Skip to end of metadata
Go to start of metadata

Overview

WCS allows mixing streams of active broadcasts. The output stream of the mixer can be recordedplayed or republished using any of technologies supported by WCS.
Mixing is controlled using settings and REST API.

Supported protocols of input streams

  • WebRTC
  • RTMP
  • RTSP

Output stream control capabilities

The mixer allows custom placing of video streams in the output frame. The stream with a certain name (by default desktop) is seen as screensharing and hence is placed in the center of the frame:

Automatically create a mixer when publishing the stream

If the name of the published RTMP stream has the '#' symbol, the server treats everything after that symbols as the name of the mixer that will be created when the stream is published. For instance, for the user1#room1 stream, the room1 mixer is created, and the stream is added to this mixer then. The stream name can also include the screen sharing keyword, for example, user1#room1#desktop

Operation flowchart

  1. The browser connects to the server via the Websocket protocol and sends the publish command.
  2. The browser sends the WebRTC stream1 to the server.
  3. Flash Player establishes a connection via RTMP and sends the publish command.
  4. Flash Player sends the RTMP stream2 to the server.
  5. The REST client creates a mixer with the output stream3 using the query: /mixer/startup
  6. The REST client adds stream1 to the mixer
  7. The REST client adds stream2 to the mixer
  8. The second browser establishes a connection via Websocket and sends the play command.
  9. The second browser receives the WebRTC audio stream stream3 and plays that stream on the page.

REST queries

A REST-query must be an HTTP/HTTPS POST query in the following form:

Here:

  • streaming.flashphoner.com - is the address of the WCS server
  • 8081 - the standard REST / HTTP port of the WCS server
  • 8444 - the standard HTTPS port
  • rest-api - the required prefix
  • mixer/startup - the REST-method used

REST-methods and response statuses

REST-method

Example of REST query

Example of response

Response statuses

Description

/mixer/startup

{
  "uri": "mixer://mixer1",
  "localStreamName": "stream3",
  "hasVideo": "true",
  "hasAudio": "true",
  "watermark": "watermark.png",
  "background": "background.png",
  "mixerLayoutClass": "com.flashphoner.mixerlayout.TestLayout"
}

200 - OK

400 - Bad request

409 - Conflict

500 - Internal error

Creates a mixer the provided stream is published for

/mixer/add

{
  "uri": "mixer://mixer1",
  "remoteStreamName": "stream1",
  "hasVideo": "true",
  "hasAudio": "true"
}

200 - OK

404 - Mixer not found

404 - Stream not found

500 - Internal error

Add the RTMP stream to the mixer

/mixer/remove

{
  "uri": "mixer://mixer1",
  "remoteStreamName": "stream1"
}

200 - OK

404 - Mixer not found

404 - Stream not found

500 - Internal error

Remove the RTMP stream from the mixer

/mixer/find_all


{
"localMediaSessionId": "ce92b134-2468-4460-8d06-1ea3c5aabace",
"remoteMediaSessionId": null,
"localStreamName": "mixer1",
"remoteStreamName": null,
"uri": "mixer://mixer1",
"status": "PROCESSED_LOCAL",
"mediaSessions": [
  "95bf2be8-f459-4f62-9a7f-c588f33e0ad3",
  "693781de-cada-4589-abe1-c3ee55c66901"
],
}

200 - OK

404 - Not found

500 - Internal error

Find all mixers

/mixer/terminate

{
  "uri": "mixer://mixer1"
}

200 - OK

404 - Not found

500 - Internal error

Terminate operation of the mixer

/stream/startRecording

{
  "mediaSessionId": "23d07fa1-3c74-4d6f-a0de-9b4bf83ce665"
}

200 - OK

404 - Not found

500 - Internal error

Start recording of the stream in the given media session

/stream/stopRecording

{
  "mediaSessionId": "23d07fa1-3c74-4d6f-a0de-9b4bf83ce665"
}

200 - OK

404 - Not found

500 - Internal error

Stop recording the stream in the given media session

/mixer/setAudioVideo

{
 "uri": "mixer://m1",
 "streams": "^stream.*",
 "audioLevel": 0,
 "videoMuted" true
}

200 - OK

400 - Bad request

404 - Not found

500 - Internal error

Mute/unmute video or change audio level for mexer incoming stream

Parameters

Parameter name

Description

Example

uri

Unique identifier of the mixer

mixer://mixer1

localStreamName

Name of the output stream of the mixer

stream3

hasVideoMix videotrue
hasAudioMix audiotrue

remoteStreamName

Name of the stream added to the mixer

stream1

rtmp://rtmp.flashphoner.com:1935/live/rtmp_stream1

mediaSessionId

Media session identifier

ce92b134-2468-4460-8d06-1ea3c5aabace

status

Stream status

PROCESSED_LOCAL

backgroundMixer backgroundbackground.png
watermarkMixer watermarkwatermark.png
mixerLayoutClassMixer layoutcom.flashphoner.mixerlayout.TestLayout
streamsStreams list or regular expression for search

^stream.*

["stream1", "stream2"]

audioLevelIncoming stream audio level0
videoMutedMute videotrue

Sending the REST query to the WCS server

To send the REST query to the WCS server use a REST-client.

Known limits

Mixer output stream name must be defined while creating mixer by REST API (localStreamName parameter). Otherwise, server returns 400 Bad request with message "No localStreamName given".

Configuration

Mixing can be configured using the following parameters in the flashphoner.properties settings file

Parameter

Default value

Description

mixer_video_desktop_layout_inline_padding

10

Distance (padding) between windows of video streams in the lower line (below the screen sharing window)

mixer_video_desktop_layout_padding

30

Distance (padding) between the screen sharing window and the lower line (the rest streams)

mixer_video_enabled

true

Enables (by default) or disables video mixing

mixer_video_grid_layout_middle_padding

10

Distance between windows of video streams in one line (without screen sharing window)

mixer_video_grid_layout_padding

30

Distance between lines of windows (without screen sharing window)

mixer_video_height

720

The image height of the mixer output stream, should have an even value. An uneven value will be auto decremented by one: e.g., if set to 481, mixer height will be 480.

mixer_video_layout_desktop_key_word

desktop

Keyword for the screen sharing stream

mixer_video_width

1280

The image width of the mixer output stream, should have an even value. An uneven value will be auto decremented by one: e.g., if set to 641, mixer width will be 640.

record_mixer_streams

false

Turns on or off (default) recording of all mixer output streams

Automatic mixer creation configuration

Automatic creation of mixers for streams with the '#' symbol in their name requires the application that handles input streams to register the handler: 'com.flashphoner.server.client.handler.wcs4.FlashRoomRecordingStreamingHandler'. Registering the handler can be done using the command line interface. For instance, for the flashStreamingApp application used to publish incoming RTMP streams this can be done with the following command:

update app -m com.flashphoner.server.client.handler.wcs4.FlashRoomRecordingStreamingHandler -c com.flashphoner.server.client.handler.wcs4.FlashStreamingCallbackHandler flashStreamingApp

You can read more about managing applications using the command line of the WCS server here.

Audio and video mixing configuration

By default, both video and audio streams are mixed. If audio only mixing is necessary, it should be set on mixer creation

{
 "uri": "mixer://mixer1",
 "localStreamName": "stream3",
 "hasVideo": "false"
}

To switch off video mixing for all streams, this parameter should be set in flashphoner.properties file

mixer_video_enabled=false

In this case video mixing can be switched on for certain mixer on its creation.

Since build 5.2.689 audio mixing can be enabled o disabled on mixer creation

{
 "uri": "mixer://mixer1",
 "localStreamName": "stream3",
 "hasAudio": "false"
}

or in server settings for all the streams

mixer_audio_enabled=false

Publishing a stream without audio or video track to AV mixer

When a stream with audio or video track only is published to AV mixer, it is necessary to disable RTP activity control with the following parameters

rtp_activity_audio=false
rtp_activity_video=false

Mixer output stream bufferization

In some cases, mixer output stream bufferization is needed. This feature is enabled with the following parameter in flashphoner.properties file

mixer_out_buffer_enabled=true

The buffer size is defined in milliseconds with parameter

mixer_out_buffer_start_size=400

In this case, the buffer size is 400 ms.

Stream data fetching from buffer and sending period is defined in milliseconds with parameter

mixer_out_buffer_polling_time=20

In this case, the period is 20 ms.

Changing bitrate of mixer output stream

When OpenH264 codec is used for transcoding, it is possible to change bitrate of mixer output stream with the following parameter in flashphoner.properties file

mixer_video_bitrate_kbps=2000

By default, mixer output stream bitrate is set to 2 Mbps. If a channel bandwidth between server and viewer is not enough, bitrate can be reduced, for example

encoder_priority=OPENH264
mixer_video_bitrate_kbps=1500

If picture quality with default bitrate is low, or distortion occurs, it is recommended to rise mixer output stream bitrate to 3-5 Mbps

encoder_priority=OPENH264
mixer_video_bitrate_kbps=5000

Mixer output stream sound management

By default, mixer output stream sound is encoded to Opus with sample rate 48 kHz. These settings may be changed using the parameters in flashphoner.properties file. For example, to use mixer output stream in SIP call the following value can be set:

audio_mixer_output_codec=pcma
audio_mixer_output_sample_rate=8000

In this case, sound will be encoded to PCMA (alaw) with sample rate 8 kHz.

Using custom lossless videoprocessor for incoming streams handling

To handle mixer incoming streams, if additional bufferizing or audio and video tracks syncronizing is required for example, the custom lossless videoprocessor may be used. This feature is enabled with the following parameter in flashphoner.properties file

mixer_lossless_video_processor_enabled=true

The maximum size of mixer buffer in milliseconds is set with this parameter

mixer_lossless_video_processor_max_mixer_buffer_size_ms=200

By default, maximum mixer buffer size is 200 ms. After filling this buffer, the custom lossless videoprocessor uses its own buffer and waits for mixer buffer freeing. The period of mixer buffer checking is set in milliseconds with this parameter

mixer_lossless_video_processor_wait_time_ms=20

By default, the mixer buffer checking period is 20 ms.

Note that using the custom lossless videoprocessor may degrade realtime performance.

When custom lossless videoprocessor is used, it is necessary to stop mixer with REST query /mixer/terminate to free all consumed resources. Mixer can be stopped also by stopping all incoming streams, in this case mixer will stop when following timeout in milliseconds expires

mixer_idle_timeout=60000

By default, mixer will stop after 60 seconds if there are no active incoming streams.

Mixer output stream layout management

By default, three mixer output stream layouts are implemented:

Grid layout

This layout can be enabled with the following parameter in flashphoner.properties file

mixer_layout_class=com.flashphoner.media.mixer.video.presentation.GridLayout

Zero padding grid layout

This layout can be enabled with the following parameter

mixer_layout_class=com.flashphoner.media.mixer.video.presentation.CenterNoPaddingGridLayout

and works for input streams of equal resolution with the same aspect ratio only

Desktop (screen sharing) layout

This layout is enabled if one of mixer input streams has a name defined in the following parameter

mixer_video_layout_desktop_key_word=desktop

By default, desktop name is used for screen sharing stream.

Since build 5.2.710 it is possible to change screen sharing picture placement with the following parameter:

mixer_desktop_align=TOP

By default, screen sharing picture is placed above the other pictures

The following placements are supported:

PlacementDescription
TOPScreen above
LEFTScreen on left side
RIGHTSreen on right side
BOTTOMScreen below
CENTERScreen in center surronded by other pictures

For example, the following parameter

mixer_desktop_align=RIGHT

places the screen picture on the right side of mixer output stream picture

If a number of screen sharing streams are published to the same mixer, the first published stream take the main place. If the stream stops after that, the main place will be taken by the following screen share stream in alphabetical order.

Picture in picture

This layout is added since build 5.2.852. In this case a stream that has a name defined in the following parameter

mixer_video_layout_desktop_key_word=desktop

will be background for other streams in mixer. This layout can be enabled with the following parameter

mixer_video_desktop_fullscreen=true

Custom mixer layout implementation

For more fine tuning of mixer layout, custom Java class should be developed to implement IVideoMixerLayout interface, for example

TestLayout.java
// Package name should be strictly defined as com.flashphoner.mixerlayout
package com.flashphoner.mixerlayout;

// Import mixer layout interface
import com.flashphoner.sdk.media.IVideoMixerLayout;
// Import YUV frame description
import com.flashphoner.sdk.media.YUVFrame;

// Import Java packages to use
import java.awt.*;
import java.util.ArrayList;

/**
 * Custom mixer layout implementation example
 */
public class TestLayout implements IVideoMixerLayout {

    // Pictures padding, should be even (or zero if no padding needed)
    private static final int PADDING = 4;

    /**
     * Function to compute layout, will be called by mixer before encoding output stream picture
     * This example computes grid layout
     * @param yuvFrames - incoming streams raw pictures array in YUV format
     * @param strings - incoming streams names array
     * @param canvasWidth - mixer output picture canwas width
     * @param canvasHeight - mixer output picture canwas heigth
     * @return array of pictures layouts
     */
    @Override
    public Layout[] computeLayout(YUVFrame[] yuvFrames, String[] strings, int canvasWidth, int canvasHeight) {
        // Declare picture layouts list to fill
        ArrayList<IVideoMixerLayout.Layout> layout = new ArrayList<>();

        // Find canvas center height
        int canvasCenter = canvasHeight / 2;
        // Find frame center
        int frameCenter = canvasCenter - (canvasHeight / yuvFrames.length) / 2;

        // Find every picture dimensions (this are the same for all the pictures because this is grid layout)
        int layoutWidth = canvasWidth / yuvFrames.length - PADDING;
        int layoutHeight = canvasHeight / yuvFrames.length;

        // Iterate through incoming stream pictures array
        // Note: streams pictures order corresponds to stream names array, use this to reorder pictures if needed
        for (int c = 0; c < yuvFrames.length; c++) {
            // Use Java AWT Point and Dimension
            Point prevPoint = new Point();
            Dimension prevDimension = new Dimension();
            if (layout.size() > 0) {
                // Find previous picture location to calculate next one
                prevPoint.setLocation(layout.get(c - 1).getPoint());
                prevDimension.setSize(layout.get(c - 1).getDimension());
            }
            // Set starting point of the picture
            Point currentPoint = new Point((int) (prevPoint.getX() + prevDimension.getWidth() + PADDING),
                    frameCenter);
            // Create the picture layout passing starting point, dimensions and raw picture YUV frames
            layout.add(new IVideoMixerLayout.Layout(currentPoint, new Dimension(layoutWidth,
                    layoutHeight), yuvFrames[c]));
        }
        // Return pictures layouts calculated as array
        return layout.toArray(new IVideoMixerLayout.Layout[layout.size()]);
    }
}

Then the class should be complied into byte code. To do this, create folder tree accordind to TestLayout class package name

mkdir -p com/flashphoner/mixerlayout

and execute the command

javac -cp /usr/local/FlashphonerWebCallServer/lib/wcs-core.jar ./com/flashphoner/mixerlayout/TestLayout.java

Now, pack the code compiled to jar file

jar -cf testlayout.jar ./com/flashphoner/mixerlayout/TestLayout.class

and copy this file to WCS libraries folder

cp testlayout.jar /usr/local/FlashphonerWebCallServer/lib

To use custom mixer layout class, set it to the following parameter in flashphoner.properties file

mixer_layout_class=com.flashphoner.mixerlayout.TestLayout

and restart WCS.

With this custom layout, mixer output stream for three input streams will look like:

Mixer layout management while creating mixer

Since build 5.2.693 mixer layout can be defined when creating mixer with REST API, for example

{
 "uri": "mixer://mixer1",
 "localStreamName": "mixer1",
 "mixerLayoutClass": "com.flashphoner.mixerlayout.TestLayout"
}

Then, layout can be defib=ned for every mixer separately

Mixer output stream encoding profile management

Some browsers do not support playback for H264 streams encoded by certain profiles. To solve it, the following parameter is added since build 5.2.414 to set mixer output stream encoding profile

mixer_video_profile_level=42c02a

By default, constrainted baseline level 4.2 profile is set.

Mixer background management and watermarking

Since build 5.2.693 mixer backgroung can be defined and watermark can be added when creating mixer with REST API, for example

{
 "uri": "mixer://mixer1",
 "localStreamName": "mixer1",
 "watermark": "watermark.png",
 "background": "background.png"
}

By default, files should be placed to /usr/local/FlashphonerWebCallServer/conf folder. Full path to the files can also be set, for example

{
 "uri": "mixer://mixer1",
 "localStreamName": "mixer1",
 "watermark": "/opt/media/watermark.png",
 "background": "/opt/media/background.png"
}

MCU support

Mixer MCU support for audio can be enabled with the following parameter

mixer_mcu_audio=true

In this case for three input streams stream1, stream2, stream3 the following output streams will be generated in mixer mixer1:

Output stream namestream1 stream2stream3
audiovideoaudiovideoaudiovideo
mixer1++++++
mixer1-stream1--+-+-
mixer1-stream2+---+-
mixer1-stream3+-+---

Thus, each of the additional streams contains audio of all streams in the mixer, except for one. This allows for example to eliminate echo for conference participants.

This feature can be also enabled for video with the following parameter

mixer_mcu_video=true

In this case for three input streams stream1, stream2, stream3 the following output streams will be generated in mixer mixer1:

Output stream namestream1 stream2stream3
audiovideoaudiovideoaudiovideo
mixer1++++++
mixer1-stream1--++++
mixer1-stream2++--++
mixer1-stream3++++--

Thus, each of the additional streams contains audio and video of all streams in the mixer, except for one. This allows to arrange a full-fledged chat room based on mixer.

In this case, if mixer recording is enabled with the parameter

record_mixer_streams=true

the main mixer output stream will be only recorded (mixer1 in the example above).

If video MCU support is enables, channel losses can affect mixer output stream quality. Custom lossless videoprocessor can be used to improve quality, that can make an additional latency.

Incoming streams audio and video management

Since build 5.2.835 it is possible to change audio level and mute video for mixer incoming streams. In this case, the original stream remains unchanged. Video track  can be muted (black screen) and then unmuted. For audio track, volume level can be set in percent up to 100, or sound can be muted by setting level to 0.

Incoming streams are managed using REST API query /mixer/setAudioVideo.

For example, create a mixer and add 3 streams to it: 2 participants and 1 speaker

curl -H "Content-Type: application/json" -X POST http://localhost:8081/rest-api/mixer/startup -d '{"uri": "mixer://m1", "localStreamName":"m1"}'
curl -H "Content-Type: application/json" -X POST http://localhost:8081/rest-api/mixer/add -d '{"uri": "mixer://m1", "remoteStreamName": "stream1"}'
curl -H "Content-Type: application/json" -X POST http://localhost:8081/rest-api/mixer/add -d '{"uri": "mixer://m1", "remoteStreamName": "stream2"}'
curl -H "Content-Type: application/json" -X POST http://localhost:8081/rest-api/mixer/add -d '{"uri": "mixer://m1", "remoteStreamName": "desktop"}'

Mute all participants excluding speaker

POST /rest-api/mixer/setAudioVideo HTTP/1.1
User-Agent: curl/7.29.0
Host: localhost:8081
Accept: */*
Content-Type: application/json
Content-Length: 62

{
 "uri": "mixer://m1",
 "streams": "^stream.*",
 "audioLevel": 0
}

REST query streams parameter may be set either as regular expression to match streams by name, or as streams list. Use regular expression syntax supported by Java, the examples can be found here.

Mute stream1 video

POST /rest-api/mixer/setAudioVideo HTTP/1.1
User-Agent: curl/7.29.0
Host: localhost:8081
Accept: */*
Content-Type: application/json
Content-Length: 65

{
 "uri": "mixer://m1",
 "streams": ["stream1"],
 "videoMuted": true
}

Check streams state by /mixer/find_all query

POST /rest-api/mixer/find_all HTTP/1.1
User-Agent: curl/7.29.0
Host: localhost:8081
Accept: */*
Content-Type: application/json
HTTP/1.1 200 OK
Access-Control-Allow-Origin: *
Access-Control-Allow-Credentials: true
Content-Type: application/json
Content-Length: 574

[
  {
    "localMediaSessionId": "e2fa5c8b-16f3-4917-9d5f-557dde75db07",
    "localStreamName": "m1",
    "uri": "mixer://m1",
    "status": "PROCESSED_LOCAL",
    "hasAudio": true,
    "hasVideo": true,
    "record": false,
    "mediaSessions": [
      {
        "localMediaSessionId": "3dd763b0-2ae7-11eb-aa72-37b2cbcbf6b9",
        "audioLevel": 0,
        "videoMuted": true,
        "localStreamName": "stream1"
      },
      {
        "localMediaSessionId": "8af64760-2ae7-11eb-b086-cdf035231b9d",
        "audioLevel": 100,
        "videoMuted": false,
        "localStreamName": "desktop"
      },
      {
        "localMediaSessionId": "7cc4b410-2ae7-11eb-b34c-a5240fe9f151",
        "audioLevel": 0,
        "videoMuted": false,
        "localStreamName": "stream2"
      }
    ]
  }
]

Quick manual on testing

1. For this test we use:

2. Open the page of the Two Way Streaming application. Publish the stream named stream1:


3. In another tab open the page of the Two Way Streaming application. Publish the stream named desktop:


4. Open the REST client. Send the /mixer/startup query and specify the URI of the mixer mixer://mixer1 and the output stream name stream3 in its parameters:


5. Send the /mixer/add query and specify the URI of the mixer mixer://mixer1 and the input stream name stream1 in its parameters:


6. Open the Player web application, specify the name of the output stream of the mixer stream3 in the Stream field and click Start:


7. Send /mixer/add and specify the URI of the mixer mixer://mixer1 and the input stream name desktop in its parameters:


8. In the output stream of the mixer you should see the desktop stream that imitates screen sharing and the stream stream1:

Call flow

Below is the call flow when using the mixer.


1. Publishing of the WebRTC stream stream1

2. Sending the WebRTC stream to the server

3. Publishing the RTMP stream stream2

4. Sending the RTMP stream to the server

5. Sending the /mixer/startup query to create the mixer://stream3 mixer with the output stream3

http://demo.flashphoner.com:9091/rest-api/mixer/startup
{
 "uri": "mixer://stream3",
 "localStreamName": "stream3"
}

6. Sending the /mixer/add query to add stream1 to the mixer://stream3 mixer

http://demo.flashphoner.com:9091/rest-api/mixer/add
{
 "uri": "mixer://stream3",
 "localStreamName": "stream3"
 "remoteStreamName": "stream1"
}

7. Sending the /mixer/add query to add stream2 to the mixer://stream3 mixer

http://demo.flashphoner.com:9091/rest-api/mixer/add
{
 "uri": "mixer://stream3",
 "localStreamName": "stream3"
 "remoteStreamName": "stream2"
}

8. Playing the WebRTC stream stream3

9. Sending the WebRTC audio stream to the client

Mixer REST hooks

By default, all the standard REST hooks are invoked for mixer stream. The /connect REST hook is invoked while creating a mixer.

A separate server application can be created for mixer REST hooks handling if necessary. To send REST hooks to this application, the following parameter can be used since build 5.2.634

mixer_app_name=defaultApp

By default, mixer REST hooks are handled by defaultApp like another streams REST hooks.

Known issues

1. A mixer is not created is the name of the mixer contains symbols restricted for URI.

Symptoms: a mixer with the name like test_mixer does not create.

Solution: do not use disallowed symbols in the name of a mixer or a stream, especially if automatic mixer creation option is enabled. For instance, the name

user_1#my_room

cannot be used.
If streams of chat rooms are mixed, room names also cannot use restricted symbols.

2. Mixer output stream will be empty if transcoding is enabled on server on demand only.

Symptoms: video streams mixer created successfully, but black screen is played in mixer output stream.

Solution: for stream mixer to work transcoding should be enabled on server with the following parameter in flashphoner.properties file

streaming_video_decoder_fast_start=true

3. The same stream cannot be added to two or more mixers if non-realtime mixer is used

Symptoms: when the stream is added to second mixer, the first mixer output stream stops

Solution: do not use the same stream in more than one mixer