mirrors/gstreamer
mirrors/gstreamer
1
1
Fork 0
mirror of https://gitlab.freedesktop.org/gstreamer/gstreamer.git synced 2024-11-30 21:51:09 +00:00
Code Issues 1 Projects Releases Wiki Activity
gstreamer/gst/timecode/gsttimecodestamper.c

1991 lines
69 KiB
C
Raw Normal View History

timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
/*
* GStreamer
* Copyright (C) 2016 Vivia Nikolaidou <vivia@toolsonair.com>
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
* Copyright (C) 2019 Sebastian Dröge <sebastian@centricular.com>
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
*
* gsttimecodestamper.c
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library 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
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
* Boston, MA 02110-1301, USA.
*/
/**
* SECTION:element-timecodestamper
docs: Port all docstring to gtk-doc markdown
2017-03-08 18:01:13 +00:00
* @title: timecodestamper
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
* @short_description: Attach a timecode into incoming video frames
*
* This element attaches a timecode into every incoming video frame. It starts
* counting from the stream time of each segment start, which it converts into
* a timecode.
*
docs: Port all docstring to gtk-doc markdown
2017-03-08 18:01:13 +00:00
* ## Example launch line
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
* |[
* gst-launch-1.0 videotestsrc ! timecodestamper ! autovideosink
* ]|
docs: Port all docstring to gtk-doc markdown
2017-03-08 18:01:13 +00:00
*
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "gsttimecodestamper.h"
#include <gst/gst.h>
#include <gst/video/video.h>
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
#include <gst/audio/audio.h>
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
#include <stdlib.h>
#include <string.h>
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
#define ABSDIFF(a,b) (((a) > (b)) ? ((a) - (b)) : ((b) - (a)))
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
GST_DEBUG_CATEGORY_STATIC (timecodestamper_debug);
#define GST_CAT_DEFAULT timecodestamper_debug
/* GstTimeCodeStamper properties */
enum
{
PROP_0,
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
PROP_SOURCE,
PROP_SET,
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
PROP_AUTO_RESYNC,
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
PROP_TIMEOUT,
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
PROP_DROP_FRAME,
timecodestamper: First timecode property Add an new property to start from a given timecode instead of zero. https://bugzilla.gnome.org/show_bug.cgi?id=778703
2017-02-22 21:26:57 +00:00
PROP_POST_MESSAGES,
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
PROP_SET_INTERNAL_TIMECODE,
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
PROP_LTC_DAILY_JAM,
PROP_LTC_AUTO_RESYNC,
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
PROP_LTC_EXTRA_LATENCY,
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
PROP_LTC_TIMEOUT,
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
PROP_RTC_MAX_DRIFT,
PROP_RTC_AUTO_RESYNC,
PROP_TIMECODE_OFFSET
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
};
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
#define DEFAULT_SOURCE GST_TIME_CODE_STAMPER_SOURCE_INTERNAL
#define DEFAULT_SET GST_TIME_CODE_STAMPER_SET_KEEP
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
#define DEFAULT_AUTO_RESYNC TRUE
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
#define DEFAULT_TIMEOUT GST_CLOCK_TIME_NONE
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
#define DEFAULT_DROP_FRAME FALSE
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
#define DEFAULT_POST_MESSAGES FALSE
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
#define DEFAULT_SET_INTERNAL_TIMECODE NULL
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#define DEFAULT_LTC_DAILY_JAM NULL
#define DEFAULT_LTC_AUTO_RESYNC TRUE
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
#define DEFAULT_LTC_TIMEOUT GST_CLOCK_TIME_NONE
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
#define DEFAULT_LTC_EXTRA_LATENCY (150 * GST_MSECOND)
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
#define DEFAULT_RTC_MAX_DRIFT 250000000
#define DEFAULT_RTC_AUTO_RESYNC TRUE
#define DEFAULT_TIMECODE_OFFSET 0
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#define DEFAULT_LTC_QUEUE 100
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
static GstStaticPadTemplate gst_timecodestamper_src_template =
GST_STATIC_PAD_TEMPLATE ("src",
GST_PAD_SRC,
GST_PAD_ALWAYS,
timecodestamper: Require a non-0/1 framerate on the pad templates We reject caps with other framerates as it's impossible to generate timecodes unless we actually know a constant framerate. Reflect this also in the pad template caps.
2019-08-06 15:51:54 +00:00
GST_STATIC_CAPS ("video/x-raw, framerate=[1/2147483647, 2147483647/1]")
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
);
static GstStaticPadTemplate gst_timecodestamper_sink_template =
GST_STATIC_PAD_TEMPLATE ("sink",
GST_PAD_SINK,
GST_PAD_ALWAYS,
timecodestamper: Require a non-0/1 framerate on the pad templates We reject caps with other framerates as it's impossible to generate timecodes unless we actually know a constant framerate. Reflect this also in the pad template caps.
2019-08-06 15:51:54 +00:00
GST_STATIC_CAPS ("video/x-raw, framerate=[1/2147483647, 2147483647/1]")
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
static GstStaticPadTemplate gst_timecodestamper_ltc_template =
GST_STATIC_PAD_TEMPLATE ("ltc_sink",
GST_PAD_SINK,
GST_PAD_REQUEST,
GST_STATIC_CAPS ("audio/x-raw,format=U8,rate=[1,max],channels=1")
);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
static void gst_timecodestamper_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec);
static void gst_timecodestamper_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec);
static void gst_timecodestamper_dispose (GObject * object);
static gboolean gst_timecodestamper_sink_event (GstBaseTransform * trans,
GstEvent * event);
static GstFlowReturn gst_timecodestamper_transform_ip (GstBaseTransform *
vfilter, GstBuffer * buffer);
static gboolean gst_timecodestamper_stop (GstBaseTransform * trans);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
static gboolean gst_timecodestamper_start (GstBaseTransform * trans);
static GstPad *gst_timecodestamper_request_new_pad (GstElement * element,
GstPadTemplate * temp, const gchar * unused, const GstCaps * caps);
static void gst_timecodestamper_release_pad (GstElement * element,
GstPad * pad);
#if HAVE_LTC
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
typedef struct
{
GstClockTime running_time;
GstVideoTimeCode timecode;
} TimestampedTimecode;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
static gboolean gst_timecodestamper_query (GstBaseTransform * trans,
GstPadDirection direction, GstQuery * query);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
static GstFlowReturn gst_timecodestamper_ltcpad_chain (GstPad * pad,
GstObject * parent, GstBuffer * buffer);
static gboolean gst_timecodestamper_ltcpad_event (GstPad * pad,
GstObject * parent, GstEvent * event);
static gboolean gst_timecodestamper_ltcpad_query (GstPad * pad,
GstObject * parent, GstQuery * query);
static gboolean gst_timecodestamper_ltcpad_activatemode (GstPad * pad,
GstObject * parent, GstPadMode mode, gboolean active);
static gboolean gst_timecodestamper_videopad_activatemode (GstPad * pad,
GstObject * parent, GstPadMode mode, gboolean active);
static GstIterator *gst_timecodestamper_src_iterate_internal_link (GstPad * pad,
GstObject * parent);
#endif
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
static void gst_timecodestamper_update_drop_frame (GstTimeCodeStamper *
timecodestamper);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
G_DEFINE_TYPE (GstTimeCodeStamper, gst_timecodestamper,
GST_TYPE_BASE_TRANSFORM);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GType
gst_timecodestamper_source_get_type (void)
{
static GType type = 0;
static const GEnumValue values[] = {
{GST_TIME_CODE_STAMPER_SOURCE_INTERNAL,
"Use internal timecode counter, starting at zero or value set by property",
"internal"},
{GST_TIME_CODE_STAMPER_SOURCE_ZERO,
"Always use zero", "zero"},
{GST_TIME_CODE_STAMPER_SOURCE_LAST_KNOWN,
"Count up from the last known upstream timecode or internal if unknown",
"last-known"},
timecodestamper: Add last-known-or-zero mode This uses the last known upstream timecode (counted up per frame), or otherwise zero if none was known. The normal last-known timestamp uses the internal timecode as fallback if no upstream timecode was ever known.
2019-10-23 11:31:16 +00:00
{GST_TIME_CODE_STAMPER_SOURCE_LAST_KNOWN_OR_ZERO,
"Count up from the last known upstream timecode or zero if unknown",
"last-known-or-zero"},
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
{GST_TIME_CODE_STAMPER_SOURCE_LTC,
"Linear timecode from an audio device", "ltc"},
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
{GST_TIME_CODE_STAMPER_SOURCE_RTC,
"Timecode from real time clock", "rtc"},
{0, NULL, NULL},
};
if (!type) {
type = g_enum_register_static ("GstTimeCodeStamperSource", values);
}
return type;
}
GType
gst_timecodestamper_set_get_type (void)
{
static GType type = 0;
static const GEnumValue values[] = {
{GST_TIME_CODE_STAMPER_SET_NEVER,
"Never set timecodes", "never"},
{GST_TIME_CODE_STAMPER_SET_KEEP,
"Keep upstream timecodes and only set if no upstream timecode", "keep"},
{GST_TIME_CODE_STAMPER_SET_ALWAYS,
"Always set timecode and remove upstream timecode", "always"},
{0, NULL, NULL},
};
if (!type) {
type = g_enum_register_static ("GstTimeCodeStamperSet", values);
}
return type;
}
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
static void
gst_timecodestamper_class_init (GstTimeCodeStamperClass * klass)
{
GObjectClass *gobject_class = (GObjectClass *) klass;
GstElementClass *element_class = (GstElementClass *) klass;
GstBaseTransformClass *trans_class = (GstBaseTransformClass *) klass;
GST_DEBUG_CATEGORY_INIT (timecodestamper_debug, "timecodestamper", 0,
"timecodestamper");
gst_element_class_set_static_metadata (element_class, "Timecode stamper",
"Filter/Video", "Attaches a timecode meta into each video frame",
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
"Vivia Nikolaidou <vivia@toolsonair.com>");
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
gobject_class->set_property = gst_timecodestamper_set_property;
gobject_class->get_property = gst_timecodestamper_get_property;
gobject_class->dispose = gst_timecodestamper_dispose;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
g_object_class_install_property (gobject_class, PROP_SOURCE,
g_param_spec_enum ("source", "Timecode Source",
"Choose from what source the timecode should be taken",
GST_TYPE_TIME_CODE_STAMPER_SOURCE,
DEFAULT_SOURCE, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
g_object_class_install_property (gobject_class, PROP_SET,
g_param_spec_enum ("set", "Timecode Set",
"Choose whether timecodes should be overridden or not",
GST_TYPE_TIME_CODE_STAMPER_SET,
DEFAULT_SET, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
g_object_class_install_property (gobject_class, PROP_AUTO_RESYNC,
g_param_spec_boolean ("auto-resync",
"Auto Resync",
"If true resync last known timecode from upstream, otherwise only "
"count up from the last known one",
DEFAULT_AUTO_RESYNC, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
g_object_class_install_property (gobject_class, PROP_TIMEOUT,
g_param_spec_uint64 ("timeout",
"Timeout",
"Time out upstream timecode if no new timecode was detected after this time",
0, G_MAXUINT64, DEFAULT_TIMEOUT,
G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
g_object_class_install_property (gobject_class, PROP_DROP_FRAME,
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
g_param_spec_boolean ("drop-frame", "Drop Frame",
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
"Use drop-frame timecodes for 29.97 and 59.94 FPS",
DEFAULT_DROP_FRAME, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
g_object_class_install_property (gobject_class, PROP_POST_MESSAGES,
g_param_spec_boolean ("post-messages", "Post element message",
"Post element message containing the current timecode",
DEFAULT_POST_MESSAGES, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
g_object_class_install_property (gobject_class, PROP_SET_INTERNAL_TIMECODE,
g_param_spec_boxed ("set-internal-timecode",
"Set Internal Timecode",
"If set, take this timecode as the internal timecode for the first "
"frame and increment from it. Only the values itself and daily jam are taken, "
"flags and frame rate are always determined by timecodestamper "
timecodestamper: Update set-tc property documentation with latest version of reality
2019-10-23 11:28:24 +00:00
"itself. If unset, the internal timecode will start at 0 with the daily jam "
"being the current real-time clock time",
timecodestamper: First timecode property Add an new property to start from a given timecode instead of zero. https://bugzilla.gnome.org/show_bug.cgi?id=778703
2017-02-22 21:26:57 +00:00
GST_TYPE_VIDEO_TIME_CODE,
G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_object_class_install_property (gobject_class, PROP_LTC_DAILY_JAM,
g_param_spec_boxed ("ltc-daily-jam",
"LTC Daily jam",
"The daily jam of the LTC timecode",
G_TYPE_DATE_TIME, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
g_object_class_install_property (gobject_class, PROP_LTC_AUTO_RESYNC,
g_param_spec_boolean ("ltc-auto-resync",
"LTC Auto Resync",
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
"If true the LTC timecode will be automatically resynced if it drifts, "
"otherwise it will only be counted up from the last known one",
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
DEFAULT_LTC_AUTO_RESYNC, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
g_object_class_install_property (gobject_class, PROP_LTC_EXTRA_LATENCY,
g_param_spec_uint64 ("ltc-extra-latency", "LTC Extra Latency",
"Extra latency to introduce for waiting for LTC timecodes",
0, G_MAXUINT64, DEFAULT_LTC_EXTRA_LATENCY,
G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
g_object_class_install_property (gobject_class, PROP_LTC_TIMEOUT,
g_param_spec_uint64 ("ltc-timeout", "LTC Timeout",
"Time out LTC timecode if no new timecode was detected after this time",
0, G_MAXUINT64, DEFAULT_LTC_TIMEOUT,
G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
g_object_class_install_property (gobject_class, PROP_RTC_MAX_DRIFT,
g_param_spec_uint64 ("rtc-max-drift",
"RTC Maximum Offset",
"Maximum number of nanoseconds the RTC clock is allowed to drift from "
"the video before it is resynced",
0, G_MAXUINT64, DEFAULT_RTC_MAX_DRIFT,
G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
g_object_class_install_property (gobject_class, PROP_RTC_AUTO_RESYNC,
g_param_spec_boolean ("rtc-auto-resync",
"RTC Auto Resync",
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
"If true the RTC timecode will be automatically resynced if it drifts, "
"otherwise it will only be counted up from the last known one",
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
DEFAULT_RTC_AUTO_RESYNC, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
g_object_class_install_property (gobject_class, PROP_TIMECODE_OFFSET,
g_param_spec_int ("timecode-offset",
"Timecode Offset",
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
"Add this offset in frames to internal, LTC or RTC timecode, "
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
"useful if there is an offset between the timecode source and video",
G_MININT, G_MAXINT, 0, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
gst_element_class_add_pad_template (element_class,
gst_static_pad_template_get (&gst_timecodestamper_sink_template));
gst_element_class_add_pad_template (element_class,
gst_static_pad_template_get (&gst_timecodestamper_src_template));
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_element_class_add_pad_template (element_class,
gst_static_pad_template_get (&gst_timecodestamper_ltc_template));
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
element_class->request_new_pad =
GST_DEBUG_FUNCPTR (gst_timecodestamper_request_new_pad);
element_class->release_pad =
GST_DEBUG_FUNCPTR (gst_timecodestamper_release_pad);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
trans_class->sink_event = GST_DEBUG_FUNCPTR (gst_timecodestamper_sink_event);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
#if HAVE_LTC
trans_class->query = GST_DEBUG_FUNCPTR (gst_timecodestamper_query);
#endif
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
trans_class->stop = GST_DEBUG_FUNCPTR (gst_timecodestamper_stop);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
trans_class->start = GST_DEBUG_FUNCPTR (gst_timecodestamper_start);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
trans_class->transform_ip =
GST_DEBUG_FUNCPTR (gst_timecodestamper_transform_ip);
}
static void
gst_timecodestamper_init (GstTimeCodeStamper * timecodestamper)
{
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
timecodestamper->ltcpad = NULL;
timecodestamper->tc_source = GST_TIME_CODE_STAMPER_SOURCE_INTERNAL;
timecodestamper->tc_set = GST_TIME_CODE_STAMPER_SET_KEEP;
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
timecodestamper->tc_auto_resync = DEFAULT_AUTO_RESYNC;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->tc_timeout = DEFAULT_TIMEOUT;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
timecodestamper->drop_frame = DEFAULT_DROP_FRAME;
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
timecodestamper->post_messages = DEFAULT_POST_MESSAGES;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
timecodestamper->set_internal_tc = NULL;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper->ltc_daily_jam = DEFAULT_LTC_DAILY_JAM;
timecodestamper->ltc_auto_resync = DEFAULT_LTC_AUTO_RESYNC;
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
timecodestamper->ltc_extra_latency = DEFAULT_LTC_EXTRA_LATENCY;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->ltc_timeout = DEFAULT_LTC_TIMEOUT;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
timecodestamper->rtc_max_drift = DEFAULT_RTC_MAX_DRIFT;
timecodestamper->rtc_auto_resync = DEFAULT_RTC_AUTO_RESYNC;
timecodestamper->timecode_offset = 0;
timecodestamper->internal_tc = NULL;
timecodestamper->last_tc = NULL;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->last_tc_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
timecodestamper->rtc_tc = NULL;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
g_mutex_init (&timecodestamper->mutex);
g_cond_init (&timecodestamper->ltc_cond_video);
g_cond_init (&timecodestamper->ltc_cond_audio);
gst_segment_init (&timecodestamper->ltc_segment, GST_FORMAT_UNDEFINED);
timecodestamper->ltc_first_running_time = GST_CLOCK_TIME_NONE;
timecodestamper->ltc_current_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
g_queue_init (&timecodestamper->ltc_current_tcs);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper->ltc_internal_tc = NULL;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->ltc_internal_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper->ltc_dec = NULL;
timecodestamper->ltc_total = 0;
timecodestamper->ltc_eos = TRUE;
timecodestamper->ltc_flushing = TRUE;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->audio_live = FALSE;
timecodestamper->audio_latency = GST_CLOCK_TIME_NONE;
timecodestamper->video_live = FALSE;
timecodestamper->video_latency = GST_CLOCK_TIME_NONE;
timecodestamper->latency = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper->video_activatemode_default =
GST_PAD_ACTIVATEMODEFUNC (GST_BASE_TRANSFORM_SINK_PAD (timecodestamper));
GST_PAD_ACTIVATEMODEFUNC (GST_BASE_TRANSFORM_SINK_PAD (timecodestamper)) =
gst_timecodestamper_videopad_activatemode;
gst_pad_set_iterate_internal_links_function (GST_BASE_TRANSFORM_SRC_PAD
(timecodestamper), gst_timecodestamper_src_iterate_internal_link);
#endif
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
static void
gst_timecodestamper_dispose (GObject * object)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (object);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
if (timecodestamper->ltc_daily_jam) {
g_date_time_unref (timecodestamper->ltc_daily_jam);
timecodestamper->ltc_daily_jam = NULL;
}
if (timecodestamper->internal_tc != NULL) {
gst_video_time_code_free (timecodestamper->internal_tc);
timecodestamper->internal_tc = NULL;
}
if (timecodestamper->set_internal_tc != NULL) {
gst_video_time_code_free (timecodestamper->set_internal_tc);
timecodestamper->set_internal_tc = NULL;
}
if (timecodestamper->last_tc != NULL) {
gst_video_time_code_free (timecodestamper->last_tc);
timecodestamper->last_tc = NULL;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->last_tc_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
if (timecodestamper->rtc_tc != NULL) {
gst_video_time_code_free (timecodestamper->rtc_tc);
timecodestamper->rtc_tc = NULL;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
g_cond_clear (&timecodestamper->ltc_cond_video);
g_cond_clear (&timecodestamper->ltc_cond_audio);
g_mutex_clear (&timecodestamper->mutex);
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
{
TimestampedTimecode *tc;
while ((tc = g_queue_pop_tail (&timecodestamper->ltc_current_tcs))) {
gst_video_time_code_clear (&tc->timecode);
g_free (tc);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
if (timecodestamper->ltc_internal_tc != NULL) {
gst_video_time_code_free (timecodestamper->ltc_internal_tc);
timecodestamper->ltc_internal_tc = NULL;
}
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->ltc_internal_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
if (timecodestamper->ltc_dec) {
ltc_decoder_free (timecodestamper->ltc_dec);
timecodestamper->ltc_dec = NULL;
}
if (timecodestamper->stream_align) {
gst_audio_stream_align_free (timecodestamper->stream_align);
timecodestamper->stream_align = NULL;
}
#endif
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
G_OBJECT_CLASS (gst_timecodestamper_parent_class)->dispose (object);
}
static void
gst_timecodestamper_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (object);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GST_OBJECT_LOCK (timecodestamper);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
switch (prop_id) {
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_SOURCE:
timecodestamper->tc_source = (GstTimeCodeStamperSource)
g_value_get_enum (value);
break;
case PROP_SET:
timecodestamper->tc_set = (GstTimeCodeStamperSet)
g_value_get_enum (value);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
break;
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
case PROP_AUTO_RESYNC:
timecodestamper->tc_auto_resync = g_value_get_boolean (value);
break;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
case PROP_TIMEOUT:
timecodestamper->tc_timeout = g_value_get_uint64 (value);
break;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
case PROP_DROP_FRAME:
timecodestamper->drop_frame = g_value_get_boolean (value);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_timecodestamper_update_drop_frame (timecodestamper);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
break;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
case PROP_LTC_DAILY_JAM:
if (timecodestamper->ltc_daily_jam)
g_date_time_unref (timecodestamper->ltc_daily_jam);
timecodestamper->ltc_daily_jam = g_value_dup_boxed (value);
#if HAVE_LTC
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
{
GList *l;
for (l = timecodestamper->ltc_current_tcs.head; l; l = l->next) {
TimestampedTimecode *tc = l->data;
if (tc->timecode.config.latest_daily_jam) {
g_date_time_unref (tc->timecode.config.latest_daily_jam);
}
tc->timecode.config.latest_daily_jam =
g_date_time_ref (timecodestamper->ltc_daily_jam);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
}
if (timecodestamper->ltc_internal_tc) {
if (timecodestamper->ltc_internal_tc->config.latest_daily_jam) {
g_date_time_unref (timecodestamper->ltc_internal_tc->config.
latest_daily_jam);
}
timecodestamper->ltc_internal_tc->config.latest_daily_jam =
g_date_time_ref (timecodestamper->ltc_daily_jam);
}
#endif
break;
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
case PROP_POST_MESSAGES:
timecodestamper->post_messages = g_value_get_boolean (value);
break;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_SET_INTERNAL_TIMECODE:{
if (timecodestamper->set_internal_tc)
gst_video_time_code_free (timecodestamper->set_internal_tc);
timecodestamper->set_internal_tc = g_value_dup_boxed (value);
/* Reset the internal timecode on the next opportunity if a new
* timecode was set here. If none was set we just continue counting
* from the previous one */
if (timecodestamper->set_internal_tc && timecodestamper->internal_tc) {
gst_video_time_code_free (timecodestamper->internal_tc);
timecodestamper->internal_tc = NULL;
}
break;
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
case PROP_LTC_AUTO_RESYNC:
timecodestamper->ltc_auto_resync = g_value_get_boolean (value);
break;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
case PROP_LTC_TIMEOUT:
timecodestamper->ltc_timeout = g_value_get_uint64 (value);
break;
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
case PROP_LTC_EXTRA_LATENCY:
timecodestamper->ltc_extra_latency = g_value_get_uint64 (value);
break;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_RTC_MAX_DRIFT:
timecodestamper->rtc_max_drift = g_value_get_uint64 (value);
break;
case PROP_RTC_AUTO_RESYNC:
timecodestamper->rtc_auto_resync = g_value_get_boolean (value);
timecodestamper: First timecode property Add an new property to start from a given timecode instead of zero. https://bugzilla.gnome.org/show_bug.cgi?id=778703
2017-02-22 21:26:57 +00:00
break;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_TIMECODE_OFFSET:
timecodestamper->timecode_offset = g_value_get_int (value);
timecodestamper: First timecode from current system time Add a new flag which automatically sets this first timecode to the current system time in local time zone. https://bugzilla.gnome.org/show_bug.cgi?id=778703
2017-02-22 21:30:17 +00:00
break;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GST_OBJECT_UNLOCK (timecodestamper);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
static void
gst_timecodestamper_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (object);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GST_OBJECT_LOCK (timecodestamper);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
switch (prop_id) {
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_SOURCE:
g_value_set_enum (value, timecodestamper->tc_source);
break;
case PROP_SET:
g_value_set_enum (value, timecodestamper->tc_set);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
break;
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
case PROP_AUTO_RESYNC:
g_value_set_boolean (value, timecodestamper->tc_auto_resync);
break;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
case PROP_TIMEOUT:
g_value_set_uint64 (value, timecodestamper->tc_timeout);
break;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
case PROP_DROP_FRAME:
g_value_set_boolean (value, timecodestamper->drop_frame);
break;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
case PROP_LTC_DAILY_JAM:
g_value_set_boxed (value, timecodestamper->ltc_daily_jam);
break;
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
case PROP_POST_MESSAGES:
g_value_set_boolean (value, timecodestamper->post_messages);
break;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_SET_INTERNAL_TIMECODE:
g_value_set_boxed (value, timecodestamper->set_internal_tc);
break;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
case PROP_LTC_AUTO_RESYNC:
g_value_set_boolean (value, timecodestamper->ltc_auto_resync);
break;
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
case PROP_LTC_TIMEOUT:
g_value_set_uint64 (value, timecodestamper->ltc_timeout);
break;
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
case PROP_LTC_EXTRA_LATENCY:
g_value_set_uint64 (value, timecodestamper->ltc_extra_latency);
break;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_RTC_MAX_DRIFT:
g_value_set_uint64 (value, timecodestamper->rtc_max_drift);
timecodestamper: First timecode property Add an new property to start from a given timecode instead of zero. https://bugzilla.gnome.org/show_bug.cgi?id=778703
2017-02-22 21:26:57 +00:00
break;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case PROP_RTC_AUTO_RESYNC:
g_value_set_boolean (value, timecodestamper->rtc_auto_resync);
break;
case PROP_TIMECODE_OFFSET:
g_value_set_int (value, timecodestamper->timecode_offset);
timecodestamper: First timecode from current system time Add a new flag which automatically sets this first timecode to the current system time in local time zone. https://bugzilla.gnome.org/show_bug.cgi?id=778703
2017-02-22 21:30:17 +00:00
break;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GST_OBJECT_UNLOCK (timecodestamper);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
static gboolean
gst_timecodestamper_stop (GstBaseTransform * trans)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (trans);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_flushing = TRUE;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->video_current_running_time = GST_CLOCK_TIME_NONE;
if (timecodestamper->video_clock_id)
gst_clock_id_unschedule (timecodestamper->video_clock_id);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper->ltc_flushing = TRUE;
g_cond_signal (&timecodestamper->ltc_cond_video);
g_cond_signal (&timecodestamper->ltc_cond_audio);
g_mutex_unlock (&timecodestamper->mutex);
#endif
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
gst_video_info_init (&timecodestamper->vinfo);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
if (timecodestamper->internal_tc != NULL) {
gst_video_time_code_free (timecodestamper->internal_tc);
timecodestamper->internal_tc = NULL;
}
if (timecodestamper->rtc_tc != NULL) {
gst_video_time_code_free (timecodestamper->rtc_tc);
timecodestamper->rtc_tc = NULL;
}
if (timecodestamper->last_tc != NULL) {
gst_video_time_code_free (timecodestamper->last_tc);
timecodestamper->last_tc = NULL;
}
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->last_tc_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
g_mutex_lock (&timecodestamper->mutex);
gst_audio_info_init (&timecodestamper->ainfo);
gst_segment_init (&timecodestamper->ltc_segment, GST_FORMAT_UNDEFINED);
timecodestamper->ltc_first_running_time = GST_CLOCK_TIME_NONE;
timecodestamper->ltc_current_running_time = GST_CLOCK_TIME_NONE;
if (timecodestamper->ltc_internal_tc != NULL) {
gst_video_time_code_free (timecodestamper->ltc_internal_tc);
timecodestamper->ltc_internal_tc = NULL;
}
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->ltc_internal_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
{
TimestampedTimecode *tc;
while ((tc = g_queue_pop_tail (&timecodestamper->ltc_current_tcs))) {
gst_video_time_code_clear (&tc->timecode);
g_free (tc);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
if (timecodestamper->ltc_dec) {
ltc_decoder_free (timecodestamper->ltc_dec);
timecodestamper->ltc_dec = NULL;
}
if (timecodestamper->stream_align) {
gst_audio_stream_align_free (timecodestamper->stream_align);
timecodestamper->stream_align = NULL;
}
timecodestamper->ltc_total = 0;
g_mutex_unlock (&timecodestamper->mutex);
#endif
return TRUE;
}
static gboolean
gst_timecodestamper_start (GstBaseTransform * trans)
{
#if HAVE_LTC
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (trans);
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_flushing = FALSE;
timecodestamper->video_eos = FALSE;
g_mutex_unlock (&timecodestamper->mutex);
#endif
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
return TRUE;
}
/* Must be called with object lock */
static void
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_timecodestamper_update_drop_frame (GstTimeCodeStamper * timecodestamper)
{
if (timecodestamper->drop_frame && timecodestamper->vinfo.fps_d == 1001 &&
(timecodestamper->vinfo.fps_n == 30000 ||
timecodestamper->vinfo.fps_n == 60000)) {
if (timecodestamper->internal_tc)
timecodestamper->internal_tc->config.flags |=
GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
if (timecodestamper->rtc_tc)
timecodestamper->rtc_tc->config.flags |=
GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
{
GList *l;
for (l = timecodestamper->ltc_current_tcs.head; l; l = l->next) {
TimestampedTimecode *tc = l->data;
tc->timecode.config.flags |= GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
}
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
if (timecodestamper->ltc_internal_tc)
timecodestamper->ltc_internal_tc->config.flags |=
GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
#endif
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
} else {
if (timecodestamper->internal_tc)
timecodestamper->internal_tc->config.flags &=
~GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
if (timecodestamper->rtc_tc)
timecodestamper->rtc_tc->config.flags &=
~GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
{
GList *l;
for (l = timecodestamper->ltc_current_tcs.head; l; l = l->next) {
TimestampedTimecode *tc = l->data;
tc->timecode.config.flags &= ~GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
}
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
if (timecodestamper->ltc_internal_tc)
timecodestamper->ltc_internal_tc->config.flags &=
~GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
#endif
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
}
}
static void
gst_timecodestamper_update_timecode_framerate (GstTimeCodeStamper *
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
timecodestamper, const GstVideoInfo * vinfo, GstVideoTimeCode * timecode,
gboolean is_ltc)
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
{
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
guint64 nframes;
GstClockTime time;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
GDateTime *jam = NULL;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GstVideoTimeCodeFlags tc_flags = 0;
if (!timecode)
return;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
if (timecodestamper->vinfo.interlace_mode !=
GST_VIDEO_INTERLACE_MODE_PROGRESSIVE)
tc_flags |= GST_VIDEO_TIME_CODE_FLAGS_INTERLACED;
if (timecodestamper->drop_frame && timecodestamper->vinfo.fps_d == 1001 &&
(timecodestamper->vinfo.fps_n == 30000 ||
timecodestamper->vinfo.fps_n == 60000))
tc_flags |= GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
/* If this is an LTC timecode and we have no framerate yet in there then
* just do nothing. We're going to set the framerate at a later time */
if (timecode->config.fps_d != 0 || !is_ltc) {
nframes = gst_video_time_code_frames_since_daily_jam (timecode);
time =
gst_util_uint64_scale (nframes,
GST_SECOND * timecodestamper->vinfo.fps_d,
timecodestamper->vinfo.fps_n);
jam =
timecode->config.latest_daily_jam ? g_date_time_ref (timecode->
config.latest_daily_jam) : NULL;
gst_video_time_code_clear (timecode);
gst_video_time_code_init (timecode, timecodestamper->vinfo.fps_n,
timecodestamper->vinfo.fps_d, jam, tc_flags, 0, 0, 0, 0, 0);
if (jam)
g_date_time_unref (jam);
nframes =
gst_util_uint64_scale (time, vinfo->fps_n, GST_SECOND * vinfo->fps_d);
gst_video_time_code_add_frames (timecode, nframes);
}
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
}
/* Must be called with object lock */
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
static gboolean
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_timecodestamper_update_framerate (GstTimeCodeStamper * timecodestamper,
const GstVideoInfo * vinfo)
{
/* Nothing changed */
if (vinfo->fps_n == timecodestamper->vinfo.fps_n &&
vinfo->fps_d == timecodestamper->vinfo.fps_d)
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
return FALSE;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_timecodestamper_update_timecode_framerate (timecodestamper, vinfo,
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
timecodestamper->internal_tc, FALSE);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_timecodestamper_update_timecode_framerate (timecodestamper, vinfo,
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
timecodestamper->last_tc, FALSE);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_timecodestamper_update_timecode_framerate (timecodestamper, vinfo,
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
timecodestamper->rtc_tc, FALSE);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
{
GList *l;
for (l = timecodestamper->ltc_current_tcs.head; l; l = l->next) {
TimestampedTimecode *tc = l->data;
gst_timecodestamper_update_timecode_framerate (timecodestamper, vinfo,
&tc->timecode, TRUE);
}
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
gst_timecodestamper_update_timecode_framerate (timecodestamper, vinfo,
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
timecodestamper->ltc_internal_tc, FALSE);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#endif
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
return TRUE;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
static gboolean
gst_timecodestamper_sink_event (GstBaseTransform * trans, GstEvent * event)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (trans);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
gboolean ret = FALSE;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
GST_DEBUG_OBJECT (trans, "received event %" GST_PTR_FORMAT, event);
switch (GST_EVENT_TYPE (event)) {
case GST_EVENT_SEGMENT:
{
GstSegment segment;
gst_event_copy_segment (event, &segment);
if (segment.format != GST_FORMAT_TIME) {
GST_ERROR_OBJECT (timecodestamper, "Invalid segment format");
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_event_unref (event);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
return FALSE;
}
break;
}
case GST_EVENT_CAPS:
{
GstCaps *caps;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GstVideoInfo info;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
gboolean latency_changed;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
Revert "WIP: Revert "Revert "timecodestamper: LTC from audio""" This reverts commit e0be05dc7059cc97dceb70a48ca9cad4ee2edce6.
2018-01-04 13:41:49 +00:00
GST_OBJECT_LOCK (timecodestamper);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
gst_event_parse_caps (event, &caps);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
if (!gst_video_info_from_caps (&info, caps)) {
GST_OBJECT_UNLOCK (timecodestamper);
gst_event_unref (event);
return FALSE;
}
if (info.fps_n == 0) {
GST_WARNING_OBJECT (timecodestamper,
"Non-constant frame rate found. Refusing to create a timecode");
Revert "WIP: Revert "Revert "timecodestamper: LTC from audio""" This reverts commit e0be05dc7059cc97dceb70a48ca9cad4ee2edce6.
2018-01-04 13:41:49 +00:00
GST_OBJECT_UNLOCK (timecodestamper);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gst_event_unref (event);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
return FALSE;
}
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
latency_changed =
gst_timecodestamper_update_framerate (timecodestamper, &info);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
timecodestamper->vinfo = info;
Revert "WIP: Revert "Revert "timecodestamper: LTC from audio""" This reverts commit e0be05dc7059cc97dceb70a48ca9cad4ee2edce6.
2018-01-04 13:41:49 +00:00
GST_OBJECT_UNLOCK (timecodestamper);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
if (latency_changed)
gst_element_post_message (GST_ELEMENT_CAST (timecodestamper),
gst_message_new_latency (GST_OBJECT_CAST (timecodestamper)));
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
break;
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
case GST_EVENT_FLUSH_START:
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_flushing = TRUE;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->video_current_running_time = GST_CLOCK_TIME_NONE;
if (timecodestamper->video_clock_id)
gst_clock_id_unschedule (timecodestamper->video_clock_id);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_cond_signal (&timecodestamper->ltc_cond_video);
g_mutex_unlock (&timecodestamper->mutex);
break;
case GST_EVENT_FLUSH_STOP:
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_flushing = FALSE;
timecodestamper->video_eos = FALSE;
g_mutex_unlock (&timecodestamper->mutex);
break;
case GST_EVENT_EOS:
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_eos = TRUE;
g_cond_signal (&timecodestamper->ltc_cond_audio);
g_mutex_unlock (&timecodestamper->mutex);
break;
#endif
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
default:
break;
}
ret =
GST_BASE_TRANSFORM_CLASS (gst_timecodestamper_parent_class)->sink_event
(trans, event);
return ret;
}
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
#if HAVE_LTC
static gboolean
gst_timecodestamper_query (GstBaseTransform * trans,
GstPadDirection direction, GstQuery * query)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (trans);
if (direction == GST_PAD_SINK)
return
GST_BASE_TRANSFORM_CLASS (gst_timecodestamper_parent_class)->query
(trans, direction, query);
switch (GST_QUERY_TYPE (query)) {
case GST_QUERY_LATENCY:{
gboolean res;
gboolean live;
GstClockTime min_latency, max_latency;
GstClockTime latency;
res =
gst_pad_query_default (GST_BASE_TRANSFORM_SRC_PAD (trans),
GST_OBJECT_CAST (trans), query);
g_mutex_lock (&timecodestamper->mutex);
if (res && timecodestamper->vinfo.fps_n && timecodestamper->vinfo.fps_d) {
gst_query_parse_latency (query, &live, &min_latency, &max_latency);
timecodestamper: Only allow requesting LTC audio pad in NULL/READY states And don't introduce any latency at all if not LTC audio pad was requested.
2020-01-09 13:27:28 +00:00
if (live && timecodestamper->ltcpad) {
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
/* Introduce additional LTC for waiting for LTC timecodes. The
* LTC library introduces some as well as the encoding of the LTC
* signal. */
latency = timecodestamper->ltc_extra_latency *
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
gst_util_uint64_scale_int_ceil (GST_SECOND,
timecodestamper->vinfo.fps_d, timecodestamper->vinfo.fps_n);
min_latency += latency;
if (max_latency != GST_CLOCK_TIME_NONE)
max_latency += latency;
timecodestamper->latency = min_latency;
timecodestamper: In live mode wait correctly for the latency to pass And also introduce 6 instead of 2 frames of latency compared to the LTC audio input as that seems to be an upper bound for how much the LTC library is lagging behind.
2020-01-09 13:23:20 +00:00
gst_query_set_latency (query, live, min_latency, max_latency);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
} else {
timecodestamper->latency = 0;
}
} else if (res) {
GST_ERROR_OBJECT (timecodestamper,
"Need a known, non-variable framerate to answer LATENCY query");
res = FALSE;
timecodestamper->latency = GST_CLOCK_TIME_NONE;
}
g_mutex_unlock (&timecodestamper->mutex);
return res;
}
default:
return
GST_BASE_TRANSFORM_CLASS (gst_timecodestamper_parent_class)->query
(trans, direction, query);
}
}
#endif
timecodestamper: Remove all existing timecode metas if requested to override existing
2016-11-24 12:41:59 +00:00
static gboolean
remove_timecode_meta (GstBuffer * buffer, GstMeta ** meta, gpointer user_data)
{
if (meta && *meta && (*meta)->info->api == GST_VIDEO_TIME_CODE_META_API_TYPE) {
*meta = NULL;
}
return TRUE;
}
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
#if HAVE_LTC
static void
gst_timecodestamper_update_latency (GstTimeCodeStamper * timecodestamper,
GstPad * pad, gboolean * live, GstClockTime * latency)
{
GstQuery *query;
query = gst_query_new_latency ();
if (!gst_pad_peer_query (pad, query)) {
GST_WARNING_OBJECT (pad, "Failed to query latency");
gst_pad_mark_reconfigure (pad);
return;
}
g_mutex_lock (&timecodestamper->mutex);
gst_query_parse_latency (query, live, latency, NULL);
/* If we're not live, consider a latency of 0 */
if (!*live)
*latency = 0;
timecodestamper: Add some more debug output
2020-01-13 10:54:35 +00:00
GST_DEBUG_OBJECT (pad,
"Queried latency: live %d, min latency %" GST_TIME_FORMAT, *live,
GST_TIME_ARGS (*latency));
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
g_mutex_unlock (&timecodestamper->mutex);
}
#endif
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
static GstFlowReturn
gst_timecodestamper_transform_ip (GstBaseTransform * vfilter,
GstBuffer * buffer)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (vfilter);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GstClockTime running_time, base_time, clock_time;
GstClock *clock;
GstClockTime clock_time_now;
GDateTime *dt_now, *dt_frame;
GstVideoTimeCode *tc = NULL;
gboolean free_tc = FALSE;
Revert "WIP: Revert "Revert "timecodestamper: LTC from audio""" This reverts commit e0be05dc7059cc97dceb70a48ca9cad4ee2edce6.
2018-01-04 13:41:49 +00:00
GstVideoTimeCodeMeta *tc_meta;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GstFlowReturn flow_ret = GST_FLOW_OK;
GstVideoTimeCodeFlags tc_flags = 0;
if (timecodestamper->vinfo.fps_n == 0 || timecodestamper->vinfo.fps_d == 0
|| !GST_BUFFER_PTS_IS_VALID (buffer)) {
gst_buffer_unref (buffer);
return GST_FLOW_NOT_NEGOTIATED;
}
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
#if HAVE_LTC
if (timecodestamper->video_latency == -1
|| gst_pad_check_reconfigure (GST_BASE_TRANSFORM_SINK_PAD (vfilter))) {
gst_timecodestamper_update_latency (timecodestamper,
GST_BASE_TRANSFORM_SINK_PAD (vfilter), &timecodestamper->video_live,
&timecodestamper->video_latency);
}
#endif
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
/* Collect all the current times */
base_time = gst_element_get_base_time (GST_ELEMENT (timecodestamper));
clock = gst_element_get_clock (GST_ELEMENT (timecodestamper));
if (clock) {
clock_time_now = gst_clock_get_time (clock);
gst_object_unref (clock);
} else {
clock_time_now = GST_CLOCK_TIME_NONE;
}
dt_now = g_date_time_new_now_local ();
running_time =
gst_segment_to_running_time (&vfilter->segment, GST_FORMAT_TIME,
GST_BUFFER_PTS (buffer));
if (clock_time_now != GST_CLOCK_TIME_NONE) {
gdouble seconds_diff;
clock_time = running_time + base_time;
if (clock_time_now > clock_time) {
seconds_diff = (clock_time_now - clock_time) / -1000000000.0;
} else {
seconds_diff = (clock_time - clock_time_now) / 1000000000.0;
}
dt_frame = g_date_time_add_seconds (dt_now, seconds_diff);
} else {
/* If we have no clock we can't really know the time of the frame */
dt_frame = g_date_time_ref (dt_now);
}
GST_DEBUG_OBJECT (timecodestamper,
"Handling video frame with running time %" GST_TIME_FORMAT,
GST_TIME_ARGS (running_time));
Revert "WIP: Revert "Revert "timecodestamper: LTC from audio""" This reverts commit e0be05dc7059cc97dceb70a48ca9cad4ee2edce6.
2018-01-04 13:41:49 +00:00
tc_meta = gst_buffer_get_video_time_code_meta (buffer);
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
/* Update all our internal timecodes as needed */
GST_OBJECT_LOCK (timecodestamper);
if (timecodestamper->vinfo.interlace_mode !=
GST_VIDEO_INTERLACE_MODE_PROGRESSIVE)
tc_flags |= GST_VIDEO_TIME_CODE_FLAGS_INTERLACED;
if (timecodestamper->drop_frame && timecodestamper->vinfo.fps_d == 1001 &&
(timecodestamper->vinfo.fps_n == 30000 ||
timecodestamper->vinfo.fps_n == 60000))
tc_flags |= GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
/* If we don't have an internal timecode yet then either a new one was just
* set via the property or we just started. Initialize it here, otherwise
* increment it by one */
if (!timecodestamper->internal_tc) {
gchar *tc_str;
if (timecodestamper->set_internal_tc)
timecodestamper->internal_tc =
gst_video_time_code_new (timecodestamper->vinfo.fps_n,
timecodestamper->vinfo.fps_d,
timecodestamper->set_internal_tc->config.latest_daily_jam,
tc_flags,
timecodestamper->set_internal_tc->hours,
timecodestamper->set_internal_tc->minutes,
timecodestamper->set_internal_tc->seconds,
timecodestamper->set_internal_tc->frames,
timecodestamper->set_internal_tc->field_count);
else
timecodestamper->internal_tc =
gst_video_time_code_new (timecodestamper->vinfo.fps_n,
timecodestamper->vinfo.fps_d, dt_frame, tc_flags, 0, 0, 0, 0, 0);
tc_str = gst_video_time_code_to_string (timecodestamper->internal_tc);
GST_DEBUG_OBJECT (timecodestamper, "Initialized internal timecode to %s",
tc_str);
g_free (tc_str);
} else {
gchar *tc_str;
gst_video_time_code_increment_frame (timecodestamper->internal_tc);
tc_str = gst_video_time_code_to_string (timecodestamper->internal_tc);
GST_DEBUG_OBJECT (timecodestamper, "Incremented internal timecode to %s",
tc_str);
g_free (tc_str);
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
timecodestamper: Remove all existing timecode metas if requested to override existing
2016-11-24 12:41:59 +00:00
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
/* If we have a new timecode on the incoming frame, update our last known
* timecode or otherwise increment it by one */
timecodestamper: Add new auto-resync boolean property This allows selecting whether we continue updating our last known upstream timecode whenever a new one arrives or instead only keep the last known one and from there on count up.
2019-10-23 12:03:09 +00:00
if (tc_meta && (!timecodestamper->last_tc || timecodestamper->tc_auto_resync)) {
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
gchar *tc_str;
if (timecodestamper->last_tc)
gst_video_time_code_free (timecodestamper->last_tc);
timecodestamper->last_tc = gst_video_time_code_copy (&tc_meta->tc);
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->last_tc_running_time = running_time;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
tc_str = gst_video_time_code_to_string (timecodestamper->last_tc);
GST_DEBUG_OBJECT (timecodestamper, "Updated upstream timecode to %s",
tc_str);
g_free (tc_str);
} else {
if (timecodestamper->last_tc) {
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
if (timecodestamper->tc_auto_resync
&& timecodestamper->tc_timeout != GST_CLOCK_TIME_NONE
&& (running_time + timecodestamper->tc_timeout <
timecodestamper->last_tc_running_time
|| running_time >=
timecodestamper->last_tc_running_time +
timecodestamper->tc_timeout)) {
if (timecodestamper->last_tc)
gst_video_time_code_free (timecodestamper->last_tc);
timecodestamper->last_tc = NULL;
timecodestamper->last_tc_running_time = GST_CLOCK_TIME_NONE;
GST_DEBUG_OBJECT (timecodestamper, "Upstream timecode timed out");
} else {
gchar *tc_str;
timecodestamper: Don't initialize upstream timecode with zero if none was seen Instead keep it unset and use the internal timecode wherever needed as fallback.
2019-10-23 11:29:06 +00:00
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
gst_video_time_code_increment_frame (timecodestamper->last_tc);
tc_str = gst_video_time_code_to_string (timecodestamper->last_tc);
GST_DEBUG_OBJECT (timecodestamper,
"Incremented upstream timecode to %s", tc_str);
g_free (tc_str);
}
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
} else {
timecodestamper: Don't initialize upstream timecode with zero if none was seen Instead keep it unset and use the internal timecode wherever needed as fallback.
2019-10-23 11:29:06 +00:00
GST_DEBUG_OBJECT (timecodestamper, "Never saw an upstream timecode");
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
}
}
/* Update RTC-based timecode */
{
GstVideoTimeCode rtc_timecode_now;
gchar *tc_str, *dt_str;
/* Create timecode for the current frame time */
memset (&rtc_timecode_now, 0, sizeof (rtc_timecode_now));
gst_video_time_code_init_from_date_time_full (&rtc_timecode_now,
timecodestamper->vinfo.fps_n, timecodestamper->vinfo.fps_d, dt_frame,
tc_flags, 0);
tc_str = gst_video_time_code_to_string (&rtc_timecode_now);
dt_str = g_date_time_format (dt_frame, "%F %R %z");
GST_DEBUG_OBJECT (timecodestamper,
"Created RTC timecode %s for %s (%06u us)", tc_str, dt_str,
g_date_time_get_microsecond (dt_frame));
g_free (dt_str);
g_free (tc_str);
/* If we don't have an RTC timecode yet, directly initialize with this one */
if (!timecodestamper->rtc_tc) {
timecodestamper->rtc_tc = gst_video_time_code_copy (&rtc_timecode_now);
tc_str = gst_video_time_code_to_string (timecodestamper->rtc_tc);
GST_DEBUG_OBJECT (timecodestamper, "Initialized RTC timecode to %s",
tc_str);
g_free (tc_str);
} else {
GstClockTime rtc_now_time, rtc_tc_time;
GstClockTime rtc_diff;
/* Increment the old RTC timecode to this frame */
gst_video_time_code_increment_frame (timecodestamper->rtc_tc);
/* Otherwise check if we drifted too much and need to resync */
rtc_tc_time =
gst_video_time_code_nsec_since_daily_jam (timecodestamper->rtc_tc);
rtc_now_time =
gst_video_time_code_nsec_since_daily_jam (&rtc_timecode_now);
if (rtc_tc_time > rtc_now_time)
rtc_diff = rtc_tc_time - rtc_now_time;
else
rtc_diff = rtc_now_time - rtc_tc_time;
if (timecodestamper->rtc_auto_resync
&& timecodestamper->rtc_max_drift != GST_CLOCK_TIME_NONE
&& rtc_diff > timecodestamper->rtc_max_drift) {
gst_video_time_code_free (timecodestamper->rtc_tc);
timecodestamper->rtc_tc = gst_video_time_code_copy (&rtc_timecode_now);
tc_str = gst_video_time_code_to_string (timecodestamper->rtc_tc);
GST_DEBUG_OBJECT (timecodestamper,
"Updated RTC timecode to %s (%s%" GST_TIME_FORMAT " drift)", tc_str,
(rtc_tc_time > rtc_now_time ? "-" : "+"), GST_TIME_ARGS (rtc_diff));
g_free (tc_str);
} else {
/* Else nothing to do here, we use the current one */
tc_str = gst_video_time_code_to_string (timecodestamper->rtc_tc);
GST_DEBUG_OBJECT (timecodestamper,
"Incremented RTC timecode to %s (%s%" GST_TIME_FORMAT " drift)",
tc_str, (rtc_tc_time > rtc_now_time ? "-" : "+"),
GST_TIME_ARGS (rtc_diff));
g_free (tc_str);
}
}
gst_video_time_code_clear (&rtc_timecode_now);
}
Revert "WIP: Revert "Revert "timecodestamper: LTC from audio""" This reverts commit e0be05dc7059cc97dceb70a48ca9cad4ee2edce6.
2018-01-04 13:41:49 +00:00
GST_OBJECT_UNLOCK (timecodestamper);
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
/* Update LTC-based timecode as needed */
#if HAVE_LTC
timecodestamper: Only allow requesting LTC audio pad in NULL/READY states And don't introduce any latency at all if not LTC audio pad was requested.
2020-01-09 13:27:28 +00:00
if (timecodestamper->ltcpad) {
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
GstClockTime frame_duration;
gchar *tc_str;
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
TimestampedTimecode *ltc_tc;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
gboolean updated_internal = FALSE;
frame_duration = gst_util_uint64_scale_int_ceil (GST_SECOND,
timecodestamper->vinfo.fps_d, timecodestamper->vinfo.fps_n);
g_mutex_lock (&timecodestamper->mutex);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->video_current_running_time = running_time;
timecodestamper: In live mode wait correctly for the latency to pass And also introduce 6 instead of 2 frames of latency compared to the LTC audio input as that seems to be an upper bound for how much the LTC library is lagging behind.
2020-01-09 13:23:20 +00:00
/* Wait to compensate for the latency we introduce and to allow the LTC
* audio to provide enough audio to extract timecodes, or until the video
* pad is flushing or the LTC pad is EOS.
* In non-live mode we introduce 4 frames of latency compared to the LTC
* audio, see LATENCY query handling for details. */
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
if (timecodestamper->video_live) {
GstClock *clock =
gst_element_get_clock (GST_ELEMENT_CAST (timecodestamper));
if (clock) {
GstClockID clock_id;
GstClockTime base_time =
gst_element_get_base_time (GST_ELEMENT_CAST (timecodestamper));
timecodestamper: Fix waiting for the first video frame in case of live video input
2020-01-10 09:33:28 +00:00
GstClockTime wait_time;
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
/* If we have no latency yet then wait at least for the LTC extra
* latency. See LATENCY query handling for details. */
timecodestamper: Fix waiting for the first video frame in case of live video input
2020-01-10 09:33:28 +00:00
if (timecodestamper->latency == GST_CLOCK_TIME_NONE) {
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
wait_time =
base_time + running_time + timecodestamper->ltc_extra_latency;
timecodestamper: Fix waiting for the first video frame in case of live video input
2020-01-10 09:33:28 +00:00
} else {
wait_time = base_time + running_time + timecodestamper->latency;
}
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
GST_TRACE_OBJECT (timecodestamper,
timecodestamper: Add some more debug output
2020-01-13 10:54:35 +00:00
"Waiting for clock to reach %" GST_TIME_FORMAT
" (base time %" GST_TIME_FORMAT
" + running time %" GST_TIME_FORMAT
" + latency %" GST_TIME_FORMAT
"), now %" GST_TIME_FORMAT,
GST_TIME_ARGS (wait_time),
GST_TIME_ARGS (base_time),
GST_TIME_ARGS (running_time),
GST_TIME_ARGS (timecodestamper->latency ==
timecodestamper: Add property to set the extra latency to introduce for waiting for LTC timecodes Default to 150ms instead of 8 frames, which seems to work in the majority of cases.
2020-01-13 15:10:39 +00:00
GST_CLOCK_TIME_NONE ? timecodestamper->ltc_extra_latency :
timecodestamper->latency),
timecodestamper: Add some more debug output
2020-01-13 10:54:35 +00:00
GST_TIME_ARGS (gst_clock_get_time (clock))
);
timecodestamper: Fix waiting for the first video frame in case of live video input
2020-01-10 09:33:28 +00:00
clock_id = gst_clock_new_single_shot_id (clock, wait_time);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->video_clock_id = clock_id;
g_mutex_unlock (&timecodestamper->mutex);
gst_clock_id_wait (clock_id, NULL);
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_clock_id = NULL;
gst_clock_id_unref (clock_id);
gst_object_unref (clock);
} else {
GST_WARNING_OBJECT (timecodestamper,
"No clock in live mode, not waiting");
}
} else {
while ((timecodestamper->ltc_current_running_time == GST_CLOCK_TIME_NONE
|| timecodestamper->ltc_current_running_time <
timecodestamper: In live mode wait correctly for the latency to pass And also introduce 6 instead of 2 frames of latency compared to the LTC audio input as that seems to be an upper bound for how much the LTC library is lagging behind.
2020-01-09 13:23:20 +00:00
running_time + 8 * frame_duration)
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
&& !timecodestamper->video_flushing && !timecodestamper->ltc_eos) {
GST_TRACE_OBJECT (timecodestamper,
"Waiting for LTC audio to advance, EOS or flushing");
g_cond_wait (&timecodestamper->ltc_cond_video, &timecodestamper->mutex);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
if (timecodestamper->video_flushing) {
g_mutex_unlock (&timecodestamper->mutex);
flow_ret = GST_FLOW_FLUSHING;
goto out;
}
GST_OBJECT_LOCK (timecodestamper);
/* Take timecodes out of the queue until we're at the current video
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
* position. */
while ((ltc_tc = g_queue_pop_head (&timecodestamper->ltc_current_tcs))) {
/* First update framerate and flags according to the video stream if not
* done yet */
if (ltc_tc->timecode.config.fps_d == 0) {
gint fps_n_div =
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
((gdouble) timecodestamper->vinfo.fps_n) /
timecodestamper->vinfo.fps_d > 30 ? 2 : 1;
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
ltc_tc->timecode.config.flags = tc_flags;
ltc_tc->timecode.config.fps_n =
timecodestamper->vinfo.fps_n / fps_n_div;
ltc_tc->timecode.config.fps_d = timecodestamper->vinfo.fps_d;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
tc_str = gst_video_time_code_to_string (&ltc_tc->timecode);
GST_INFO_OBJECT (timecodestamper,
"Retrieved LTC timecode %s at %" GST_TIME_FORMAT
" (%u timecodes queued)", tc_str,
GST_TIME_ARGS (ltc_tc->running_time),
g_queue_get_length (&timecodestamper->ltc_current_tcs));
g_free (tc_str);
timecodestamper: Skip over invalid LTC timecodes immediately
2020-01-10 11:08:38 +00:00
if (!gst_video_time_code_is_valid (&ltc_tc->timecode)) {
tc_str = gst_video_time_code_to_string (&ltc_tc->timecode);
GST_INFO_OBJECT (timecodestamper, "Invalid LTC timecode %s", tc_str);
g_free (tc_str);
gst_video_time_code_clear (&ltc_tc->timecode);
g_free (ltc_tc);
ltc_tc = NULL;
continue;
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
/* A timecode frame that starts +/- half a frame to the
* video frame is considered belonging to that video frame.
*
* If it's further ahead than half a frame duration, break out of
* the loop here and reconsider on the next frame. */
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
if (ABSDIFF (running_time, ltc_tc->running_time) <= frame_duration / 2) {
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
/* If we're resyncing LTC in general, directly replace the current
* LTC timecode with the new one we read. Otherwise we'll continue
* counting based on the previous timecode we had
*/
if (timecodestamper->ltc_auto_resync) {
if (timecodestamper->ltc_internal_tc)
gst_video_time_code_free (timecodestamper->ltc_internal_tc);
timecodestamper: Skip over invalid LTC timecodes immediately
2020-01-10 11:08:38 +00:00
timecodestamper->ltc_internal_tc =
gst_video_time_code_copy (&ltc_tc->timecode);
timecodestamper->ltc_internal_running_time = ltc_tc->running_time;
updated_internal = TRUE;
GST_INFO_OBJECT (timecodestamper, "Resynced internal LTC counter");
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
/* And store it back for the next frame in case it has more or less
* the same running time */
g_queue_push_head (&timecodestamper->ltc_current_tcs,
g_steal_pointer (&ltc_tc));
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
break;
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
} else if (ltc_tc->running_time > running_time
&& ltc_tc->running_time - running_time > frame_duration / 2) {
/* Store it back for the next frame */
g_queue_push_head (&timecodestamper->ltc_current_tcs,
g_steal_pointer (&ltc_tc));
ltc_tc = NULL;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
break;
}
/* otherwise it's in the past and we need to consider the next
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
* timecode. Read a new one */
gst_video_time_code_clear (&ltc_tc->timecode);
g_free (ltc_tc);
ltc_tc = NULL;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
/* If we didn't update from LTC above, increment our internal timecode
* for this frame */
if (!updated_internal && timecodestamper->ltc_internal_tc) {
gst_video_time_code_increment_frame (timecodestamper->ltc_internal_tc);
}
if (timecodestamper->ltc_internal_tc) {
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
if (timecodestamper->ltc_auto_resync
&& timecodestamper->ltc_timeout != GST_CLOCK_TIME_NONE
&& (running_time + timecodestamper->ltc_timeout <
timecodestamper->ltc_internal_running_time
|| running_time >=
timecodestamper->ltc_internal_running_time +
timecodestamper->ltc_timeout)) {
if (timecodestamper->ltc_internal_tc)
gst_video_time_code_free (timecodestamper->ltc_internal_tc);
timecodestamper->ltc_internal_tc = NULL;
GST_DEBUG_OBJECT (timecodestamper, "LTC timecode timed out");
timecodestamper->ltc_internal_running_time = GST_CLOCK_TIME_NONE;
} else {
tc_str =
gst_video_time_code_to_string (timecodestamper->ltc_internal_tc);
GST_DEBUG_OBJECT (timecodestamper, "Updated LTC timecode to %s",
tc_str);
g_free (tc_str);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
} else {
GST_DEBUG_OBJECT (timecodestamper, "Have no LTC timecode yet");
}
GST_OBJECT_UNLOCK (timecodestamper);
g_cond_signal (&timecodestamper->ltc_cond_audio);
g_mutex_unlock (&timecodestamper->mutex);
}
#endif
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
GST_OBJECT_LOCK (timecodestamper);
switch (timecodestamper->tc_source) {
case GST_TIME_CODE_STAMPER_SOURCE_INTERNAL:
tc = timecodestamper->internal_tc;
break;
case GST_TIME_CODE_STAMPER_SOURCE_ZERO:
tc = gst_video_time_code_new (timecodestamper->vinfo.fps_n,
timecodestamper->vinfo.fps_d, NULL, tc_flags, 0, 0, 0, 0, 0);
free_tc = TRUE;
break;
case GST_TIME_CODE_STAMPER_SOURCE_LAST_KNOWN:
tc = timecodestamper->last_tc;
timecodestamper: Don't initialize upstream timecode with zero if none was seen Instead keep it unset and use the internal timecode wherever needed as fallback.
2019-10-23 11:29:06 +00:00
if (!tc)
tc = timecodestamper->internal_tc;
break;
timecodestamper: Add last-known-or-zero mode This uses the last known upstream timecode (counted up per frame), or otherwise zero if none was known. The normal last-known timestamp uses the internal timecode as fallback if no upstream timecode was ever known.
2019-10-23 11:31:16 +00:00
case GST_TIME_CODE_STAMPER_SOURCE_LAST_KNOWN_OR_ZERO:
tc = timecodestamper->last_tc;
if (!tc) {
tc = gst_video_time_code_new (timecodestamper->vinfo.fps_n,
timecodestamper->vinfo.fps_d, NULL, tc_flags, 0, 0, 0, 0, 0);
free_tc = TRUE;
}
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
break;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
case GST_TIME_CODE_STAMPER_SOURCE_LTC:
#if HAVE_LTC
timecodestamper: Use the internal LTC timecode tracker instead of the last one we retrieved Otherwise we don't interpolate between LTC timecodes but only ever put an LTC timecode on buffers once we actually received one.
2020-01-09 13:16:02 +00:00
if (timecodestamper->ltc_internal_tc)
tc = timecodestamper->ltc_internal_tc;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#endif
if (!tc) {
tc = gst_video_time_code_new (timecodestamper->vinfo.fps_n,
timecodestamper->vinfo.fps_d, NULL, tc_flags, 0, 0, 0, 0, 0);
free_tc = TRUE;
}
break;
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
case GST_TIME_CODE_STAMPER_SOURCE_RTC:
tc = timecodestamper->rtc_tc;
break;
}
switch (timecodestamper->tc_set) {
case GST_TIME_CODE_STAMPER_SET_NEVER:
break;
case GST_TIME_CODE_STAMPER_SET_KEEP:
if (!tc_meta && tc) {
gchar *tc_str;
if (timecodestamper->timecode_offset) {
if (!free_tc) {
tc = gst_video_time_code_copy (tc);
free_tc = TRUE;
}
gst_video_time_code_add_frames (tc, timecodestamper->timecode_offset);
}
tc_str = gst_video_time_code_to_string (tc);
GST_DEBUG_OBJECT (timecodestamper, "Storing timecode %s", tc_str);
g_free (tc_str);
gst_buffer_add_video_time_code_meta (buffer, tc);
}
break;
case GST_TIME_CODE_STAMPER_SET_ALWAYS:
gst_buffer_foreach_meta (buffer, remove_timecode_meta, NULL);
if (tc) {
gchar *tc_str;
if (timecodestamper->timecode_offset) {
if (!free_tc) {
tc = gst_video_time_code_copy (tc);
free_tc = TRUE;
}
gst_video_time_code_add_frames (tc, timecodestamper->timecode_offset);
}
tc_str = gst_video_time_code_to_string (tc);
GST_DEBUG_OBJECT (timecodestamper, "Storing timecode %s", tc_str);
g_free (tc_str);
gst_buffer_add_video_time_code_meta (buffer, tc);
}
break;
}
GST_OBJECT_UNLOCK (timecodestamper);
if (timecodestamper->post_messages && tc) {
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
GstClockTime stream_time, running_time, duration;
GstStructure *s;
GstMessage *msg;
running_time =
gst_segment_to_running_time (&vfilter->segment, GST_FORMAT_TIME,
GST_BUFFER_PTS (buffer));
stream_time =
gst_segment_to_stream_time (&vfilter->segment, GST_FORMAT_TIME,
GST_BUFFER_PTS (buffer));
duration =
gst_util_uint64_scale_int (GST_SECOND, timecodestamper->vinfo.fps_d,
timecodestamper->vinfo.fps_n);
s = gst_structure_new ("timecodestamper", "timestamp", G_TYPE_UINT64,
GST_BUFFER_PTS (buffer), "stream-time", G_TYPE_UINT64, stream_time,
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
"running-time", G_TYPE_UINT64, running_time, "duration",
G_TYPE_UINT64, duration, "timecode", GST_TYPE_VIDEO_TIME_CODE, tc,
NULL);
timecodestamper: Post element message with current timecode timecodestamper will post an element message which contains the current timecode it just stamped. If a timecode was already found and not replaced, it will still post it in a message. https://bugzilla.gnome.org/show_bug.cgi?id=777048
2017-01-09 15:53:38 +00:00
msg = gst_message_new_element (GST_OBJECT (timecodestamper), s);
gst_element_post_message (GST_ELEMENT (timecodestamper), msg);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
#if HAVE_LTC
out:
#endif
timecodestamper: Rewrite element API and code flow We now have a single property to select the timecode source that should be applied, and for each timecode source the timecode is updated at every frame. Then based on a set mode, the timecode is added to the frame if none exists already or all existing timecodes are removed and the timecode is added. In addition the real-time clock is considered a proper timecode source now instead of only allowing to initialize once in the beginning with it, and also instead of just taking the current time we now take the current time at the clock time of the video frame.
2019-07-01 10:42:16 +00:00
if (dt_now)
g_date_time_unref (dt_now);
if (dt_frame)
g_date_time_unref (dt_frame);
if (free_tc && tc)
gst_video_time_code_free (tc);
return flow_ret;
timecodestamper: New element to attach SMPTE timecode to buffers The timecodestamper element attaches a SMPTE timecode to each video buffer. This timecode corresponds to the current stream time. https://bugzilla.gnome.org/show_bug.cgi?id=766419
2016-05-20 15:17:52 +00:00
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
static GstPad *
gst_timecodestamper_request_new_pad (GstElement * element,
GstPadTemplate * templ, const gchar * name_templ, const GstCaps * caps)
{
#if HAVE_LTC
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (element);
GST_OBJECT_LOCK (timecodestamper);
if (timecodestamper->ltcpad) {
GST_OBJECT_UNLOCK (timecodestamper);
return NULL;
}
timecodestamper: Only allow requesting LTC audio pad in NULL/READY states And don't introduce any latency at all if not LTC audio pad was requested.
2020-01-09 13:27:28 +00:00
if (GST_STATE (timecodestamper) > GST_STATE_READY ||
GST_STATE_TARGET (timecodestamper) > GST_STATE_READY) {
GST_ERROR_OBJECT (timecodestamper,
"LTC audio pad can only be requested in NULL or READY state");
GST_OBJECT_UNLOCK (timecodestamper);
return NULL;
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper->ltcpad = gst_pad_new_from_static_template
timecodestamper: Create LTC sink pad with the correct name according to the template Should be "ltc_sink" and not just "ltc"
2019-11-19 10:18:39 +00:00
(&gst_timecodestamper_ltc_template, "ltc_sink");
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
gst_pad_set_chain_function (timecodestamper->ltcpad,
GST_DEBUG_FUNCPTR (gst_timecodestamper_ltcpad_chain));
gst_pad_set_event_function (timecodestamper->ltcpad,
GST_DEBUG_FUNCPTR (gst_timecodestamper_ltcpad_event));
gst_pad_set_query_function (timecodestamper->ltcpad,
GST_DEBUG_FUNCPTR (gst_timecodestamper_ltcpad_query));
gst_pad_set_activatemode_function (timecodestamper->ltcpad,
GST_DEBUG_FUNCPTR (gst_timecodestamper_ltcpad_activatemode));
GST_OBJECT_UNLOCK (timecodestamper);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->audio_live = FALSE;
timecodestamper->audio_latency = GST_CLOCK_TIME_NONE;
g_mutex_unlock (&timecodestamper->mutex);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
gst_element_add_pad (element, timecodestamper->ltcpad);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
gst_element_post_message (GST_ELEMENT_CAST (timecodestamper),
gst_message_new_latency (GST_OBJECT_CAST (timecodestamper)));
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
return timecodestamper->ltcpad;
#else
return NULL;
#endif
}
static void
gst_timecodestamper_release_pad (GstElement * element, GstPad * pad)
{
#if HAVE_LTC
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (element);
GST_OBJECT_LOCK (timecodestamper);
if (timecodestamper->ltcpad != pad) {
GST_OBJECT_UNLOCK (timecodestamper);
return;
}
timecodestamper->ltcpad = NULL;
if (timecodestamper->ltc_internal_tc != NULL) {
gst_video_time_code_free (timecodestamper->ltc_internal_tc);
timecodestamper->ltc_internal_tc = NULL;
}
timecodestamper: Add properties to time out cached upstream/LTC timecodes after a while By default we never time them out and simply continue couting up with each frame forever.
2019-10-23 13:47:43 +00:00
timecodestamper->ltc_internal_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
{
TimestampedTimecode *tc;
while ((tc = g_queue_pop_tail (&timecodestamper->ltc_current_tcs))) {
gst_video_time_code_clear (&tc->timecode);
g_free (tc);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
GST_OBJECT_UNLOCK (timecodestamper);
gst_pad_set_active (pad, FALSE);
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->ltc_flushing = TRUE;
timecodestamper->ltc_eos = TRUE;
g_cond_signal (&timecodestamper->ltc_cond_video);
g_cond_signal (&timecodestamper->ltc_cond_audio);
gst_audio_info_init (&timecodestamper->ainfo);
gst_segment_init (&timecodestamper->ltc_segment, GST_FORMAT_UNDEFINED);
timecodestamper->ltc_first_running_time = GST_CLOCK_TIME_NONE;
timecodestamper->ltc_current_running_time = GST_CLOCK_TIME_NONE;
if (timecodestamper->ltc_dec) {
ltc_decoder_free (timecodestamper->ltc_dec);
timecodestamper->ltc_dec = NULL;
}
if (timecodestamper->stream_align) {
gst_audio_stream_align_free (timecodestamper->stream_align);
timecodestamper->stream_align = NULL;
}
timecodestamper->ltc_total = 0;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->audio_live = FALSE;
timecodestamper->audio_latency = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_mutex_unlock (&timecodestamper->mutex);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
gst_element_post_message (GST_ELEMENT_CAST (timecodestamper),
gst_message_new_latency (GST_OBJECT_CAST (timecodestamper)));
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
gst_element_remove_pad (element, pad);
#endif
}
#if HAVE_LTC
static GstFlowReturn
gst_timecodestamper_ltcpad_chain (GstPad * pad,
GstObject * parent, GstBuffer * buffer)
{
GstFlowReturn fr = GST_FLOW_OK;
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (parent);
GstMapInfo map;
GstClockTime timestamp, running_time, duration;
guint nsamples;
gboolean discont;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
if (timecodestamper->audio_latency == -1 || gst_pad_check_reconfigure (pad)) {
gst_timecodestamper_update_latency (timecodestamper, pad,
&timecodestamper->audio_live, &timecodestamper->audio_latency);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_mutex_lock (&timecodestamper->mutex);
if (timecodestamper->ltc_flushing) {
g_mutex_unlock (&timecodestamper->mutex);
gst_buffer_unref (buffer);
return GST_FLOW_FLUSHING;
}
nsamples = gst_buffer_get_size (buffer) /
GST_AUDIO_INFO_BPF (&timecodestamper->ainfo);
if (!timecodestamper->stream_align) {
timecodestamper->stream_align =
timecodestamper: Clean up old LTC timecodes on LTC discontinuity We might have some old timecodes that are in the future now and have to drop those to make sure that our queue is correctly ordered and we don't have multiple timecodes for the same running time.
2020-01-10 09:40:54 +00:00
gst_audio_stream_align_new (timecodestamper->ainfo.rate,
500 * GST_MSECOND, 20 * GST_MSECOND);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
discont =
gst_audio_stream_align_process (timecodestamper->stream_align,
GST_BUFFER_IS_DISCONT (buffer), GST_BUFFER_PTS (buffer), nsamples,
&timestamp, &duration, NULL);
if (discont) {
timecodestamper: Clean up old LTC timecodes on LTC discontinuity We might have some old timecodes that are in the future now and have to drop those to make sure that our queue is correctly ordered and we don't have multiple timecodes for the same running time.
2020-01-10 09:40:54 +00:00
if (timecodestamper->ltc_dec) {
GST_WARNING_OBJECT (timecodestamper, "Got discont at %" GST_TIME_FORMAT,
GST_TIME_ARGS (timestamp));
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
ltc_decoder_queue_flush (timecodestamper->ltc_dec);
timecodestamper: Clean up old LTC timecodes on LTC discontinuity We might have some old timecodes that are in the future now and have to drop those to make sure that our queue is correctly ordered and we don't have multiple timecodes for the same running time.
2020-01-10 09:40:54 +00:00
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
timecodestamper->ltc_total = 0;
}
if (!timecodestamper->ltc_dec) {
gint samples_per_frame = 1920;
GST_OBJECT_LOCK (timecodestamper);
/* This is only for initialization and needs to be somewhat close to the
* real value. It will be tracked automatically afterwards */
if (timecodestamper->vinfo.fps_n) {
samples_per_frame = timecodestamper->ainfo.rate *
timecodestamper->vinfo.fps_d / timecodestamper->vinfo.fps_n;
}
GST_OBJECT_UNLOCK (timecodestamper);
timecodestamper->ltc_dec =
ltc_decoder_create (samples_per_frame, DEFAULT_LTC_QUEUE);
timecodestamper->ltc_total = 0;
}
running_time = gst_segment_to_running_time (&timecodestamper->ltc_segment,
GST_FORMAT_TIME, timestamp);
GST_DEBUG_OBJECT (timecodestamper,
"Handling LTC audio buffer at %" GST_TIME_FORMAT " - %" GST_TIME_FORMAT
" (offset %" G_GUINT64_FORMAT ")",
GST_TIME_ARGS (running_time),
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
GST_TIME_ARGS (running_time + duration),
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
(guint64) timecodestamper->ltc_total);
if (timecodestamper->ltc_total == 0) {
timecodestamper->ltc_first_running_time = running_time;
}
gst_buffer_map (buffer, &map, GST_MAP_READ);
ltc_decoder_write (timecodestamper->ltc_dec, map.data, map.size,
timecodestamper->ltc_total);
timecodestamper->ltc_total += map.size;
gst_buffer_unmap (buffer, &map);
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
/* Now read all the timecodes from the decoder that are currently available
* and store them in our own queue, which gives us more control over how
* things are working. */
{
LTCFrameExt ltc_frame;
while (ltc_decoder_read (timecodestamper->ltc_dec, &ltc_frame) == 1) {
SMPTETimecode stc;
TimestampedTimecode *ltc_tc;
GstClockTime ltc_running_time;
if (ltc_frame.off_start < 0) {
GstClockTime offset =
gst_util_uint64_scale (GST_SECOND, -ltc_frame.off_start,
timecodestamper->ainfo.rate);
if (offset > timecodestamper->ltc_first_running_time)
ltc_running_time = 0;
else
ltc_running_time = timecodestamper->ltc_first_running_time - offset;
} else {
ltc_running_time = timecodestamper->ltc_first_running_time +
gst_util_uint64_scale (GST_SECOND, ltc_frame.off_start,
timecodestamper->ainfo.rate);
}
ltc_frame_to_time (&stc, &ltc_frame.ltc, 0);
GST_INFO_OBJECT (timecodestamper,
"Got LTC timecode %02d:%02d:%02d:%02d at %" GST_TIME_FORMAT,
stc.hours, stc.mins, stc.secs, stc.frame,
GST_TIME_ARGS (ltc_running_time));
ltc_tc = g_new0 (TimestampedTimecode, 1);
ltc_tc->running_time = ltc_running_time;
/* We fill in the framerate and other metadata later */
gst_video_time_code_init (&ltc_tc->timecode,
0, 0, timecodestamper->ltc_daily_jam, 0,
stc.hours, stc.mins, stc.secs, stc.frame, 0);
timecodestamper: Clean up old LTC timecodes on LTC discontinuity We might have some old timecodes that are in the future now and have to drop those to make sure that our queue is correctly ordered and we don't have multiple timecodes for the same running time.
2020-01-10 09:40:54 +00:00
/* If we have a discontinuity it might happen that we're getting
* timecodes that are in the past relative to timecodes we already have
* in our queue. We have to get rid of all the timecodes that are in the
* future now. */
if (discont) {
TimestampedTimecode *tmp;
while ((tmp = g_queue_peek_tail (&timecodestamper->ltc_current_tcs)) &&
tmp->running_time >= ltc_running_time) {
gst_video_time_code_clear (&tmp->timecode);
g_free (tmp);
g_queue_pop_tail (&timecodestamper->ltc_current_tcs);
}
g_queue_push_tail (&timecodestamper->ltc_current_tcs,
g_steal_pointer (&ltc_tc));
} else {
g_queue_push_tail (&timecodestamper->ltc_current_tcs,
g_steal_pointer (&ltc_tc));
}
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
}
}
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
/* Notify the video streaming thread that new data is available */
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_cond_signal (&timecodestamper->ltc_cond_video);
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
/* Wait until video has caught up, if needed */
if (timecodestamper->audio_live) {
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
/* In live-mode, do no waiting as we're guaranteed to be more or less in
* sync (~latency) with the video */
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
} else {
/* If we're ahead of the video, wait until the video has caught up.
* Otherwise don't wait and drop any too old items from the ringbuffer */
while ((timecodestamper->video_current_running_time == GST_CLOCK_TIME_NONE
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
|| running_time + duration >=
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->video_current_running_time)
&& timecodestamper->ltc_dec
timecodestamper: Fix up handling/queueing of LTC timecodes Directly read them out of the decoder as soon as we passed audio and then store them in a queue that we handle internally together with their timestamps. This cleans up memory management and gives us proper control over the queue instead of guessing how the queue inside the LTC decoder actually works and when it overflows.
2020-01-09 15:41:55 +00:00
&& g_queue_get_length (&timecodestamper->ltc_current_tcs) >
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
DEFAULT_LTC_QUEUE / 2 && !timecodestamper->video_eos
&& !timecodestamper->ltc_flushing) {
GST_TRACE_OBJECT (timecodestamper,
"Waiting for video to advance, EOS or flushing");
g_cond_wait (&timecodestamper->ltc_cond_audio, &timecodestamper->mutex);
}
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
}
if (timecodestamper->ltc_flushing)
fr = GST_FLOW_FLUSHING;
else
fr = GST_FLOW_OK;
g_mutex_unlock (&timecodestamper->mutex);
gst_buffer_unref (buffer);
return fr;
}
static gboolean
gst_timecodestamper_ltcpad_event (GstPad * pad,
GstObject * parent, GstEvent * event)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (parent);
GstCaps *caps;
gboolean ret = TRUE;
switch (GST_EVENT_TYPE (event)) {
case GST_EVENT_CAPS:
gst_event_parse_caps (event, &caps);
if (!gst_audio_info_from_caps (&timecodestamper->ainfo, caps)) {
gst_event_unref (event);
return FALSE;
}
if (timecodestamper->stream_align) {
gst_audio_stream_align_set_rate (timecodestamper->stream_align,
timecodestamper->ainfo.rate);
}
break;
case GST_EVENT_SEGMENT:
gst_event_copy_segment (event, &timecodestamper->ltc_segment);
break;
case GST_EVENT_FLUSH_START:
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->ltc_flushing = TRUE;
g_cond_signal (&timecodestamper->ltc_cond_audio);
g_mutex_unlock (&timecodestamper->mutex);
break;
case GST_EVENT_FLUSH_STOP:
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->ltc_flushing = FALSE;
timecodestamper->ltc_eos = FALSE;
gst_segment_init (&timecodestamper->ltc_segment, GST_FORMAT_UNDEFINED);
g_mutex_unlock (&timecodestamper->mutex);
break;
case GST_EVENT_EOS:
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->ltc_eos = TRUE;
g_cond_signal (&timecodestamper->ltc_cond_video);
g_mutex_unlock (&timecodestamper->mutex);
break;
default:
break;
}
gst_event_unref (event);
return ret;
}
static gboolean
gst_timecodestamper_ltcpad_query (GstPad * pad,
GstObject * parent, GstQuery * query)
{
GstCaps *caps, *filter, *tcaps;
switch (GST_QUERY_TYPE (query)) {
case GST_QUERY_CAPS:
gst_query_parse_caps (query, &filter);
tcaps = gst_pad_get_pad_template_caps (pad);
if (filter)
caps = gst_caps_intersect_full (tcaps, filter,
GST_CAPS_INTERSECT_FIRST);
else
caps = gst_caps_ref (tcaps);
gst_query_set_caps_result (query, caps);
gst_caps_unref (tcaps);
gst_caps_unref (caps);
return TRUE;
default:
return gst_pad_query_default (pad, parent, query);
}
}
static gboolean
gst_timecodestamper_ltcpad_activatemode (GstPad * pad,
GstObject * parent, GstPadMode mode, gboolean active)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (parent);
if (active) {
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->ltc_flushing = FALSE;
timecodestamper->ltc_eos = FALSE;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->audio_live = FALSE;
timecodestamper->audio_latency = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_mutex_unlock (&timecodestamper->mutex);
} else {
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->ltc_flushing = TRUE;
timecodestamper->ltc_eos = TRUE;
g_cond_signal (&timecodestamper->ltc_cond_audio);
g_mutex_unlock (&timecodestamper->mutex);
}
return TRUE;
}
static gboolean
gst_timecodestamper_videopad_activatemode (GstPad * pad,
GstObject * parent, GstPadMode mode, gboolean active)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (parent);
if (active) {
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_flushing = FALSE;
timecodestamper->video_eos = FALSE;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->video_live = FALSE;
timecodestamper->video_latency = GST_CLOCK_TIME_NONE;
timecodestamper->video_current_running_time = GST_CLOCK_TIME_NONE;
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_mutex_unlock (&timecodestamper->mutex);
} else {
g_mutex_lock (&timecodestamper->mutex);
timecodestamper->video_flushing = TRUE;
timecodestamper: Refactor LTC audio waiting and properly handle live inputs If one of the inputs is live, add a latency of 2 frames to the video stream and wait on the clock for that much time to pass to allow for the LTC audio to be ahead. In case of live LTC, don't do any waiting but only ensure that we don't overflow the LTC queue. Also in non-live LTC audio mode, flush too old items from the LTC queue if the video is actually ahead instead of potentially waiting forever. This could've happened if there was a bigger gap in the video stream.
2019-12-27 14:56:32 +00:00
timecodestamper->video_current_running_time = GST_CLOCK_TIME_NONE;
if (timecodestamper->video_clock_id)
gst_clock_id_unschedule (timecodestamper->video_clock_id);
timecodestamper: Add support for linear timecode (LTC) from an audio stream Based on a patch by Georg Lippitsch <glippitsch@toolsonair.com> Vivia Nikolaidou <vivia@toolsonair.com> Using libltc from https://github.com/x42/libltc
2019-07-01 10:43:28 +00:00
g_cond_signal (&timecodestamper->ltc_cond_video);
g_mutex_unlock (&timecodestamper->mutex);
}
return timecodestamper->video_activatemode_default (pad, parent, mode,
active);
}
static GstIterator *
gst_timecodestamper_src_iterate_internal_link (GstPad * pad, GstObject * parent)
{
GstTimeCodeStamper *timecodestamper = GST_TIME_CODE_STAMPER (parent);
GValue value = G_VALUE_INIT;
GstIterator *it;
g_value_init (&value, GST_TYPE_PAD);
g_value_set_object (&value, GST_BASE_TRANSFORM_SINK_PAD (timecodestamper));
it = gst_iterator_new_single (GST_TYPE_PAD, &value);
g_value_unset (&value);
return it;
}
#endif
Reference in a new issue Copy permalink