Go to the documentation of this file.
45 #define MAX_CHANNELS 63
48 #define ABS_UP_THRES 10
49 #define HIST_GRAIN 100
50 #define HIST_SIZE ((ABS_UP_THRES - ABS_THRES) * HIST_GRAIN + 1)
156 #define OFFSET(x) offsetof(EBUR128Context, x)
157 #define A AV_OPT_FLAG_AUDIO_PARAM
158 #define V AV_OPT_FLAG_VIDEO_PARAM
159 #define F AV_OPT_FLAG_FILTERING_PARAM
164 {
"framelog",
"force frame logging level",
OFFSET(loglevel),
AV_OPT_TYPE_INT, {.i64 = -1}, INT_MIN, INT_MAX,
A|
V|
F,
"level" },
173 {
"panlaw",
"set a specific pan law for dual-mono files",
OFFSET(pan_law),
AV_OPT_TYPE_DOUBLE, {.dbl = -3.01029995663978}, -10.0, 0.0,
A|
F },
174 {
"target",
"set a specific target level in LUFS (-23 to 0)",
OFFSET(target),
AV_OPT_TYPE_INT, {.i64 = -23}, -23, 0,
V|
F },
211 const int above_opt_max = y > ebur128->
y_opt_max;
212 const int below_opt_min = y < ebur128->
y_opt_min;
213 const int reached = y >= v;
215 const int colorid = 8*below_opt_min+ 4*
line + 2*reached + above_opt_max;
221 v += 2 * ebur128->
meter;
251 for (
i = 0; buf[
i];
i++) {
253 uint8_t *p = pic->
data[0] + y*pic->
linesize[0] + (x +
i*8)*3;
255 for (char_y = 0; char_y < font_height; char_y++) {
257 if (font[buf[
i] * font_height + char_y] &
mask)
260 memcpy(p,
"\x00\x00\x00", 3);
273 for (
i = 0;
i <
len;
i++) {
274 memcpy(p,
"\x00\xff\x00", 3);
288 if (ebur128->
w < 640 || ebur128->
h < 480) {
290 "minimum size is 640x480\n", ebur128->
w, ebur128->
h);
293 outlink->
w = ebur128->
w;
294 outlink->
h = ebur128->
h;
301 ebur128->
text.
y = 40;
302 ebur128->
text.
w = 3 * 8;
334 memset(outpicref->
data[0], 0, ebur128->
h * outpicref->
linesize[0]);
340 x =
PAD + (i < 10 && i > -10) * 8;
344 "%c%d", i < 0 ? '-' : i > 0 ?
'+' :
' ',
FFABS(
i));
353 for (y = 0; y < ebur128->
graph.
h; y++) {
356 for (x = 0; x < ebur128->
graph.
w; x++)
357 memcpy(p + x*3,
c, 3);
362 #define DRAW_RECT(r) do { \
363 drawline(outpicref, r.x, r.y - 1, r.w, 3); \
364 drawline(outpicref, r.x, r.y + r.h, r.w, 3); \
365 drawline(outpicref, r.x - 1, r.y, r.h, outpicref->linesize[0]); \
366 drawline(outpicref, r.x + r.w, r.y, r.h, outpicref->linesize[0]); \
382 double f0 = 1681.974450955533;
383 double G = 3.999843853973347;
384 double Q = 0.7071752369554196;
386 double K = tan(
M_PI * f0 / (
double)
inlink->sample_rate);
387 double Vh = pow(10.0,
G / 20.0);
388 double Vb = pow(Vh, 0.4996667741545416);
390 double a0 = 1.0 + K / Q + K * K;
392 ebur128->
pre_b[0] = (Vh + Vb * K / Q + K * K) /
a0;
393 ebur128->
pre_b[1] = 2.0 * (K * K - Vh) /
a0;
394 ebur128->
pre_b[2] = (Vh - Vb * K / Q + K * K) /
a0;
395 ebur128->
pre_a[1] = 2.0 * (K * K - 1.0) /
a0;
396 ebur128->
pre_a[2] = (1.0 - K / Q + K * K) /
a0;
398 f0 = 38.13547087602444;
399 Q = 0.5003270373238773;
400 K = tan(
M_PI * f0 / (
double)
inlink->sample_rate);
402 ebur128->
rlb_b[0] = 1.0;
403 ebur128->
rlb_b[1] = -2.0;
404 ebur128->
rlb_b[2] = 1.0;
405 ebur128->
rlb_a[1] = 2.0 * (K * K - 1.0) / (1.0 + K / Q + K * K);
406 ebur128->
rlb_a[2] = (1.0 - K / Q + K * K) / (1.0 + K / Q + K * K);
427 #define BACK_MASK (AV_CH_BACK_LEFT |AV_CH_BACK_CENTER |AV_CH_BACK_RIGHT| \
428 AV_CH_TOP_BACK_LEFT|AV_CH_TOP_BACK_CENTER|AV_CH_TOP_BACK_RIGHT| \
429 AV_CH_SIDE_LEFT |AV_CH_SIDE_RIGHT| \
430 AV_CH_SURROUND_DIRECT_LEFT |AV_CH_SURROUND_DIRECT_RIGHT)
437 #define I400_BINS(x) ((x) * 4 / 10)
438 #define I3000_BINS(x) ((x) * 3)
463 #if CONFIG_SWRESAMPLE
471 if (!ebur128->swr_buf || !ebur128->
true_peaks ||
498 #define ENERGY(loudness) (ff_exp10(((loudness) + 0.691) / 10.))
499 #define LOUDNESS(energy) (-0.691 + 10 * log10(energy))
500 #define DBFS(energy) (20 * log10(energy))
532 "True-peak mode requires libswresample to be performed\n");
574 #define HIST_POS(power) (int)(((power) - ABS_THRES) * HIST_GRAIN)
582 double relative_threshold;
593 if (!relative_threshold)
594 relative_threshold = 1e-12;
598 return gate_hist_pos;
603 int i, ch, idx_insample;
608 const double *
samples = (
double *)insamples->
data[0];
611 #
if CONFIG_SWRESAMPLE
613 const double *swr_samples = ebur128->swr_buf;
614 int ret =
swr_convert(ebur128->swr_ctx, (uint8_t**)&ebur128->swr_buf, 19200,
615 (
const uint8_t **)insamples->
data, nb_samples);
620 for (idx_insample = 0; idx_insample <
ret; idx_insample++) {
631 for (idx_insample = 0; idx_insample < nb_samples; idx_insample++) {
635 #define MOVE_TO_NEXT_CACHED_ENTRY(time) do { \
636 ebur128->i##time.cache_pos++; \
637 if (ebur128->i##time.cache_pos == \
638 ebur128->i##time.cache_size) { \
639 ebur128->i##time.filled = 1; \
640 ebur128->i##time.cache_pos = 0; \
659 #define FILTER(Y, X, NUM, DEN) do { \
660 double *dst = ebur128->Y + ch*3; \
661 double *src = ebur128->X + ch*3; \
664 dst[0] = src[0]*NUM[0] + src[1]*NUM[1] + src[2]*NUM[2] \
665 - dst[1]*DEN[1] - dst[2]*DEN[2]; \
670 ebur128->
x[ch * 3 + 2] = ebur128->
x[ch * 3 + 1];
671 ebur128->
x[ch * 3 + 1] = ebur128->
x[ch * 3 ];
674 bin = ebur128->
z[ch * 3] * ebur128->
z[ch * 3];
682 ebur128->
i400.
cache [ch][bin_id_400 ] = bin;
690 double loudness_400, loudness_3000;
691 double power_400 = 1e-12, power_3000 = 1e-12;
693 const int64_t
pts = insamples->
pts +
699 #define COMPUTE_LOUDNESS(m, time) do { \
700 if (ebur128->i##time.filled) { \
702 for (ch = 0; ch < nb_channels; ch++) \
703 power_##time += ebur128->ch_weighting[ch] * ebur128->i##time.sum[ch]; \
704 power_##time /= I##time##_BINS(inlink->sample_rate); \
706 loudness_##time = LOUDNESS(power_##time); \
713 #define I_GATE_THRES -10 // initially defined to -8 LU in the first EBU standard
716 double integrated_sum = 0;
717 int nb_integrated = 0;
725 nb_integrated += nb_v;
738 #define LRA_GATE_THRES -20
739 #define LRA_LOWER_PRC 10
740 #define LRA_HIGHER_PRC 95
783 loudness_400 -= ebur128->
pan_law;
784 loudness_3000 -= ebur128->
pan_law;
787 #define LOG_FMT "TARGET:%d LUFS M:%6.1f S:%6.1f I:%6.1f %s LRA:%6.1f LU"
795 int y_loudness_lu_graph, y_loudness_lu_gauge;
798 gauge_value = loudness_400 - ebur128->
target;
800 gauge_value = loudness_3000 - ebur128->
target;
803 y_loudness_lu_graph =
lu_to_y(ebur128, loudness_3000 - ebur128->
target);
804 y_loudness_lu_gauge =
lu_to_y(ebur128, gauge_value);
807 p = pic->data[0] + ebur128->
graph.
y*pic->linesize[0] + ebur128->
graph.
x*3;
808 for (y = 0; y < ebur128->
graph.
h; y++) {
811 memmove(p, p + 3, (ebur128->
graph.
w - 1) * 3);
812 memcpy(p + (ebur128->
graph.
w - 1) * 3,
c, 3);
813 p += pic->linesize[0];
817 p = pic->data[0] + ebur128->
gauge.
y*pic->linesize[0] + ebur128->
gauge.
x*3;
818 for (y = 0; y < ebur128->
gauge.
h; y++) {
821 for (x = 0; x < ebur128->
gauge.
w; x++)
822 memcpy(p + x*3,
c, 3);
823 p += pic->linesize[0];
830 ebur128->
target, loudness_400, loudness_3000,
851 #define META_PREFIX "lavfi.r128."
853 #define SET_META(name, var) do { \
854 snprintf(metabuf, sizeof(metabuf), "%.3f", var); \
855 av_dict_set(&insamples->metadata, name, metabuf, 0); \
858 #define SET_META_PEAK(name, ptype) do { \
859 if (ebur128->peak_mode & PEAK_MODE_ ## ptype ## _PEAKS) { \
861 for (ch = 0; ch < nb_channels; ch++) { \
862 snprintf(key, sizeof(key), \
863 META_PREFIX AV_STRINGIFY(name) "_peaks_ch%d", ch); \
864 SET_META(key, ebur128->name##_peaks[ch]); \
883 ebur128->
target, loudness_400, loudness_3000,
892 #define PRINT_PEAKS(str, sp, ptype) do { \
893 if (ebur128->peak_mode & PEAK_MODE_ ## ptype ## _PEAKS) { \
894 av_log(ctx, ebur128->loglevel, " " str ":"); \
895 for (ch = 0; ch < nb_channels; ch++) \
896 av_log(ctx, ebur128->loglevel, " %5.1f", DBFS(sp[ch])); \
897 av_log(ctx, ebur128->loglevel, " dBFS"); \
928 outlink =
ctx->outputs[1];
966 " Integrated loudness:\n"
968 " Threshold: %5.1f LUFS\n\n"
971 " Threshold: %5.1f LUFS\n"
972 " LRA low: %5.1f LUFS\n"
973 " LRA high: %5.1f LUFS",
978 #define PRINT_PEAK_SUMMARY(str, sp, ptype) do { \
982 if (ebur128->peak_mode & PEAK_MODE_ ## ptype ## _PEAKS) { \
983 for (ch = 0; ch < ebur128->nb_channels; ch++) \
984 maxpeak = FFMAX(maxpeak, sp[ch]); \
985 av_log(ctx, AV_LOG_INFO, "\n\n " str " peak:\n" \
986 " Peak: %5.1f dBFS", \
1007 #if CONFIG_SWRESAMPLE
1031 .priv_class = &ebur128_class,
AVFrame * ff_get_video_buffer(AVFilterLink *link, int w, int h)
Request a picture buffer with a specific set of permissions.
A list of supported channel layouts.
AVPixelFormat
Pixel format.
int dual_mono
whether or not to treat single channel input files as dual-mono
double * ch_weighting
channel weighting mapping
int y_opt_min
the y value (pixel position) for -1 LU
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
int filled
1 if the cache is completely filled, 0 otherwise
int gauge_type
whether gauge shows momentary or short
int y_opt_max
the y value (pixel position) for 1 LU
#define SET_META_PEAK(name, ptype)
#define PRINT_PEAKS(str, sp, ptype)
static const uint8_t * get_graph_color(const EBUR128Context *ebur128, int v, int y)
#define AV_CH_LOW_FREQUENCY_2
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
static enum AVSampleFormat sample_fmts[]
enum MovChannelLayoutTag * layouts
int do_video
1 if video output enabled, 0 otherwise
int sample_count
sample count used for refresh frequency, reset at refresh
The exact code depends on how similar the blocks are and how related they are to the and needs to apply these operations to the correct inlink or outlink if there are several Macros are available to factor that when no extra processing is inlink
int count
how many times the corresponding value occurred
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
This structure describes decoded (raw) audio or video data.
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
trying all byte sequences megabyte in length and selecting the best looking sequence will yield cases to try But a word about which is also called distortion Distortion can be quantified by almost any quality measurement one chooses the sum of squared differences is used but more complex methods that consider psychovisual effects can be used as well It makes no difference in this discussion First step
AVFILTER_DEFINE_CLASS(ebur128)
#define FILTER_QUERY_FUNC(func)
static const AVOption ebur128_options[]
#define AV_LOG_VERBOSE
Detailed information.
static av_cold int init(AVFilterContext *ctx)
const char * name
Filter name.
static int config_audio_output(AVFilterLink *outlink)
double sum_kept_powers
sum of the powers (weighted sums) above absolute threshold
double z[MAX_CHANNELS *3]
3 RLB-filter samples cache for each channel
A link between two filters.
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
double * cache[MAX_CHANNELS]
window of filtered samples (N ms)
static av_cold void uninit(AVFilterContext *ctx)
double pan_law
pan law value used to calculate dual-mono measurements
double rlb_b[3]
rlb-filter numerator coefficients
static void drawline(AVFrame *pic, int x, int y, int len, int step)
static struct hist_entry * get_histogram(void)
int scale
display scale type of statistics
static int query_formats(AVFilterContext *ctx)
double rlb_a[3]
rlb-filter denominator coefficients
static av_always_inline float scale(float x, float s)
int cache_pos
focus on the last added bin in the cache array
A filter pad used for either input or output.
static const uint8_t font_colors[]
int metadata
whether or not to inject loudness results in frames
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
#define MOVE_TO_NEXT_CACHED_ENTRY(time)
#define FILTER(Y, X, NUM, DEN)
struct rect graph
rectangle for the main graph in the center
static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
av_cold int swr_init(struct SwrContext *s)
Initialize context after user parameters have been set.
#define AV_CH_LOW_FREQUENCY
static int config_video_output(AVFilterLink *outlink)
static const uint16_t mask[17]
double integrated_loudness
integrated loudness in LUFS (I)
int target
target level in LUFS used to set relative zero LU in visualization
AVRational sample_aspect_ratio
agreed upon sample aspect ratio
av_cold struct SwrContext * swr_alloc(void)
Allocate SwrContext.
#define av_assert0(cond)
assert() equivalent, that is always enabled.
static enum AVPixelFormat pix_fmts[]
AVFrame * av_frame_clone(const AVFrame *src)
Create a new frame that references the same data as src.
int64_t av_rescale_q(int64_t a, AVRational bq, AVRational cq)
Rescale a 64-bit integer by 2 rational numbers.
static void drawtext(AVFrame *pic, int x, int y, int ftid, const uint8_t *color, const char *fmt,...)
The libswresample context.
double loudness
L = -0.691 + 10 * log10(E)
#define FILTER_INPUTS(array)
double * true_peaks_per_frame
true peaks in a frame per channel
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
double * sample_peaks
sample peaks per channel
#define HIST_GRAIN
defines histogram precision
Describe the class of an AVClass context structure.
static __device__ float fabs(float a)
int peak_mode
enabled peak modes
Rational number (pair of numerator and denominator).
@ AV_OPT_TYPE_IMAGE_SIZE
offset must point to two consecutive integers
double pre_b[3]
pre-filter numerator coefficients
uint64_t channel_layout
channel layout of current buffer (see libavutil/channel_layout.h)
struct rect gauge
rectangle for the gauge on the right
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
int av_get_channel_layout_nb_channels(uint64_t channel_layout)
Return the number of channels in the channel layout.
int av_opt_set_int(void *obj, const char *name, int64_t val, int search_flags)
#define AVFILTER_FLAG_DYNAMIC_OUTPUTS
The number of the filter outputs is not determined just by AVFilter.outputs.
double pre_a[3]
pre-filter denominator coefficients
int y_zero_lu
the y value (pixel position) for 0 LU
struct integrator i3000
3s integrator, used for Short term loudness (S), and Loudness Range (LRA)
@ AV_PIX_FMT_RGB24
packed RGB 8:8:8, 24bpp, RGBRGB...
#define av_ts2timestr(ts, tb)
Convenience macro, the return value should be used only directly in function arguments but never stan...
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
int format
agreed upon media format
av_cold void swr_free(SwrContext **ss)
Free the given SwrContext and set the pointer to NULL.
int attribute_align_arg swr_convert(struct SwrContext *s, uint8_t **out_arg, int out_count, const uint8_t **in_arg, int in_count)
Convert audio.
AVFilterContext * src
source filter
AVFilterFormatsConfig incfg
Lists of supported formats / etc.
AVFrame * outpicref
output picture reference, updated regularly
double y[MAX_CHANNELS *3]
3 pre-filter samples cache for each channel
int h
size of the video output
#define AV_LOG_INFO
Standard information.
#define PRINT_PEAK_SUMMARY(str, sp, ptype)
int sample_rate
samples per second
int nb_samples
number of audio samples (per channel) described by this frame
const uint8_t avpriv_vga16_font[4096]
#define i(width, name, range_min, range_max)
struct rect text
rectangle for the LU legend on the left
uint64_t av_channel_layout_extract_channel(uint64_t channel_layout, int index)
Get the channel with the given index in channel_layout.
int w
agreed upon image width
int * y_line_ref
y reference values for drawing the LU lines in the graph and the gauge
#define av_malloc_array(a, b)
AVSampleFormat
Audio sample formats.
double rel_threshold
relative threshold
#define ABS_THRES
silence gate: we discard anything below this absolute (LUFS) threshold
const char * name
Pad name.
int nb_kept_powers
number of sum above absolute threshold
double * true_peaks
true peaks per channel
void * av_calloc(size_t nmemb, size_t size)
double loudness_range
loudness range in LU (LRA)
double energy
E = 10^((L + 0.691) / 10)
static const uint8_t graph_colors[]
double lra_high
low and high LRA values
AVRational sample_aspect_ratio
Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
const AVFilter ff_af_ebur128
int scale_range
the range of LU values according to the meter
static float power(float r, float g, float b, float max)
#define SET_META(name, var)
int h
agreed upon image height
int meter
select a EBU mode between +9 and +18
Filter the word “frame” indicates either a video frame or a group of audio samples
AVRational time_base
Define the time base used by the PTS of the frames/samples which will pass through this link.
static const AVFilterPad ebur128_inputs[]
int ff_append_outpad(AVFilterContext *f, AVFilterPad *p)
const uint8_t avpriv_cga_font[2048]
int nb_channels
number of channels in the input
struct integrator i400
400ms integrator, used for Momentary loudness (M), and Integrated loudness (I)
int linesize[AV_NUM_DATA_POINTERS]
For video, a positive or negative value, which is typically indicating the size in bytes of each pict...
static int config_audio_input(AVFilterLink *inlink)
struct hist_entry * histogram
histogram of the powers, used to compute LRA and I
@ AV_SAMPLE_FMT_DBL
double
static int lu_to_y(const EBUR128Context *ebur128, double v)
int av_opt_set_sample_fmt(void *obj, const char *name, enum AVSampleFormat fmt, int search_flags)
double sum[MAX_CHANNELS]
sum of the last N ms filtered samples (cache content)
int loglevel
log level for frame logging
double x[MAX_CHANNELS *3]
3 input samples cache for each channel
@ PEAK_MODE_SAMPLES_PEAKS
static int gate_update(struct integrator *integ, double power, double loudness, int gate_thres)
#define COMPUTE_LOUDNESS(m, time)
A histogram is an array of HIST_SIZE hist_entry storing all the energies recorded (with an accuracy o...