Re: [MSE] Ad Insertion (was: [MSE] Timestamp offset mechanism)

Duncan,

Because the required behaviour for ad insertion is so dependent on business rules, we should be thinking in terms of providing applications with pretty low level tools that they can use to implement the required behaviour. Otherwise we will inadvertently leave some use-case unsupported.

I particularly think we should avoid having the Media Element manage multiple timelines - for example a discontinuous 'main media' timeline overlaid on a 'real time' timeline which also includes ads.

Despite the overlap rules, I think it's reasonable that if you are going to do ad insertion with a single track, then it happens at segment boundaries. In this case, ad insertion must happen immediately before a random access point, because you are using a single decoder and it needs to start at a RAP after the advert. The content must be encoded with ad insertion in mind and precise knowledge of the ad insertion points. The app must know exactly where these points are. Given this it's no big deal to put a segment boundary there.

The Media Source extension with the timing offset feature supports the use-cases you described, with the assumptions above, with one small exception.

I can place adverts into the video elements timeline. I can render my own controls which show only the main media timeline if I wish and map this to the media element timeline in my app. I can fetch new adverts to replace the previous one when the user seeks, all according to my business rules.

The only issue I see is that when replacing a part of the media element timeline with new material, I have no way to 'purge' what was there before. If the network stalls then playback could spin straight off the end of the 'new' material and into the 'old' material.

For more advanced scenarios - for example where ad insertion happens somewhere else than an I-Frame I think  you do need separate tracks or even separate elements and frame-accurate switching.

Btw, why do you BBC guys care about ads?? ;-)

...Mark

Sent from my iPhone

On Jul 11, 2012, at 8:06 AM, "Duncan Rowden" <Duncan.Rowden@bbc.co.uk<mailto:Duncan.Rowden@bbc.co.uk>> wrote:

Hi Aaron,

Thanks for your comments.

Your assumptions on my first comment are correct, I believe a frame accurate pause and track switch would be very useful. This was what I was going for on my follow-up email when I suggested using a TextTrackCue to signal the switch: http://lists.w3.org/Archives/Public/public-html-media/2012Jun/0109.html/. Kevin S and Steven R mentioned on this point that you could get a non-deterministic delay if you use this method, but I don’t think this is a problem as long as the content is paused frame-accurately. A viewer won’t notice a delay if the switch happens in less than 0.5s, but maybe this non-deterministic delay is much bigger than this? If this is the case, then as you stated this may be something for the HTML-WG to look at?

To answer your question on seeking back to an advert, this should be dictated through business rules. A Content Provider may wish to only show adverts at a given point in the timeline once. If you skip back to the ad-break point or before and press play again you won’t see any adverts at that ad-break point. It’s also conceivable though that you’re always shown the same ad’s at a given ad-break or that new ones are acquired each time.
On a side note: From personal experience I never really want to watch an advert again, but if the content provider wished to do so they could make references to viewed adverts available within their application.

With regards to the overlap rules, maybe I have misunderstood them. If for example you have a segment that is placed on the time-line at t=100s -> t=110s and you want insert a 10s ad at t=105s, how would you achieve this? From my understanding of the rules the last 5 seconds of the segment would be overwritten by the ad and you’d somehow have to acquire those last 5 seconds of that segment again and add them to the timeline.

In conclusion it appears to me that we are not being as comprehensive in covering as many use-cases as we could by adding the complexity of the timestamp offset mechanism.

Regards
Duncan

From: Aaron Colwell [mailto:acolwell@google.com]
Sent: 10 July 2012 19:15
To: Duncan Rowden
Cc: public-html-media@w3.org<mailto:public-html-media@w3.org>
Subject: [MSE] Ad Insertion (was: [MSE] Timestamp offset mechanism)


Hi Duncan,

Sorry for the huge delay in my response. I was solely focusing on the object oriented API spec changes for a while. I'll be more responsive in the future. :)

Comments inline...
On Mon, Jun 18, 2012 at 4:25 AM, Duncan Rowden <Duncan.Rowden@bbc.co.uk<mailto:Duncan.Rowden@bbc.co.uk>> wrote:
Hi Aaron,

I agree that the use-cases you state are important and need to be catered for, but I have some reservations about the suggested API’s.


-          Ideally, when inserting an advert, this won’t contaminate the main media item’s timeline. The reasoning for this is that when ads are inserted, a content provider may only wish these to be viewed once. So when rewinding back through a given point either another ad can be inserted or playback can occur without showing an ad.
[acolwell] These seem like policy decisions best handled by the web application instead of the browser itself. I don't quite understand how the ad won't contaminate the media timeline since at a minimum it would increase the duration of playback by the duration of the ad. I also don't quite understand how seeking is supposed to work. What if I decide I want to seek back into the ad after it has completed and the main presentation has resumed? Are you saying this wouldn't be allowed or could cause me to see a different ad? All this sounds to me like you need a different feature than what I am proposing here. It sounds like you want some sort of way
to trigger a frame accurate track switch or pause.

-          The point at which ads need to be inserted may not necessarily coincide with a segment boundary, which as I understand it would be required using this proposal.
[acolwell] The ad does not need to be appended at a segment boundary. The standard overlap rules<http://dvcs.w3.org/hg/html-media/raw-file/tip/media-source/media-source.html#source-buffer-overlapping-segments> would apply which should allow one to place the ad anywhere in the timeline. Care would have to be taken though if any end overlaps exist because they may cause a gap in the timeline.

As an alternative suggestion, I’d recommend keeping the separate media items on their own timeline and define a way to switch at particular points between the items?

[acolwell] I think your best option would be to propose a frame accurate track switch feature to the HTML-WG since this doesn't sound like a MediaSource specific feature. If you had this then you could simply use two source buffers (1 for media and 1 for ad) for this use case and then use the track switch feature to switch between the tracks associated with these buffers. You'd still need the timestamp offset mechanism though so you could place the ad in the proper location in the timeline, but using 2 source buffers would allow you to avoid the ad clobbering the main presentation media.

Aaron


Regards
Duncan

From: Aaron Colwell [mailto:acolwell@google.com<mailto:acolwell@google.com>]
Sent: 15 June 2012 23:41
To: public-html-media@w3.org<mailto:public-html-media@w3.org>
Subject: [MSE] Timestamp offset mechanism

Hi,

Bug 17004<https://www.w3.org/Bugs/Public/show_bug.cgi?id=17004> proposes a mechanism for adjusting timestamps inside media segments. This enables seamlessly splicing content together that doesn't originally exist on a common presentation timeline. Inserting advertisements into content is the most straight forward use case. There has been some useful discussion in the bug comments, but we haven't decided on which method signature to use for specifying the timestamp adjustments.

Here are the proposed signatures at this point:

void sourceTimestampMapping(in double presentationTimestamp, in double segmentTimestamp)
 -- or --
void sourceTimestampOffset(in double timestampOffset)

sourceTimestampMapping() just provides a way to say, "I want this timestamp in the segment to map to this presentation time."
sourceTimestampOffset() is basically just saying, "Add this offset to any timestamp in the segment to get the desired presentation time."

Fundamentally they do the exact same thing, but I'm looking for input on which option people feel more comfortable with.

Which do you prefer?
Are you ok with the semantics outlined in the bug description<https://www.w3.org/Bugs/Public/show_bug.cgi?id=17004#c0>?

Aaron



http://www.bbc.co.uk
This e-mail (and any attachments) is confidential and may contain personal views which are not the views of the BBC unless specifically stated.
If you have received it in error, please delete it from your system.
Do not use, copy or disclose the information in any way nor act in reliance on it and notify the sender immediately.
Please note that the BBC monitors e-mails sent or received.
Further communication will signify your consent to this.



http://www.bbc.co.uk
This e-mail (and any attachments) is confidential and may contain personal views which are not the views of the BBC unless specifically stated.
If you have received it in error, please delete it from your system.
Do not use, copy or disclose the information in any way nor act in reliance on it and notify the sender immediately.
Please note that the BBC monitors e-mails sent or received.
Further communication will signify your consent to this.