WebRTC - Change device/camera in realtime - camera

I'm having a problem trying to change my camera in real time, It works for the local video, but the remote person cannot see the new camera, and still sees the old one. I tried to stop the stream and init again but still not working. This is just some of my code.
I have searched everywhere and I can't find a solution. Can someone help me out?
function init() {
getUserMedia(constraints, connect, fail);
}
$(".webcam-devices").on('change', function() {
var deviceID = this.value;
constraints.video = {
optional: [{
sourceId: deviceID
}]
};
stream.getTracks().forEach(function (track) { track.stop(); });
init();
});

You need to actually change the track you're sending in the PeerConnection. In Firefox, you can use RTPSender.replaceTrack(new_track); to change without renegotiation (this is being added to the spec now). Otherwise, you need to add the new stream/track to the RTCPeerConnection, and remove the old one, and then process the onnegotiationneeded event and renegotatiate
See one of #jib's fiddles: Jib's replaceTrack() fiddle:
function flip() {
flipped = 1 - flipped;
return pc1.getSenders()[0].replaceTrack(streams[flipped].getVideoTracks()[0])
.then(() => log("Flip! (notice change in dimensions & framerate!)"))
.catch(failed);
}

Related

MediaRecorder has a delay of multiple seconda

I'm trying to use a MediaRecorder to record a MediaStream and display it in a video element using a MediaSource. So the setup looks like:
Request a MediaStream from the browser
Add it to the MediaRecorder
Add the recorded blobs to the MediaSource Buffer
The result looks very good but there is one problem: There is a delay in the playback.
When displaying the MediaStream directly there is no delay so I sorted out the first bulletpoint as the problem.
Nevertheless, it seems like either the MediaRecorder or the MediaSource is adding a delay of about 3 seconds to the stream.
this.screenRecording = await mediaDevices.getDisplayMedia({ video: { frameRate: 60, resizeMode: 'none' } });
const mediaRecorder = new MediaRecorder(this.screenRecording);
mediaRecorder.ondataavailable = async (event: any) => {
if (this.screenReceiving.readyState === 'open') {
if (this.screenReceivingBuffer == null) {
this.screenReceivingBuffer = this.screenReceiving.addSourceBuffer('video/webm;codecs=vp8');
}
if (!this.screenReceivingBuffer.updating) {
this.screenReceivingBuffer.appendBuffer(await new Response(event.data).arrayBuffer());
}
}
};
mediaRecorder.start(16);
The above code is only copy & paste from the actual project so please don't expect it to work by copy & paste ;)
Does anyone have an idea why this delay exists?
Any ideas on how to tweak the browser to not add this delay?

Using video.js is it possible to get current HLS timestamp?

I have an application which is embedding a live stream in it. To cater for delays I'd like to know what is the current timestamp of the stream and compare it with the time on the server.
What I have tested up till now is checking the difference between the buffered time of the video with the current time of the video:
player.bufferedEnd() - player.currentTime()
However I'd like to compare the time with the server instead and to do so I need to get the timestamp of the last requested .ts file.
So, my question is using video.js, is there some sort of hook to get the timestamp of the last requested .ts file?
Video.js version: 7.4.1
I had managed to solve this issue, however please bear with me I don't remember where I had found the documentation for this bit of code.
In my case I was working in an Angular application, I had a video component responsible for loading a live stream with the use of video.js. Anyway let's see some code...
Video initialisation
private videoInit() {
this.player = videojs('video', {
aspectRatio: this.videoStream.aspectRatio,
controls: true,
autoplay: false,
muted: true,
html5: {
hls: {
overrideNative: true
}
}
});
this.player.src({
src: '://some-stream-url.com',
type: 'application/x-mpegURL'
});
// on video play callback
this.player.on('play', () => {
this.saveHlsObject();
});
}
Save HLS Object
private saveHlsObject() {
if (this.player !== undefined) {
this.playerHls = (this.player.tech() as any).hls;
// get and syncing server time...
// make some request to get server time...
// then calculate difference...
this.diff = serverTime.getTime() - this.getVideoTime().getTime();
}
}
Get Timestamp of Video Segment
// access the player's playlists, get the last segment and extract time
// in my case URI of segments were for example: 1590763989033.ts
private getVideoTime(): Date {
const targetMedia = this.playerHls.playlists.media();
const lastSegment = targetMedia.segments[0];
const uri: string = lastSegment.uri;
const segmentTimestamp: number = +uri.substring(0, uri.length - 3);
return new Date(segmentTimestamp);
}
So above the main point is the getVideoTime function. The time of a segment can be found in the segment URI, so that function extracts the time from the segment URI and then converts it to a Date object. Now to be honest, I don't know if this URI format is something that's a standard for HLS or something that was set for the particular stream I was connecting to. Hope this helps, and sorry I don't have any more specific information!

WebRTC: Switch from Video Sharing to Screen sharing during call

Initially, I had two different webpages:
One was to do Video Call and
Other was to do Screen Sharing
Now, I want to do both of them in one page.
Here is the scenario:
During Live call, a user wants to stop sharing his/her video and start sharing screen.
Afterwards, again he/she wishes to turn off screen sharing and start video sharing.
For clarity, here are some questions I want to ask:
On Caller Side:
1) How can I change my local stream from video to screen and vice versa?
2) Once it is done, how can I assign it to the local video element?
On Callee Side:
1) How do I handle if the current stream I am receiving is changed from video to screen?
2) How do I handle if the stream I am receiving has stopped? I mean, now I can receive neither video nor screen (just audio)
Kindly, help me in this regards. If there are any open source codes available, kindly share their links too.
Just for your reference, I was trying to handle it using following code. (i know this is naive and won't work)
function handleUserMedia(newStream){
var localvideo = document.getElementById("localvideo");
localvideo.src = URL.createObjectURL(newStream);
localStream = newStream;
sendMessage('got user media');
if (isInitiator) {
maybeStart();
}
}
function handleUserMediaError(error){
console.log(error);
}
var video_constraints = {video: true, audio: true};
var screen_constraints = {video: { mandatory: { chromeMediaSource: 'screen' } }};
getUserMedia(video_constraints, handleUserMedia, handleUserMediaError);
//getUserMedia(screen_constraints, handleUserMedia, handleUserMediaError);
$scope.btnLabel = 'Share Screen';
$scope.toggleSelected = function () {
$scope.selected = !$scope.selected;
if($scope.selected)
{
getUserMedia(screen_constraints, handleUserMedia, handleUserMediaError);
$scope.btnLabel = 'Share Video';
}
else
{
getUserMedia(video_constraints, handleUserMedia, handleUserMediaError);
$scope.btnLabel = 'Share Screen';
}
};
Check this demo:
https://www.webrtc-experiment.com/demos/switch-streams.html
and the relevant tutorial:
https://www.webrtc-experiment.com/docs/how-to-switch-streams.html
simply renegotiate peer connections on both users' side!

WinJS Listview shows undefined when navigating quickly

I have a WinJS application with listviews in which if quickly navigate between pages before the listview is fully loaded, the next page shows the listview with all elements in it bound as "undefined".
So say I have a hub page with a "to do" that is filtered to only show 6 items, and there is a header that navigates to the full "to do" page, when the hub page is displayed but before it is fully loaded I click on the header link to the "to do" page, the app then goes to the "to do" page, but the items show up with all the properties in the tile as "undefined".
I am using IndexedDB as my data store.
My home page code looks like this:
WinJS.UI.Pages.define("/pages/home/home.html", {
ready: function (element, options) {
WinJS.Utilities.query("a").listen("click", function (e) {
e.preventDefault();
WinJS.Navigation.navigate(e.currentTarget.href);
}, false);
viewModel = new HomeViewModel(element);
viewModel.load(); //loads from indexed db
},
//etc...
To Do Page:
WinJS.UI.Pages.define("/pages/ToDo/ToDo.html", {
ready: function (element, options) {
viewModel = new ToDoViewModel(element);
viewModel.load();
},
etc//
I know there isn't much to go off, but any ideas would be appreciated.
Also tips on how to debug something like this would be great.
Update
I narrowed it down to this one line from the Hub Page:
myLib.GetData(todaysDate, function (result) {
that.trendsModel.today = result;
WinJS.Binding.processAll(that.el.querySelector("#dataPanel"), that.trendsModel); //<--Right Here
});
If I remove that, then when I load the second page the data doesn't show as undefined. What is interesting is the data initially shows correctly on the second page and then it changes to "undefined".
Solution
My fix:
myLib.GetData(todaysDate, function (result) {
var element = that.el.querySelector("#dataPanel");
that.trendsModel.today = result;
if(element) {
WinJS.Binding.processAll(element, that.trendsModel);
}
});
At the point when when the callback returns, I am already on the second page. So the selector was not found returning null. If you pass null to processAll it tries to bind the whole page which is why I was able to see the correct data for a second then it changes to undefined...Wow, what a doozy. I guess it makes sense but what a pain to find.
Hope it helps someone in the future :)
Your ToDoViewModel, and HomeViewModel need to be observable. This means they need to mix in from WinJS.Binding.mixin, and for the properties that you pull in asynchronously, they need to call this.notify("propertyName", newVal, oldVal) from the property setter.
Note that you need to have getter/setter properties. e.g.
var bindingBase = WinJS.Class.mix(function() {}, WinJS.Binding.mixin);
WinJS.Namespace.define("YourNamespace", {
ToDoViewModel: WinJS.Class.derive(bindingBase, function constructor() {
}, {
_titleStorage: "",
title: {
get: function() { return this._titleStorage; },
set: function(newValue) {
if(newValue === this._titleStorage) {
return;
}
var old = this._titleStorage;
this._titleStorage = newValue;
this.notify("title", newValue, old);
}
}
}),
});
myLib.GetData(todaysDate, function (result) {
var element = that.el.querySelector("#dataPanel");
that.trendsModel.today = result;
if(element) {
WinJS.Binding.processAll(element, that.trendsModel);
}
});
At the point when when the callback returns, I am already on the second page. So the selector was not found returning null. If you pass null to processAll it tries to bind the whole page which is why I was able to see the correct data for a second then it change to undefined...Wow, what doozy. I guess it makes sense but what a pain to find.

Detecting browser print event

Is it possible to detect when a user is printing something from their browser?
To complicate matters, if we are presenting a user with a PDF document in a new window is it possible to detect the printing of that document ( assuming the user prints it from the browser window)?
The closest I've been able to find is if we implement custom print functionality (something like this) and track when that is invoked
I'm primarily interested in a solution that works for internet explorer (6 or later)
You can now detect a print request in IE 5+, Firefox 6+, Chrome 9+, and Safari 5+ using the following technique:
(function() {
var beforePrint = function() {
console.log('Functionality to run before printing.');
};
var afterPrint = function() {
console.log('Functionality to run after printing');
};
if (window.matchMedia) {
var mediaQueryList = window.matchMedia('print');
mediaQueryList.addListener(function(mql) {
if (mql.matches) {
beforePrint();
} else {
afterPrint();
}
});
}
window.onbeforeprint = beforePrint;
window.onafterprint = afterPrint;
}());
I go into more detail into what this is doing and what it can be used for at http://tjvantoll.com/2012/06/15/detecting-print-requests-with-javascript/.
For Internet Exploder, there are the events window.onbeforeprint and window.onafterprint but they don't work with any other browser and as a result they are usually useless.
They seem to work exactly the same for some reason, both executing their event handlers before the printing window opens.
But in case you want it anyway despite these caveats, here's an example:
window.onbeforeprint = function() {
alert("Printing shall commence!");
}
For anyone reading this on 2020.
The addListener function is mostly deprecated in favor of addEventListener except for Safari:
if (window.matchMedia) {
const media = window.matchMedia("print");
const myFunc = mediaQueryList => {
if (mediaQueryList.matches) {
doStuff();
}
};
try {
media.addEventListener("change", myFunc);
} catch (error) {
try {
media.addListener(myFunc);
} catch (error) {
console.debug('Error', error)
}
}
}
Reference: This other S.O question
If it's only for tracking purposes, perhaps you could set a background url in CSS print media to a server page (.aspx, .php, etc) and then do something on the server?
This guy claims it works.
This is not as versitile as TJ's solution, but it may be less buggy (see TJs blog post for issues he found) when only tracking is needed.