FreeRDP/channels/tsmf/client/gstreamer/tsmf_gstreamer.c

1071 lines
35 KiB
C
Raw Normal View History

/*
2012-10-09 07:02:04 +04:00
* FreeRDP: A Remote Desktop Protocol Implementation
* Video Redirection Virtual Channel - GStreamer Decoder
*
2014-05-23 15:46:37 +04:00
* (C) Copyright 2012 HP Development Company, LLC
* (C) Copyright 2014 Thincast Technologies GmbH
* (C) Copyright 2014 Armin Novak <armin.novak@thincast.com>
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
2019-11-06 17:24:51 +03:00
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
2021-06-09 15:03:34 +03:00
#include <winpr/assert.h>
2012-10-09 22:26:07 +04:00
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
2012-10-09 22:26:07 +04:00
2015-07-03 14:26:15 +03:00
#include <winpr/string.h>
#if __clang__
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wparentheses-equality"
#endif /* __clang__ */
#include <gst/gst.h>
#if __clang__
#pragma clang diagnostic pop
#endif /* __clang__ */
#include <gst/app/gstappsrc.h>
#include <gst/app/gstappsink.h>
#include "tsmf_constants.h"
#include "tsmf_decoder.h"
2014-05-23 15:46:37 +04:00
#include "tsmf_platform.h"
2012-10-09 22:52:07 +04:00
#ifdef HAVE_INTTYPES_H
#include <inttypes.h>
#endif
/* 1 second = 10,000,000 100ns units*/
2019-11-06 17:24:51 +03:00
#define SEEK_TOLERANCE 10 * 1000 * 1000
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
2014-11-10 22:02:54 +03:00
static BOOL tsmf_gstreamer_pipeline_build(TSMFGstreamerDecoder* mdecoder);
static void tsmf_gstreamer_clean_up(TSMFGstreamerDecoder* mdecoder);
static int tsmf_gstreamer_pipeline_set_state(TSMFGstreamerDecoder* mdecoder,
2019-11-06 17:24:51 +03:00
GstState desired_state);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
static BOOL tsmf_gstreamer_buffer_level(ITSMFDecoder* decoder);
2019-11-20 13:30:14 +03:00
static const char* get_type(TSMFGstreamerDecoder* mdecoder)
2014-05-23 15:46:37 +04:00
{
2014-11-10 22:02:54 +03:00
if (!mdecoder)
return NULL;
switch (mdecoder->media_type)
{
case TSMF_MAJOR_TYPE_VIDEO:
return "VIDEO";
case TSMF_MAJOR_TYPE_AUDIO:
return "AUDIO";
default:
return "UNKNOWN";
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
}
2019-11-06 17:24:51 +03:00
static void cb_child_added(GstChildProxy* child_proxy, GObject* object,
TSMFGstreamerDecoder* mdecoder)
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
{
DEBUG_TSMF("NAME: %s", G_OBJECT_TYPE_NAME(object));
2019-11-06 17:24:51 +03:00
if (!g_strcmp0(G_OBJECT_TYPE_NAME(object), "GstXvImageSink") ||
!g_strcmp0(G_OBJECT_TYPE_NAME(object), "GstXImageSink") ||
!g_strcmp0(G_OBJECT_TYPE_NAME(object), "GstFluVAAutoSink"))
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
{
2019-11-06 17:24:51 +03:00
gst_base_sink_set_max_lateness((GstBaseSink*)object, 10000000); /* nanoseconds */
g_object_set(G_OBJECT(object), "sync", TRUE, NULL); /* synchronize on the clock */
g_object_set(G_OBJECT(object), "async", TRUE, NULL); /* no async state changes */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
}
2019-11-06 17:24:51 +03:00
else if (!g_strcmp0(G_OBJECT_TYPE_NAME(object), "GstAlsaSink") ||
!g_strcmp0(G_OBJECT_TYPE_NAME(object), "GstPulseSink"))
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
{
2019-11-06 17:24:51 +03:00
gst_base_sink_set_max_lateness((GstBaseSink*)object, 10000000); /* nanoseconds */
g_object_set(G_OBJECT(object), "slave-method", 1, NULL);
2019-11-06 17:24:51 +03:00
g_object_set(G_OBJECT(object), "buffer-time", (gint64)20000, NULL); /* microseconds */
g_object_set(G_OBJECT(object), "drift-tolerance", (gint64)20000, NULL); /* microseconds */
g_object_set(G_OBJECT(object), "latency-time", (gint64)10000, NULL); /* microseconds */
g_object_set(G_OBJECT(object), "sync", TRUE, NULL); /* synchronize on the clock */
g_object_set(G_OBJECT(object), "async", TRUE, NULL); /* no async state changes */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
}
2014-05-23 15:46:37 +04:00
}
2019-11-06 17:24:51 +03:00
static void tsmf_gstreamer_enough_data(GstAppSrc* src, gpointer user_data)
2014-05-23 15:46:37 +04:00
{
2014-11-10 22:02:54 +03:00
TSMFGstreamerDecoder* mdecoder = user_data;
2019-11-06 17:24:51 +03:00
(void)mdecoder;
2014-05-23 15:46:37 +04:00
DEBUG_TSMF("%s", get_type(mdecoder));
}
2019-11-06 17:24:51 +03:00
static void tsmf_gstreamer_need_data(GstAppSrc* src, guint length, gpointer user_data)
2014-05-23 15:46:37 +04:00
{
2014-11-10 22:02:54 +03:00
TSMFGstreamerDecoder* mdecoder = user_data;
2019-11-06 17:24:51 +03:00
(void)mdecoder;
DEBUG_TSMF("%s length=%u", get_type(mdecoder), length);
2014-05-23 15:46:37 +04:00
}
2019-11-06 17:24:51 +03:00
static gboolean tsmf_gstreamer_seek_data(GstAppSrc* src, guint64 offset, gpointer user_data)
2014-05-23 15:46:37 +04:00
{
2014-11-10 22:02:54 +03:00
TSMFGstreamerDecoder* mdecoder = user_data;
2019-11-06 17:24:51 +03:00
(void)mdecoder;
DEBUG_TSMF("%s offset=%" PRIu64 "", get_type(mdecoder), offset);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
return TRUE;
}
static BOOL tsmf_gstreamer_change_volume(ITSMFDecoder* decoder, UINT32 newVolume, UINT32 muted)
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (!mdecoder || !mdecoder->pipe)
return TRUE;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->media_type == TSMF_MAJOR_TYPE_VIDEO)
return TRUE;
2019-11-06 17:24:51 +03:00
mdecoder->gstMuted = (BOOL)muted;
DEBUG_TSMF("mute=[%" PRId32 "]", mdecoder->gstMuted);
mdecoder->gstVolume = (double)newVolume / (double)10000;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
DEBUG_TSMF("gst_new_vol=[%f]", mdecoder->gstVolume);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (!mdecoder->volume)
return TRUE;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (!G_IS_OBJECT(mdecoder->volume))
return TRUE;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
g_object_set(mdecoder->volume, "mute", mdecoder->gstMuted, NULL);
g_object_set(mdecoder->volume, "volume", mdecoder->gstVolume, NULL);
return TRUE;
2014-05-23 15:46:37 +04:00
}
static inline GstClockTime tsmf_gstreamer_timestamp_ms_to_gst(UINT64 ms_timestamp)
{
2014-05-23 15:46:37 +04:00
/*
* Convert Microsoft 100ns timestamps to Gstreamer 1ns units.
*/
return (GstClockTime)(ms_timestamp * 100);
}
2014-11-10 22:02:54 +03:00
int tsmf_gstreamer_pipeline_set_state(TSMFGstreamerDecoder* mdecoder, GstState desired_state)
{
GstStateChangeReturn state_change;
2014-11-10 22:02:54 +03:00
const char* name;
const char* sname = get_type(mdecoder);
if (!mdecoder)
2014-05-23 15:46:37 +04:00
return 0;
if (!mdecoder->pipe)
2019-11-06 17:24:51 +03:00
return 0; /* Just in case this is called during startup or shutdown when we don't expect it
*/
if (desired_state == mdecoder->state)
2019-11-06 17:24:51 +03:00
return 0; /* Redundant request - Nothing to do */
2014-05-25 00:22:33 +04:00
name = gst_element_state_get_name(desired_state); /* For debug */
DEBUG_TSMF("%s to %s", sname, name);
2014-05-23 15:46:37 +04:00
state_change = gst_element_set_state(mdecoder->pipe, desired_state);
if (state_change == GST_STATE_CHANGE_FAILURE)
2014-11-10 22:02:54 +03:00
{
WLog_ERR(TAG, "%s: (%s) GST_STATE_CHANGE_FAILURE.", sname, name);
2014-11-10 22:02:54 +03:00
}
else if (state_change == GST_STATE_CHANGE_ASYNC)
{
WLog_ERR(TAG, "%s: (%s) GST_STATE_CHANGE_ASYNC.", sname, name);
mdecoder->state = desired_state;
}
else
2014-11-10 22:02:54 +03:00
{
mdecoder->state = desired_state;
2014-11-10 22:02:54 +03:00
}
return 0;
}
2014-11-10 22:02:54 +03:00
static GstBuffer* tsmf_get_buffer_from_data(const void* raw_data, gsize size)
{
2014-11-10 22:02:54 +03:00
GstBuffer* buffer;
2014-05-23 15:46:37 +04:00
gpointer data;
2014-11-10 22:02:54 +03:00
if (!raw_data)
return NULL;
if (size < 1)
return NULL;
2014-05-23 15:46:37 +04:00
data = g_malloc(size);
2014-05-25 00:22:33 +04:00
if (!data)
{
2019-11-06 17:24:51 +03:00
WLog_ERR(TAG, "Could not allocate %" G_GSIZE_FORMAT " bytes of data.", size);
2014-05-25 00:22:33 +04:00
return NULL;
}
2014-11-10 22:02:54 +03:00
CopyMemory(data, raw_data, size);
2014-05-25 00:22:33 +04:00
#if GST_VERSION_MAJOR > 0
buffer = gst_buffer_new_wrapped(data, size);
#else
buffer = gst_buffer_new();
2014-05-25 00:22:33 +04:00
if (!buffer)
{
WLog_ERR(TAG, "Could not create GstBuffer");
2014-05-25 00:22:33 +04:00
free(data);
return NULL;
}
GST_BUFFER_MALLOCDATA(buffer) = data;
GST_BUFFER_SIZE(buffer) = size;
GST_BUFFER_DATA(buffer) = GST_BUFFER_MALLOCDATA(buffer);
#endif
2014-11-10 22:02:54 +03:00
2014-05-25 00:22:33 +04:00
return buffer;
2014-05-23 15:46:37 +04:00
}
2014-11-10 22:02:54 +03:00
static BOOL tsmf_gstreamer_set_format(ITSMFDecoder* decoder, TS_AM_MEDIA_TYPE* media_type)
2014-05-23 15:46:37 +04:00
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
if (!mdecoder)
2014-05-23 15:46:37 +04:00
return FALSE;
2014-05-23 15:46:37 +04:00
DEBUG_TSMF("");
switch (media_type->MajorType)
{
case TSMF_MAJOR_TYPE_VIDEO:
mdecoder->media_type = TSMF_MAJOR_TYPE_VIDEO;
break;
case TSMF_MAJOR_TYPE_AUDIO:
mdecoder->media_type = TSMF_MAJOR_TYPE_AUDIO;
break;
default:
return FALSE;
}
switch (media_type->SubType)
{
case TSMF_SUB_TYPE_WVC1:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-wmv", "bitrate", G_TYPE_UINT, media_type->BitRate, "width", G_TYPE_INT,
media_type->Width, "height", G_TYPE_INT, media_type->Height, "wmvversion",
G_TYPE_INT, 3,
#if GST_VERSION_MAJOR > 0
2019-11-06 17:24:51 +03:00
"format", G_TYPE_STRING, "WVC1",
#else
2019-11-06 17:24:51 +03:00
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('W', 'V', 'C', '1'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, "pixel-aspect-ratio", GST_TYPE_FRACTION,
1, 1, NULL);
break;
case TSMF_SUB_TYPE_MP4S:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-divx", "divxversion", G_TYPE_INT, 5, "bitrate", G_TYPE_UINT,
media_type->BitRate, "width", G_TYPE_INT, media_type->Width, "height", G_TYPE_INT,
media_type->Height,
#if GST_VERSION_MAJOR > 0
2019-11-06 17:24:51 +03:00
"format", G_TYPE_STRING, "MP42",
#else
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('M', 'P', '4', '2'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, NULL);
break;
case TSMF_SUB_TYPE_MP42:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-msmpeg", "msmpegversion", G_TYPE_INT, 42, "bitrate", G_TYPE_UINT,
media_type->BitRate, "width", G_TYPE_INT, media_type->Width, "height", G_TYPE_INT,
media_type->Height,
#if GST_VERSION_MAJOR > 0
"format", G_TYPE_STRING, "MP42",
#else
2019-11-06 17:24:51 +03:00
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('M', 'P', '4', '2'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, NULL);
break;
case TSMF_SUB_TYPE_MP43:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-msmpeg", "msmpegversion", G_TYPE_INT, 43, "bitrate", G_TYPE_UINT,
media_type->BitRate, "width", G_TYPE_INT, media_type->Width, "height", G_TYPE_INT,
media_type->Height,
#if GST_VERSION_MAJOR > 0
"format", G_TYPE_STRING, "MP43",
#else
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('M', 'P', '4', '3'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, NULL);
break;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
case TSMF_SUB_TYPE_M4S2:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/mpeg", "mpegversion", G_TYPE_INT, 4, "width", G_TYPE_INT, media_type->Width,
"height", G_TYPE_INT, media_type->Height,
#if GST_VERSION_MAJOR > 0
"format", G_TYPE_STRING, "M4S2",
#else
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('M', '4', 'S', '2'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, NULL);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
break;
case TSMF_SUB_TYPE_WMA9:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"audio/x-wma", "wmaversion", G_TYPE_INT, 3, "rate", G_TYPE_INT,
media_type->SamplesPerSecond.Numerator, "channels", G_TYPE_INT,
media_type->Channels, "bitrate", G_TYPE_INT, media_type->BitRate, "depth",
G_TYPE_INT, media_type->BitsPerSample, "width", G_TYPE_INT,
media_type->BitsPerSample, "block_align", G_TYPE_INT, media_type->BlockAlign, NULL);
break;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
case TSMF_SUB_TYPE_WMA1:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"audio/x-wma", "wmaversion", G_TYPE_INT, 1, "rate", G_TYPE_INT,
media_type->SamplesPerSecond.Numerator, "channels", G_TYPE_INT,
media_type->Channels, "bitrate", G_TYPE_INT, media_type->BitRate, "depth",
G_TYPE_INT, media_type->BitsPerSample, "width", G_TYPE_INT,
media_type->BitsPerSample, "block_align", G_TYPE_INT, media_type->BlockAlign, NULL);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
break;
case TSMF_SUB_TYPE_WMA2:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"audio/x-wma", "wmaversion", G_TYPE_INT, 2, "rate", G_TYPE_INT,
media_type->SamplesPerSecond.Numerator, "channels", G_TYPE_INT,
media_type->Channels, "bitrate", G_TYPE_INT, media_type->BitRate, "depth",
G_TYPE_INT, media_type->BitsPerSample, "width", G_TYPE_INT,
media_type->BitsPerSample, "block_align", G_TYPE_INT, media_type->BlockAlign, NULL);
break;
case TSMF_SUB_TYPE_MP3:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps =
gst_caps_new_simple("audio/mpeg", "mpegversion", G_TYPE_INT, 1, "layer", G_TYPE_INT,
3, "rate", G_TYPE_INT, media_type->SamplesPerSecond.Numerator,
"channels", G_TYPE_INT, media_type->Channels, NULL);
break;
case TSMF_SUB_TYPE_WMV1:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-wmv", "bitrate", G_TYPE_UINT, media_type->BitRate, "width", G_TYPE_INT,
media_type->Width, "height", G_TYPE_INT, media_type->Height, "wmvversion",
G_TYPE_INT, 1,
#if GST_VERSION_MAJOR > 0
"format", G_TYPE_STRING, "WMV1",
#else
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('W', 'M', 'V', '1'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, NULL);
break;
case TSMF_SUB_TYPE_WMV2:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-wmv", "width", G_TYPE_INT, media_type->Width, "height", G_TYPE_INT,
media_type->Height, "wmvversion", G_TYPE_INT, 2,
#if GST_VERSION_MAJOR > 0
"format", G_TYPE_STRING, "WMV2",
#else
2019-11-06 17:24:51 +03:00
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('W', 'M', 'V', '2'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, "pixel-aspect-ratio", GST_TYPE_FRACTION,
1, 1, NULL);
break;
case TSMF_SUB_TYPE_WMV3:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-wmv", "bitrate", G_TYPE_UINT, media_type->BitRate, "width", G_TYPE_INT,
media_type->Width, "height", G_TYPE_INT, media_type->Height, "wmvversion",
G_TYPE_INT, 3,
#if GST_VERSION_MAJOR > 0
"format", G_TYPE_STRING, "WMV3",
#else
2019-11-06 17:24:51 +03:00
"format", GST_TYPE_FOURCC, GST_MAKE_FOURCC('W', 'M', 'V', '3'),
#endif
2019-11-06 17:24:51 +03:00
"framerate", GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, "pixel-aspect-ratio", GST_TYPE_FRACTION,
1, 1, NULL);
break;
case TSMF_SUB_TYPE_AVC1:
case TSMF_SUB_TYPE_H264:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-h264", "width", G_TYPE_INT, media_type->Width, "height", G_TYPE_INT,
media_type->Height, "framerate", GST_TYPE_FRACTION,
media_type->SamplesPerSecond.Numerator, media_type->SamplesPerSecond.Denominator,
"pixel-aspect-ratio", GST_TYPE_FRACTION, 1, 1, "stream-format", G_TYPE_STRING,
"byte-stream", "alignment", G_TYPE_STRING, "nal", NULL);
break;
case TSMF_SUB_TYPE_AC3:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"audio/x-ac3", "rate", G_TYPE_INT, media_type->SamplesPerSecond.Numerator,
"channels", G_TYPE_INT, media_type->Channels, NULL);
break;
case TSMF_SUB_TYPE_AAC:
/* For AAC the pFormat is a HEAACWAVEINFO struct, and the codec data
is at the end of it. See
http://msdn.microsoft.com/en-us/library/dd757806.aspx */
if (media_type->ExtraData)
{
media_type->ExtraData += 12;
media_type->ExtraDataSize -= 12;
}
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"audio/mpeg", "rate", G_TYPE_INT, media_type->SamplesPerSecond.Numerator,
"channels", G_TYPE_INT, media_type->Channels, "mpegversion", G_TYPE_INT, 4,
"framed", G_TYPE_BOOLEAN, TRUE, "stream-format", G_TYPE_STRING, "raw", NULL);
break;
case TSMF_SUB_TYPE_MP1A:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps =
gst_caps_new_simple("audio/mpeg", "mpegversion", G_TYPE_INT, 1, "channels",
G_TYPE_INT, media_type->Channels, NULL);
break;
case TSMF_SUB_TYPE_MP1V:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps =
gst_caps_new_simple("video/mpeg", "mpegversion", G_TYPE_INT, 1, "width", G_TYPE_INT,
media_type->Width, "height", G_TYPE_INT, media_type->Height,
"systemstream", G_TYPE_BOOLEAN, FALSE, NULL);
break;
case TSMF_SUB_TYPE_YUY2:
#if GST_VERSION_MAJOR > 0
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-raw", "format", G_TYPE_STRING, "YUY2", "width", G_TYPE_INT,
media_type->Width, "height", G_TYPE_INT, media_type->Height, NULL);
#else
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple(
"video/x-raw-yuv", "format", G_TYPE_STRING, "YUY2", "width", G_TYPE_INT,
media_type->Width, "height", G_TYPE_INT, media_type->Height, "framerate",
GST_TYPE_FRACTION, media_type->SamplesPerSecond.Numerator,
media_type->SamplesPerSecond.Denominator, NULL);
#endif
break;
case TSMF_SUB_TYPE_MP2V:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple("video/mpeg", "mpegversion", G_TYPE_INT, 2,
"systemstream", G_TYPE_BOOLEAN, FALSE, NULL);
break;
case TSMF_SUB_TYPE_MP2A:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps =
gst_caps_new_simple("audio/mpeg", "mpegversion", G_TYPE_INT, 1, "rate", G_TYPE_INT,
media_type->SamplesPerSecond.Numerator, "channels", G_TYPE_INT,
media_type->Channels, NULL);
break;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
case TSMF_SUB_TYPE_FLAC:
2019-11-06 17:24:51 +03:00
mdecoder->gst_caps = gst_caps_new_simple("audio/x-flac", "", NULL);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
break;
default:
WLog_ERR(TAG, "unknown format:(%d).", media_type->SubType);
return FALSE;
}
if (media_type->ExtraDataSize > 0)
2014-05-23 15:46:37 +04:00
{
2019-11-06 17:24:51 +03:00
GstBuffer* buffer;
DEBUG_TSMF("Extra data available (%" PRIu32 ")", media_type->ExtraDataSize);
2014-05-25 00:22:33 +04:00
buffer = tsmf_get_buffer_from_data(media_type->ExtraData, media_type->ExtraDataSize);
if (!buffer)
{
WLog_ERR(TAG, "could not allocate GstBuffer!");
2014-05-25 00:22:33 +04:00
return FALSE;
}
gst_caps_set_simple(mdecoder->gst_caps, "codec_data", GST_TYPE_BUFFER, buffer, NULL);
2014-05-23 15:46:37 +04:00
}
2019-11-06 17:24:51 +03:00
DEBUG_TSMF("%p format '%s'", (void*)mdecoder, gst_caps_to_string(mdecoder->gst_caps));
2014-05-23 15:46:37 +04:00
tsmf_platform_set_format(mdecoder);
2014-05-23 15:46:37 +04:00
/* Create the pipeline... */
if (!tsmf_gstreamer_pipeline_build(mdecoder))
2014-05-23 15:46:37 +04:00
return FALSE;
return TRUE;
}
void tsmf_gstreamer_clean_up(TSMFGstreamerDecoder* mdecoder)
{
if (!mdecoder || !mdecoder->pipe)
2014-05-23 15:46:37 +04:00
return;
if (mdecoder->pipe && GST_OBJECT_REFCOUNT_VALUE(mdecoder->pipe) > 0)
2014-05-23 15:46:37 +04:00
{
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_NULL);
2014-05-23 15:46:37 +04:00
gst_object_unref(mdecoder->pipe);
}
2014-05-23 15:46:37 +04:00
mdecoder->ready = FALSE;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
mdecoder->paused = FALSE;
2014-05-23 15:46:37 +04:00
mdecoder->pipe = NULL;
mdecoder->src = NULL;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
mdecoder->queue = NULL;
2014-05-23 15:46:37 +04:00
}
2014-11-10 22:02:54 +03:00
BOOL tsmf_gstreamer_pipeline_build(TSMFGstreamerDecoder* mdecoder)
2014-05-23 15:46:37 +04:00
{
#if GST_VERSION_MAJOR > 0
2019-11-06 17:24:51 +03:00
const char* video =
"appsrc name=videosource ! queue2 name=videoqueue ! decodebin name=videodecoder !";
const char* audio =
"appsrc name=audiosource ! queue2 name=audioqueue ! decodebin name=audiodecoder ! "
"audioconvert ! audiorate ! audioresample ! volume name=audiovolume !";
#else
2019-11-06 17:24:51 +03:00
const char* video =
"appsrc name=videosource ! queue2 name=videoqueue ! decodebin2 name=videodecoder !";
const char* audio =
"appsrc name=audiosource ! queue2 name=audioqueue ! decodebin2 name=audiodecoder ! "
"audioconvert ! audiorate ! audioresample ! volume name=audiovolume !";
#endif
char pipeline[1024];
if (!mdecoder)
return FALSE;
/* TODO: Construction of the pipeline from a string allows easy overwrite with arguments.
* The only fixed elements necessary are appsrc and the volume element for audio streams.
* The rest could easily be provided in gstreamer pipeline notation from command line. */
if (mdecoder->media_type == TSMF_MAJOR_TYPE_VIDEO)
2019-11-06 17:24:51 +03:00
sprintf_s(pipeline, sizeof(pipeline), "%s %s name=videosink", video,
tsmf_platform_get_video_sink());
else
2019-11-06 17:24:51 +03:00
sprintf_s(pipeline, sizeof(pipeline), "%s %s name=audiosink", audio,
tsmf_platform_get_audio_sink());
DEBUG_TSMF("pipeline=%s", pipeline);
mdecoder->pipe = gst_parse_launch(pipeline, NULL);
if (!mdecoder->pipe)
{
WLog_ERR(TAG, "Failed to create new pipe");
return FALSE;
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->media_type == TSMF_MAJOR_TYPE_VIDEO)
mdecoder->src = gst_bin_get_by_name(GST_BIN(mdecoder->pipe), "videosource");
else
mdecoder->src = gst_bin_get_by_name(GST_BIN(mdecoder->pipe), "audiosource");
if (!mdecoder->src)
{
WLog_ERR(TAG, "Failed to get appsrc");
return FALSE;
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->media_type == TSMF_MAJOR_TYPE_VIDEO)
mdecoder->queue = gst_bin_get_by_name(GST_BIN(mdecoder->pipe), "videoqueue");
else
mdecoder->queue = gst_bin_get_by_name(GST_BIN(mdecoder->pipe), "audioqueue");
if (!mdecoder->queue)
{
WLog_ERR(TAG, "Failed to get queue");
return FALSE;
}
if (mdecoder->media_type == TSMF_MAJOR_TYPE_VIDEO)
mdecoder->outsink = gst_bin_get_by_name(GST_BIN(mdecoder->pipe), "videosink");
else
mdecoder->outsink = gst_bin_get_by_name(GST_BIN(mdecoder->pipe), "audiosink");
if (!mdecoder->outsink)
2014-05-23 15:46:37 +04:00
{
WLog_ERR(TAG, "Failed to get sink");
return FALSE;
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
g_signal_connect(mdecoder->outsink, "child-added", G_CALLBACK(cb_child_added), mdecoder);
if (mdecoder->media_type == TSMF_MAJOR_TYPE_AUDIO)
{
mdecoder->volume = gst_bin_get_by_name(GST_BIN(mdecoder->pipe), "audiovolume");
if (!mdecoder->volume)
{
WLog_ERR(TAG, "Failed to get volume");
2014-05-23 15:46:37 +04:00
return FALSE;
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
2019-11-06 17:24:51 +03:00
tsmf_gstreamer_change_volume((ITSMFDecoder*)mdecoder, mdecoder->gstVolume * ((double)10000),
mdecoder->gstMuted);
}
2014-05-23 15:46:37 +04:00
tsmf_platform_register_handler(mdecoder);
/* AppSrc settings */
2019-11-06 17:24:51 +03:00
GstAppSrcCallbacks callbacks = {
tsmf_gstreamer_need_data, tsmf_gstreamer_enough_data, tsmf_gstreamer_seek_data, { NULL }
2014-05-23 15:46:37 +04:00
};
g_object_set(mdecoder->src, "format", GST_FORMAT_TIME, NULL);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
g_object_set(mdecoder->src, "is-live", FALSE, NULL);
g_object_set(mdecoder->src, "block", FALSE, NULL);
g_object_set(mdecoder->src, "blocksize", 1024, NULL);
2019-11-06 17:24:51 +03:00
gst_app_src_set_caps((GstAppSrc*)mdecoder->src, mdecoder->gst_caps);
gst_app_src_set_callbacks((GstAppSrc*)mdecoder->src, &callbacks, mdecoder, NULL);
gst_app_src_set_stream_type((GstAppSrc*)mdecoder->src, GST_APP_STREAM_TYPE_SEEKABLE);
gst_app_src_set_latency((GstAppSrc*)mdecoder->src, 0, -1);
gst_app_src_set_max_bytes((GstAppSrc*)mdecoder->src, (guint64)0); // unlimited
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
g_object_set(G_OBJECT(mdecoder->queue), "use-buffering", FALSE, NULL);
g_object_set(G_OBJECT(mdecoder->queue), "use-rate-estimate", FALSE, NULL);
g_object_set(G_OBJECT(mdecoder->queue), "max-size-buffers", 0, NULL);
g_object_set(G_OBJECT(mdecoder->queue), "max-size-bytes", 0, NULL);
2019-11-06 17:24:51 +03:00
g_object_set(G_OBJECT(mdecoder->queue), "max-size-time", (guint64)0, NULL);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
2019-11-06 17:24:51 +03:00
/* Only set these properties if not an autosink, otherwise we will set properties when real
* sinks are added */
if (!g_strcmp0(G_OBJECT_TYPE_NAME(mdecoder->outsink), "GstAutoVideoSink") &&
!g_strcmp0(G_OBJECT_TYPE_NAME(mdecoder->outsink), "GstAutoAudioSink"))
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
{
if (mdecoder->media_type == TSMF_MAJOR_TYPE_VIDEO)
{
2019-11-06 17:24:51 +03:00
gst_base_sink_set_max_lateness((GstBaseSink*)mdecoder->outsink,
10000000); /* nanoseconds */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
}
else
{
2019-11-06 17:24:51 +03:00
gst_base_sink_set_max_lateness((GstBaseSink*)mdecoder->outsink,
10000000); /* nanoseconds */
g_object_set(G_OBJECT(mdecoder->outsink), "buffer-time", (gint64)20000,
NULL); /* microseconds */
g_object_set(G_OBJECT(mdecoder->outsink), "drift-tolerance", (gint64)20000,
NULL); /* microseconds */
g_object_set(G_OBJECT(mdecoder->outsink), "latency-time", (gint64)10000,
NULL); /* microseconds */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
g_object_set(G_OBJECT(mdecoder->outsink), "slave-method", 1, NULL);
}
2019-11-06 17:24:51 +03:00
g_object_set(G_OBJECT(mdecoder->outsink), "sync", TRUE,
NULL); /* synchronize on the clock */
g_object_set(G_OBJECT(mdecoder->outsink), "async", TRUE, NULL); /* no async state changes */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
}
2014-05-23 15:46:37 +04:00
tsmf_window_create(mdecoder);
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_READY);
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_PLAYING);
mdecoder->pipeline_start_time_valid = 0;
mdecoder->shutdown = 0;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
mdecoder->paused = FALSE;
2019-11-06 17:24:51 +03:00
GST_DEBUG_BIN_TO_DOT_FILE(GST_BIN(mdecoder->pipe), GST_DEBUG_GRAPH_SHOW_ALL,
get_type(mdecoder));
return TRUE;
}
2019-11-06 17:24:51 +03:00
static BOOL tsmf_gstreamer_decodeEx(ITSMFDecoder* decoder, const BYTE* data, UINT32 data_size,
UINT32 extensions, UINT64 start_time, UINT64 end_time,
UINT64 duration)
{
2019-11-06 17:24:51 +03:00
GstBuffer* gst_buf;
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
2014-05-23 15:46:37 +04:00
UINT64 sample_time = tsmf_gstreamer_timestamp_ms_to_gst(start_time);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
BOOL useTimestamps = TRUE;
if (!mdecoder)
{
WLog_ERR(TAG, "Decoder not initialized!");
return FALSE;
}
2014-05-23 15:46:37 +04:00
/*
* This function is always called from a stream-specific thread.
* It should be alright to block here if necessary.
* We don't expect to block here often, since the pipeline should
* have more than enough buffering.
2014-05-23 15:46:37 +04:00
*/
2019-11-06 17:24:51 +03:00
DEBUG_TSMF(
"%s. Start:(%" PRIu64 ") End:(%" PRIu64 ") Duration:(%" PRIu64 ") Last Start:(%" PRIu64 ")",
get_type(mdecoder), start_time, end_time, duration, mdecoder->last_sample_start_time);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->shutdown)
{
WLog_ERR(TAG, "decodeEx called on shutdown decoder");
return TRUE;
}
if (mdecoder->gst_caps == NULL)
{
WLog_ERR(TAG, "tsmf_gstreamer_set_format not called or invalid format.");
return FALSE;
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (!mdecoder->pipe)
tsmf_gstreamer_pipeline_build(mdecoder);
if (!mdecoder->src)
{
2019-11-06 17:24:51 +03:00
WLog_ERR(
TAG,
"failed to construct pipeline correctly. Unable to push buffer to source element.");
2014-05-23 15:46:37 +04:00
return FALSE;
}
2014-05-23 15:46:37 +04:00
gst_buf = tsmf_get_buffer_from_data(data, data_size);
if (gst_buf == NULL)
{
2019-11-06 17:24:51 +03:00
WLog_ERR(TAG, "tsmf_get_buffer_from_data(%p, %" PRIu32 ") failed.", (void*)data, data_size);
return FALSE;
}
/* Relative timestamping will sometimes be set to 0
* so we ignore these timestamps just to be safe(bit 8)
*/
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (extensions & 0x00000080)
{
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
DEBUG_TSMF("Ignoring the timestamps - relative - bit 8");
useTimestamps = FALSE;
}
/* If no timestamps exist then we dont want to look at the timestamp values (bit 7) */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (extensions & 0x00000040)
{
DEBUG_TSMF("Ignoring the timestamps - none - bit 7");
useTimestamps = FALSE;
}
/* If performing a seek */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->seeking)
{
mdecoder->seeking = FALSE;
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_PAUSED);
2019-11-06 17:24:51 +03:00
mdecoder->pipeline_start_time_valid = 0;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->pipeline_start_time_valid)
{
2019-11-06 17:24:51 +03:00
DEBUG_TSMF("%s start time %" PRIu64 "", get_type(mdecoder), start_time);
/* Adjusted the condition for a seek to be based on start time only
* WMV1 and WMV2 files in particular have bad end time and duration values
* there seems to be no real side effects of just using the start time instead
*/
2019-11-06 17:24:51 +03:00
UINT64 minTime = mdecoder->last_sample_start_time - (UINT64)SEEK_TOLERANCE;
UINT64 maxTime = mdecoder->last_sample_start_time + (UINT64)SEEK_TOLERANCE;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
2019-11-06 17:24:51 +03:00
/* Make sure the minTime stops at 0 , should we be at the beginning of the stream */
if (mdecoder->last_sample_start_time < (UINT64)SEEK_TOLERANCE)
minTime = 0;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
2019-11-06 17:24:51 +03:00
/* If the start_time is valid and different from the previous start time by more than the
* seek tolerance, then we have a seek condition */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (((start_time > maxTime) || (start_time < minTime)) && useTimestamps)
{
2019-11-06 17:24:51 +03:00
DEBUG_TSMF("tsmf_gstreamer_decodeEx: start_time=[%" PRIu64
"] > last_sample_start_time=[%" PRIu64 "] OR ",
start_time, mdecoder->last_sample_start_time);
DEBUG_TSMF("tsmf_gstreamer_decodeEx: start_time=[%" PRIu64
"] < last_sample_start_time=[%" PRIu64 "] with",
start_time, mdecoder->last_sample_start_time);
DEBUG_TSMF(
"tsmf_gstreamer_decodeEX: a tolerance of more than [%lu] from the last sample",
SEEK_TOLERANCE);
DEBUG_TSMF("tsmf_gstreamer_decodeEX: minTime=[%" PRIu64 "] maxTime=[%" PRIu64 "]",
minTime, maxTime);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
mdecoder->seeking = TRUE;
2019-11-06 17:24:51 +03:00
/* since we cant make the gstreamer pipeline jump to the new start time after a seek -
* we just maintain a offset between realtime and gstreamer time
*/
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
mdecoder->seek_offset = start_time;
}
}
2014-05-23 15:46:37 +04:00
else
{
2019-11-06 17:24:51 +03:00
DEBUG_TSMF("%s start time %" PRIu64 "", get_type(mdecoder), start_time);
/* Always set base/start time to 0. Will use seek offset to translate real buffer times
2019-11-06 17:24:51 +03:00
* back to 0. This allows the video to be started from anywhere and the ability to handle
* seeks without rebuilding the pipeline, etc. since that is costly
*/
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
gst_element_set_base_time(mdecoder->pipe, tsmf_gstreamer_timestamp_ms_to_gst(0));
gst_element_set_start_time(mdecoder->pipe, tsmf_gstreamer_timestamp_ms_to_gst(0));
2014-05-23 15:46:37 +04:00
mdecoder->pipeline_start_time_valid = 1;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
/* Set the seek offset if buffer has valid timestamps. */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (useTimestamps)
mdecoder->seek_offset = start_time;
if (!gst_element_seek(mdecoder->pipe, 1.0, GST_FORMAT_TIME, GST_SEEK_FLAG_FLUSH,
2019-11-06 17:24:51 +03:00
GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, GST_CLOCK_TIME_NONE))
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
{
WLog_ERR(TAG, "seek failed");
}
}
2014-05-25 00:22:33 +04:00
#if GST_VERSION_MAJOR > 0
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (useTimestamps)
2019-11-06 17:24:51 +03:00
GST_BUFFER_PTS(gst_buf) =
sample_time - tsmf_gstreamer_timestamp_ms_to_gst(mdecoder->seek_offset);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
else
GST_BUFFER_PTS(gst_buf) = GST_CLOCK_TIME_NONE;
2014-05-25 00:22:33 +04:00
#else
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (useTimestamps)
2019-11-06 17:24:51 +03:00
GST_BUFFER_TIMESTAMP(gst_buf) =
sample_time - tsmf_gstreamer_timestamp_ms_to_gst(mdecoder->seek_offset);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
else
GST_BUFFER_TIMESTAMP(gst_buf) = GST_CLOCK_TIME_NONE;
2014-05-25 00:22:33 +04:00
#endif
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
GST_BUFFER_DURATION(gst_buf) = GST_CLOCK_TIME_NONE;
GST_BUFFER_OFFSET(gst_buf) = GST_BUFFER_OFFSET_NONE;
#if GST_VERSION_MAJOR > 0
#else
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
gst_buffer_set_caps(gst_buf, mdecoder->gst_caps);
#endif
gst_app_src_push_buffer(GST_APP_SRC(mdecoder->src), gst_buf);
/* Should only update the last timestamps if the current ones are valid */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (useTimestamps)
{
mdecoder->last_sample_start_time = start_time;
mdecoder->last_sample_end_time = end_time;
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->pipe && (GST_STATE(mdecoder->pipe) != GST_STATE_PLAYING))
{
2019-11-06 17:24:51 +03:00
DEBUG_TSMF("%s: state=%s", get_type(mdecoder),
gst_element_state_get_name(GST_STATE(mdecoder->pipe)));
2019-11-06 17:24:51 +03:00
DEBUG_TSMF("%s Paused: %" PRIi32 " Shutdown: %i Ready: %" PRIi32 "", get_type(mdecoder),
mdecoder->paused, mdecoder->shutdown, mdecoder->ready);
if (!mdecoder->paused && !mdecoder->shutdown && mdecoder->ready)
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_PLAYING);
}
return TRUE;
}
2019-11-06 17:24:51 +03:00
static BOOL tsmf_gstreamer_control(ITSMFDecoder* decoder, ITSMFControlMsg control_msg, UINT32* arg)
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
if (!mdecoder)
{
WLog_ERR(TAG, "Control called with no decoder!");
return TRUE;
}
if (control_msg == Control_Pause)
{
2014-05-23 15:46:37 +04:00
DEBUG_TSMF("Control_Pause %s", get_type(mdecoder));
if (mdecoder->paused)
{
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
WLog_ERR(TAG, "%s: Ignoring Control_Pause, already received!", get_type(mdecoder));
return TRUE;
}
2014-05-23 15:46:37 +04:00
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_PAUSED);
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
mdecoder->shutdown = 0;
2014-05-23 15:46:37 +04:00
mdecoder->paused = TRUE;
}
else if (control_msg == Control_Resume)
{
DEBUG_TSMF("Control_Resume %s", get_type(mdecoder));
if (!mdecoder->paused && !mdecoder->shutdown)
{
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
WLog_ERR(TAG, "%s: Ignoring Control_Resume, already received!", get_type(mdecoder));
return TRUE;
}
2014-05-23 15:46:37 +04:00
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
mdecoder->shutdown = 0;
mdecoder->paused = FALSE;
}
else if (control_msg == Control_Stop)
{
DEBUG_TSMF("Control_Stop %s", get_type(mdecoder));
if (mdecoder->shutdown)
{
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
WLog_ERR(TAG, "%s: Ignoring Control_Stop, already received!", get_type(mdecoder));
return TRUE;
}
/* Reset stamps, flush buffers, etc */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->pipe)
{
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_NULL);
tsmf_window_destroy(mdecoder);
tsmf_gstreamer_clean_up(mdecoder);
}
mdecoder->seek_offset = 0;
mdecoder->pipeline_start_time_valid = 0;
mdecoder->shutdown = 1;
}
else if (control_msg == Control_Restart)
{
DEBUG_TSMF("Control_Restart %s", get_type(mdecoder));
mdecoder->shutdown = 0;
mdecoder->paused = FALSE;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (mdecoder->pipeline_start_time_valid)
tsmf_gstreamer_pipeline_set_state(mdecoder, GST_STATE_PLAYING);
}
else
WLog_ERR(TAG, "Unknown control message %08x", control_msg);
return TRUE;
}
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
static BOOL tsmf_gstreamer_buffer_level(ITSMFDecoder* decoder)
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
2014-05-23 15:46:37 +04:00
DEBUG_TSMF("");
if (!mdecoder)
2014-05-23 15:46:37 +04:00
return FALSE;
guint clbuff = 0;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
if (G_IS_OBJECT(mdecoder->queue))
g_object_get(mdecoder->queue, "current-level-buffers", &clbuff, NULL);
DEBUG_TSMF("%s buffer level %u", get_type(mdecoder), clbuff);
return clbuff;
}
2014-11-10 22:02:54 +03:00
static void tsmf_gstreamer_free(ITSMFDecoder* decoder)
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
DEBUG_TSMF("%s", get_type(mdecoder));
if (mdecoder)
{
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
tsmf_window_destroy(mdecoder);
2014-05-23 15:46:37 +04:00
tsmf_gstreamer_clean_up(mdecoder);
if (mdecoder->gst_caps)
gst_caps_unref(mdecoder->gst_caps);
2014-05-23 15:46:37 +04:00
tsmf_platform_free(mdecoder);
2014-11-10 22:02:54 +03:00
ZeroMemory(mdecoder, sizeof(TSMFGstreamerDecoder));
free(mdecoder);
2014-05-23 15:46:37 +04:00
mdecoder = NULL;
}
}
2014-11-10 22:02:54 +03:00
static UINT64 tsmf_gstreamer_get_running_time(ITSMFDecoder* decoder)
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
if (!mdecoder)
return 0;
if (!mdecoder->outsink)
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
return mdecoder->last_sample_start_time;
if (!mdecoder->pipe)
return 0;
GstFormat fmt = GST_FORMAT_TIME;
gint64 pos = 0;
2014-05-25 00:22:33 +04:00
#if GST_VERSION_MAJOR > 0
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
gst_element_query_position(mdecoder->pipe, fmt, &pos);
2014-05-25 00:22:33 +04:00
#else
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
gst_element_query_position(mdecoder->pipe, &fmt, &pos);
2014-05-25 00:22:33 +04:00
#endif
2019-11-06 17:24:51 +03:00
return (UINT64)(pos / 100 + mdecoder->seek_offset);
}
2019-11-06 17:24:51 +03:00
static BOOL tsmf_gstreamer_update_rendering_area(ITSMFDecoder* decoder, int newX, int newY,
int newWidth, int newHeight, int numRectangles,
RDP_RECT* rectangles)
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
DEBUG_TSMF("x=%d, y=%d, w=%d, h=%d, rect=%d", newX, newY, newWidth, newHeight, numRectangles);
if (mdecoder->media_type == TSMF_MAJOR_TYPE_VIDEO)
2014-11-10 22:02:54 +03:00
{
2019-11-06 17:24:51 +03:00
return tsmf_window_resize(mdecoder, newX, newY, newWidth, newHeight, numRectangles,
rectangles) == 0;
2014-11-10 22:02:54 +03:00
}
return TRUE;
2014-05-23 15:46:37 +04:00
}
2019-11-20 13:30:14 +03:00
static BOOL tsmf_gstreamer_ack(ITSMFDecoder* decoder, BOOL (*cb)(void*, BOOL), void* stream)
2014-05-23 15:46:37 +04:00
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
2014-05-23 15:46:37 +04:00
DEBUG_TSMF("");
mdecoder->ack_cb = NULL;
2014-05-23 15:46:37 +04:00
mdecoder->stream = stream;
return TRUE;
}
2019-11-20 13:30:14 +03:00
static BOOL tsmf_gstreamer_sync(ITSMFDecoder* decoder, void (*cb)(void*), void* stream)
2014-05-23 15:46:37 +04:00
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* mdecoder = (TSMFGstreamerDecoder*)decoder;
2014-05-23 15:46:37 +04:00
DEBUG_TSMF("");
mdecoder->sync_cb = NULL;
mdecoder->stream = stream;
return TRUE;
}
#ifdef BUILTIN_CHANNELS
2019-11-06 17:24:51 +03:00
#define freerdp_tsmf_client_subsystem_entry gstreamer_freerdp_tsmf_client_decoder_subsystem_entry
2016-03-03 19:11:26 +03:00
#else
2019-11-06 17:24:51 +03:00
#define freerdp_tsmf_client_subsystem_entry FREERDP_API freerdp_tsmf_client_decoder_subsystem_entry
#endif
2014-11-10 22:02:54 +03:00
ITSMFDecoder* freerdp_tsmf_client_subsystem_entry(void)
{
2019-11-06 17:24:51 +03:00
TSMFGstreamerDecoder* decoder;
2019-11-06 17:24:51 +03:00
#if GST_CHECK_VERSION(0, 10, 31)
if (!gst_is_initialized())
{
2014-05-23 15:46:37 +04:00
gst_init(NULL, NULL);
}
#else
gst_init(NULL, NULL);
#endif
2014-11-10 22:02:54 +03:00
decoder = calloc(1, sizeof(TSMFGstreamerDecoder));
if (!decoder)
return NULL;
decoder->iface.SetFormat = tsmf_gstreamer_set_format;
decoder->iface.Decode = NULL;
decoder->iface.GetDecodedData = NULL;
decoder->iface.GetDecodedFormat = NULL;
decoder->iface.GetDecodedDimension = NULL;
decoder->iface.GetRunningTime = tsmf_gstreamer_get_running_time;
decoder->iface.UpdateRenderingArea = tsmf_gstreamer_update_rendering_area;
decoder->iface.Free = tsmf_gstreamer_free;
decoder->iface.Control = tsmf_gstreamer_control;
decoder->iface.DecodeEx = tsmf_gstreamer_decodeEx;
decoder->iface.ChangeVolume = tsmf_gstreamer_change_volume;
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
decoder->iface.BufferLevel = tsmf_gstreamer_buffer_level;
2014-05-23 15:46:37 +04:00
decoder->iface.SetAckFunc = tsmf_gstreamer_ack;
decoder->iface.SetSyncFunc = tsmf_gstreamer_sync;
decoder->paused = FALSE;
decoder->gstVolume = 0.5;
decoder->gstMuted = FALSE;
2019-11-06 17:24:51 +03:00
decoder->state = GST_STATE_VOID_PENDING; /* No real state yet */
- Use decodebin2 instead of old decodebin - decodebin has issues - Use autovideosink - xvimagesink does not work with cards with no xv ports available and cant be used if wanted to use the fluendo hardware accelerated playback codec - Use autoaudiosink - let gstreamer choose best audio playback plugin - Catch when autosinks add known elements so that we can manipulate properties on them - Adjust caps of various media types to work better with gstreamer, some codecs are picky about having certain fields available - Remove unneeded plugins such as "ffmpegcolorspace" and "videoscale" - these do not work correctly with fluendo hardware accelerated playback codec - Name audio/video gstreamer elements better for easier debugging - Update gstreamer pipeline and element properties to handle playback better - Detect when valid timestamps are available for buffer from server and try to account for when they are not valid - Start time is much more reliable then end time from server for various media formats, so use it when possible to make decisions instead of end time - Do not rebuild gstreamer pipeline for a seek(very expensive), instead reset gstreamer time to 0 and maintain offset between real time and gstreamer time - Change buffer filled function back to a buffer level function, so that we can use buffer level to make better choices above gstreamer decoder in tsmf - Remove ack function from gstreamer, instead rely on ack thread to handle acks - Rework X11 gstreamer code to handle various videosinks which implement the XOverlayInterface and to keep more detailed information on the sub-window that is used for display - Add check to see if a decoder is available for telling the server the client various media types - Add in support for M4S2 and WMA1 media types - Fix flush message handling, they are for individual streams and not the entire presentation - Delay eos response to try to allow more time for buffers to be loaded into decoder, as we anticipate acks to server and the server will issue stop as soon as we ack eos. - Fix issue with geometry info being ignored when resent for new streams within existing presentation - Fixed volume level initialization issue when a stream is stopped and restarted - Attempt to sync video/audio streams...because we run two different gstreamer pipelines - they can enter pause/playing states at different times and are thus not synchronized. Attempt to adjust video buffer timestamps based on difference between audio/video running time to account for this difference. This logic accounts for a huge improvement in audio/video sync(ie. lip sync to words)
2015-07-08 00:39:29 +03:00
decoder->last_sample_start_time = 0;
decoder->last_sample_end_time = 0;
decoder->seek_offset = 0;
decoder->seeking = FALSE;
2014-11-10 22:02:54 +03:00
if (tsmf_platform_create(decoder) < 0)
{
free(decoder);
return NULL;
}
2014-11-10 22:02:54 +03:00
2019-11-06 17:24:51 +03:00
return (ITSMFDecoder*)decoder;
}