CODEWITHSUNDEEP
manifestmpeg-dashadaptive-bitrate
user2948033
user2948033

Reputation: 113

MPEG-DASH - How is the content separated in AdaptationSet?

I'm trying to understand the MPD format. I read the information about AdaptationSet in the standard, but I don't understand well how the separation of the Representations into different AdaptationSets is made.

The standard points out:

Representations are arranged into Adaptation Sets according to the media content component properties of the media content components present in the Representations, namely

the language as described by the @lang attribute, 
the media component type described by the @contentType attribute, 
the picture aspect ratio as described by the @par attribute, 
the role property as described by the Role elements, 
the accessibility property as described by the Accessibility elements, 
the viewpoint property as described by the Viewpoint elements, 
the rating property as described by the Rating elements.*

For example: http://www-itec.aau.at/~cmueller/libdashtest/showcases/redbull_segment_template.mpd

There are five AdaptationSets, I think the media component type (contentType) is video for all them, although I don't understand the value of the attribute contentType="RedBull 320x240", contentType="RedBull 480x360", etc. and the picture aspect ratio is equal for the first two adaptationSet and it is equal for the last three AdaptationSet. And for other mentioned properties(role, viewpoint, etc.), no information is provided.

In another example: http://ftp-itec.uni-klu.ac.at/dash/conformance/mp4-full-gdr/mp4-full-gdr-mpd-AV-BS.mpd, the value of codecs is different for each representation into the same AdaptationSet. Therefore, this attribute doesn't affect to the separation in different AdaptationSet.

Any help?

Thank you!

Upvotes: 3

Views: 2736

Answers (1)

Alexander Wolf
Alexander Wolf

Reputation: 604

I will try to make it clear based on an example: imagine a stream with audio and video. The video has two different quality levels (lets say 500kbps at 480p and 2Mbps at 1080p), we call them Representations. Audio is further divided into two different languages (English and Spanish), both languages only come with one quality level/representation.

To describe this in an MPD you would need three AdaptationSets - one for Video and two for Audio, as you have two languages which you can select, but only one video. The video AdaptationSet would contain two Representation elements, which can be switched smoothly, as they show the same content (only at different quality levels). For the audio you need different AdaptationSets (which can be differentiated based on the lang attribute), because the audio track for English and Spanish is not the same. In case you only have one quality level for each audio track you would only have one Representation element inside the AdaptationSet.

I hope this makes things more clear for you - you may want to have also a look in different MPDs, which you can find for instance at dashjs or create some content (and MPDs) on your own with the help of MP4Box or more convenient with portal.bitcodin.com/portal.

Upvotes: 3

Related Questions