Go to the documentation of this file.
46 #define MAX_CHANNELS 63
49 #define PRE_B0 1.53512485958697
50 #define PRE_B1 -2.69169618940638
51 #define PRE_B2 1.19839281085285
52 #define PRE_A1 -1.69065929318241
53 #define PRE_A2 0.73248077421585
59 #define RLB_A1 -1.99004745483398
60 #define RLB_A2 0.99007225036621
63 #define ABS_UP_THRES 10
64 #define HIST_GRAIN 100
65 #define HIST_SIZE ((ABS_UP_THRES - ABS_THRES) * HIST_GRAIN + 1)
101 #if CONFIG_SWRESAMPLE
132 #define I400_BINS (48000 * 4 / 10)
133 #define I3000_BINS (48000 * 3)
168 #define OFFSET(x) offsetof(EBUR128Context, x)
169 #define A AV_OPT_FLAG_AUDIO_PARAM
170 #define V AV_OPT_FLAG_VIDEO_PARAM
171 #define F AV_OPT_FLAG_FILTERING_PARAM
176 {
"framelog",
"force frame logging level",
OFFSET(loglevel),
AV_OPT_TYPE_INT, {.i64 = -1}, INT_MIN, INT_MAX,
A|
V|
F,
"level" },
185 {
"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 },
186 {
"target",
"set a specific target level in LUFS (-23 to 0)",
OFFSET(target),
AV_OPT_TYPE_INT, {.i64 = -23}, -23, 0,
V|
F },
223 const int above_opt_max = y > ebur128->
y_opt_max;
224 const int below_opt_min = y < ebur128->
y_opt_min;
225 const int reached = y >= v;
227 const int colorid = 8*below_opt_min+ 4*
line + 2*reached + above_opt_max;
233 v += 2 * ebur128->
meter;
263 for (
i = 0; buf[
i];
i++) {
267 for (char_y = 0; char_y < font_height; char_y++) {
269 if (font[buf[
i] * font_height + char_y] &
mask)
272 memcpy(p,
"\x00\x00\x00", 3);
285 for (
i = 0;
i <
len;
i++) {
286 memcpy(p,
"\x00\xff\x00", 3);
300 if (ebur128->
w < 640 || ebur128->
h < 480) {
302 "minimum size is 640x480\n", ebur128->
w, ebur128->
h);
305 outlink->
w = ebur128->
w;
306 outlink->
h = ebur128->
h;
313 ebur128->
text.
y = 40;
314 ebur128->
text.
w = 3 * 8;
346 memset(outpicref->
data[0], 0, ebur128->
h * outpicref->
linesize[0]);
352 x =
PAD + (i < 10 && i > -10) * 8;
356 "%c%d", i < 0 ? '-' : i > 0 ?
'+' :
' ',
FFABS(
i));
365 for (y = 0; y < ebur128->
graph.
h; y++) {
368 for (x = 0; x < ebur128->
graph.
w; x++)
369 memcpy(p + x*3,
c, 3);
374 #define DRAW_RECT(r) do { \
375 drawline(outpicref, r.x, r.y - 1, r.w, 3); \
376 drawline(outpicref, r.x, r.y + r.h, r.w, 3); \
377 drawline(outpicref, r.x - 1, r.y, r.h, outpicref->linesize[0]); \
378 drawline(outpicref, r.x + r.w, r.y, r.h, outpicref->linesize[0]); \
411 #define BACK_MASK (AV_CH_BACK_LEFT |AV_CH_BACK_CENTER |AV_CH_BACK_RIGHT| \
412 AV_CH_TOP_BACK_LEFT|AV_CH_TOP_BACK_CENTER|AV_CH_TOP_BACK_RIGHT| \
413 AV_CH_SIDE_LEFT |AV_CH_SIDE_RIGHT| \
414 AV_CH_SURROUND_DIRECT_LEFT |AV_CH_SURROUND_DIRECT_RIGHT)
442 #if CONFIG_SWRESAMPLE
450 if (!ebur128->swr_buf || !ebur128->
true_peaks ||
477 #define ENERGY(loudness) (ff_exp10(((loudness) + 0.691) / 10.))
478 #define LOUDNESS(energy) (-0.691 + 10 * log10(energy))
479 #define DBFS(energy) (20 * log10(energy))
511 "True-peak mode requires libswresample to be performed\n");
561 #define HIST_POS(power) (int)(((power) - ABS_THRES) * HIST_GRAIN)
569 double relative_threshold;
580 if (!relative_threshold)
581 relative_threshold = 1e-12;
585 return gate_hist_pos;
590 int i, ch, idx_insample;
595 const double *
samples = (
double *)insamples->
data[0];
598 #
if CONFIG_SWRESAMPLE
600 const double *swr_samples = ebur128->swr_buf;
607 for (idx_insample = 0; idx_insample <
ret; idx_insample++) {
618 for (idx_insample = 0; idx_insample < nb_samples; idx_insample++) {
622 #define MOVE_TO_NEXT_CACHED_ENTRY(time) do { \
623 ebur128->i##time.cache_pos++; \
624 if (ebur128->i##time.cache_pos == I##time##_BINS) { \
625 ebur128->i##time.filled = 1; \
626 ebur128->i##time.cache_pos = 0; \
645 #define FILTER(Y, X, name) do { \
646 double *dst = ebur128->Y + ch*3; \
647 double *src = ebur128->X + ch*3; \
650 dst[0] = src[0]*name##_B0 + src[1]*name##_B1 + src[2]*name##_B2 \
651 - dst[1]*name##_A1 - dst[2]*name##_A2; \
656 ebur128->
x[ch * 3 + 2] = ebur128->
x[ch * 3 + 1];
657 ebur128->
x[ch * 3 + 1] = ebur128->
x[ch * 3 ];
660 bin = ebur128->
z[ch * 3] * ebur128->
z[ch * 3];
668 ebur128->
i400.
cache [ch][bin_id_400 ] = bin;
676 double loudness_400, loudness_3000;
677 double power_400 = 1e-12, power_3000 = 1e-12;
679 const int64_t
pts = insamples->
pts +
685 #define COMPUTE_LOUDNESS(m, time) do { \
686 if (ebur128->i##time.filled) { \
688 for (ch = 0; ch < nb_channels; ch++) \
689 power_##time += ebur128->ch_weighting[ch] * ebur128->i##time.sum[ch]; \
690 power_##time /= I##time##_BINS; \
692 loudness_##time = LOUDNESS(power_##time); \
699 #define I_GATE_THRES -10 // initially defined to -8 LU in the first EBU standard
702 double integrated_sum = 0;
703 int nb_integrated = 0;
711 nb_integrated += nb_v;
724 #define LRA_GATE_THRES -20
725 #define LRA_LOWER_PRC 10
726 #define LRA_HIGHER_PRC 95
769 loudness_400 -= ebur128->
pan_law;
770 loudness_3000 -= ebur128->
pan_law;
773 #define LOG_FMT "TARGET:%d LUFS M:%6.1f S:%6.1f I:%6.1f %s LRA:%6.1f LU"
781 int y_loudness_lu_graph, y_loudness_lu_gauge;
784 gauge_value = loudness_400 - ebur128->
target;
786 gauge_value = loudness_3000 - ebur128->
target;
789 y_loudness_lu_graph =
lu_to_y(ebur128, loudness_3000 - ebur128->
target);
790 y_loudness_lu_gauge =
lu_to_y(ebur128, gauge_value);
793 p = pic->data[0] + ebur128->
graph.
y*pic->linesize[0] + ebur128->
graph.
x*3;
794 for (y = 0; y < ebur128->
graph.
h; y++) {
797 memmove(p, p + 3, (ebur128->
graph.
w - 1) * 3);
798 memcpy(p + (ebur128->
graph.
w - 1) * 3,
c, 3);
799 p += pic->linesize[0];
803 p = pic->data[0] + ebur128->
gauge.
y*pic->linesize[0] + ebur128->
gauge.
x*3;
804 for (y = 0; y < ebur128->
gauge.
h; y++) {
807 for (x = 0; x < ebur128->
gauge.
w; x++)
808 memcpy(p + x*3,
c, 3);
809 p += pic->linesize[0];
816 ebur128->
target, loudness_400, loudness_3000,
837 #define META_PREFIX "lavfi.r128."
839 #define SET_META(name, var) do { \
840 snprintf(metabuf, sizeof(metabuf), "%.3f", var); \
841 av_dict_set(&insamples->metadata, name, metabuf, 0); \
844 #define SET_META_PEAK(name, ptype) do { \
845 if (ebur128->peak_mode & PEAK_MODE_ ## ptype ## _PEAKS) { \
847 for (ch = 0; ch < nb_channels; ch++) { \
848 snprintf(key, sizeof(key), \
849 META_PREFIX AV_STRINGIFY(name) "_peaks_ch%d", ch); \
850 SET_META(key, ebur128->name##_peaks[ch]); \
869 ebur128->
target, loudness_400, loudness_3000,
878 #define PRINT_PEAKS(str, sp, ptype) do { \
879 if (ebur128->peak_mode & PEAK_MODE_ ## ptype ## _PEAKS) { \
880 av_log(ctx, ebur128->loglevel, " " str ":"); \
881 for (ch = 0; ch < nb_channels; ch++) \
882 av_log(ctx, ebur128->loglevel, " %5.1f", DBFS(sp[ch])); \
883 av_log(ctx, ebur128->loglevel, " dBFS"); \
907 static const int input_srate[] = {48000, -1};
915 outlink =
ctx->outputs[1];
953 " Integrated loudness:\n"
955 " Threshold: %5.1f LUFS\n\n"
958 " Threshold: %5.1f LUFS\n"
959 " LRA low: %5.1f LUFS\n"
960 " LRA high: %5.1f LUFS",
965 #define PRINT_PEAK_SUMMARY(str, sp, ptype) do { \
969 if (ebur128->peak_mode & PEAK_MODE_ ## ptype ## _PEAKS) { \
970 for (ch = 0; ch < ebur128->nb_channels; ch++) \
971 maxpeak = FFMAX(maxpeak, sp[ch]); \
972 av_log(ctx, AV_LOG_INFO, "\n\n " str " peak:\n" \
973 " Peak: %5.1f dBFS", \
993 for (
i = 0;
i <
ctx->nb_outputs;
i++)
996 #if CONFIG_SWRESAMPLE
1021 .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
struct AVFilterChannelLayouts * in_channel_layouts
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)
@ PEAK_MODE_SAMPLES_PEAKS
#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
AVFilterFormats * in_formats
Lists of formats and channel layouts supported by the input and output filters respectively.
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
char * av_asprintf(const char *fmt,...)
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)
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.
AVFilterFormats * in_samplerates
Lists of channel layouts and sample rates used for automatic negotiation.
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
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)
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)
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 const AVFilterPad outputs[]
static enum AVPixelFormat pix_fmts[]
int attribute_align_arg swr_convert(struct SwrContext *s, uint8_t *out_arg[SWR_CH_MAX], int out_count, const uint8_t *in_arg[SWR_CH_MAX], int in_count)
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)
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.
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
uint64_t channel_layout
channel layout of current buffer (see libavutil/channel_layout.h)
struct rect gauge
rectangle for the gauge on the right
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several inputs
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.
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.
AVFilterContext * src
source filter
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
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.
int scale_range
the range of LU values according to the meter
#define SET_META(name, var)
void * av_calloc(size_t nmemb, size_t size)
Non-inlined equivalent of av_mallocz_array().
int h
agreed upon image height
int meter
select a EBU mode between +9 and +18
static int ff_insert_outpad(AVFilterContext *f, unsigned index, AVFilterPad *p)
Insert a new output pad for the filter.
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.
#define FILTER(Y, X, name)
char * av_strdup(const char *s)
Duplicate a string.
static const AVFilterPad ebur128_inputs[]
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)
#define flags(name, subs,...)
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
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
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...