb_channels; ch++) {
const float *p = (float *)in->extended_data[ch];
if (s->bypass[ch])
continue;
for (n = 0; n < win_size; n++) {
s->fft_input[ch][n].re = p[n] * s->window_func_lut[n];
s->fft_input[ch][n].im = 0;
}
}
/* run FFT on each samples set */
for (ch = 0; ch < s->nb_channels; ch++) {
if (s->bypass[ch])
continue;
s->tx_fn(s->fft, s->fft_data[ch], s->fft_input[ch], sizeof(AVComplexFloat));
}
s->pts = av_rescale_q(pts, inlink->time_base, outlink->time_base);
if (s->old_pts >= s->pts)
return 0;
s->old_pts = s->pts;
#define RE(x, ch) s->fft_data[ch][x].re
#define IM(x, ch) s->fft_data[ch][x].im
#define M(a, b) (sqrt((a) * (a) + (b) * (b)))
#define P(a, b) (atan2((b), (a)))
colors = av_strdup(s->colors);
if (!colors)
return AVERROR(ENOMEM);
out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
if (!out) {
av_free(colors);
return AVERROR(ENOMEM);
}
for (n = 0; n < outlink->h; n++)
memset(out->data[0] + out->linesize[0] * n, 0, outlink->w * 4);
for (ch = 0; ch < s->nb_channels; ch++) {
uint8_t fg[4] = { 0xff, 0xff, 0xff, 0xff };
int prev_y = -1, f;
double a;
color = av_strtok(ch == 0 ? colors : NULL, " |", &saveptr);
if (color)
av_parse_color(fg, color, -1, ctx);
if (s->bypass[ch])
continue;
switch (s->data_mode) {
case MAGNITUDE:
for (f = 0; f < s->nb_freq; f++) {
a = av_clipd(M(RE(f, ch), IM(f, ch)) / s->scale, 0, 1);
plot_freq(s, ch, a, f, fg, &prev_y, out, outlink);
}
break;
case PHASE:
for (f = 0; f < s->nb_freq; f++) {
a = av_clipd((M_PI + P(RE(f, ch), IM(f, ch))) / (2. * M_PI), 0, 1);
plot_freq(s, ch, a, f, fg, &prev_y, out, outlink);
}
break;
case DELAY:
for (f = 0; f < s->nb_freq; f++) {
a = av_clipd((M_PI - P(IM(f, ch) * RE(f-1, ch) - IM(f-1, ch) * RE(f, ch),
RE(f, ch) * RE(f-1, ch) + IM(f, ch) * IM(f-1, ch))) / (2. * M_PI), 0, 1);
plot_freq(s, ch, a, f, fg, &prev_y, out, outlink);
}
break;
}
}
av_free(colors);
out->pts = s->pts;
out->duration = 1;
out->sample_aspect_ratio = (AVRational){1,1};
return ff_filter_frame(outlink, out);
}
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
{
AVFilterContext *ctx = inlink->dst;
ShowFreqsContext *s = ctx->priv;
const int offset = s->win_size - s->hop_size;
int64_t pts = in->pts;
for (int ch = 0; ch < in->ch_layout.nb_channels; ch++) {
float *dst = (float *)s->window->extended_data[ch];
memmove(dst, &dst[s->hop_size], offset * sizeof(float));
memcpy(&dst[offset], in->extended_data[ch], in->nb_samples * sizeof(float));
memset(&dst[offset + in->nb_samples], 0, (s->hop_size - in->nb_samples) * sizeof(float));
}
av_frame_free(&in);
return plot_freqs(inlink, pts);
}
static int activate(AVFilterContext *ctx)
{
AVFilterLink *inlink = ctx->inputs[0];
AVFilterLink *outlink = ctx->outputs[0];
ShowFreqsContext *s = ctx->priv;
AVFrame *in;
int ret;
FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
ret = ff_inlink_consume_samples(inlink, s->hop_size, s->hop_size, &in);
if (ret < 0)
return ret;
if (ret > 0)
ret = filter_frame(inlink, in);
if (ret < 0)
return ret;
if (ff_inlink_queued_samples(inlink) >= s->hop_size) {
ff_filter_set_ready(ctx, 10);
return 0;
}
FF_FILTER_FORWARD_STATUS(inlink, outlink);
FF_FILTER_FORWARD_WANTED(outlink, inlink);
return FFERROR_NOT_READY;
}
static av_cold void uninit(AVFilterContext *ctx)
{
ShowFreqsContext *s = ctx->priv;
int i;
av_channel_layout_uninit(&s->ch_layout);
av_tx_uninit(&s->fft);
for (i = 0; i < s->nb_channels; i++) {
if (s->fft_input)
av_freep(&s->fft_input[i]);
if (s->fft_data)
av_freep(&s->fft_data[i]);
if (s->avg_data)
av_freep(&s->avg_data[i]);
}
av_freep(&s->bypass);
av_freep(&s->fft_input);
av_freep(&s->fft_data);
av_freep(&s->avg_data);
av_freep(&s->window_func_lut);
av_frame_free(&s->window);
}
static const AVFilterPad showfreqs_outputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
.config_props = config_output,
},
};
const AVFilter ff_avf_showfreqs = {
.name = "showfreqs",
.description = NULL_IF_CONFIG_SMALL("Convert input audio to a frequencies video output."),
.uninit = uninit,
.priv_size = sizeof(ShowFreqsContext),
.activate = activate,
FILTER_INPUTS(ff_audio_default_filterpad),
FILTER_OUTPUTS(showfreqs_outputs),
FILTER_QUERY_FUNC(query_formats),
.priv_class = &showfreqs_class,
};
Reference in New Issue View Git Blame Copy Permalink