MPTF/MPTF Discussions/ViewPort-Support

From Web and TV IG
Jump to: navigation, search

Submitter: BBC

Tracker Issue ID: ISSUE-37

Description:

The HTML5 <video> tag should allow a sub-region of decoded video to be displayed. This sub-region is also known as a View-Port.

  1. It should be possible to change the View-Port during playback without interrupting the media presentation, in particular the audio.
  2. This use-case is taking connected CE devices with limited processor power into consideration.

Motivation:

Broadcasters sometimes transmit Mosaic style pictures, from which a single element (i.e. a portion of this picture) can be scaled up to fill the whole viewing region.

Using W3C media fragments or drawing mechanisms such as the <canvas> tag are currently not adequate to meet the requirements of this use-case.

Media Fragments do allow a view-port to be defined, but the drawbacks are:

  1. A change in the media fragment parameters will in the best case scenario result in a delay to the media pipeline while the switching occurs
  2. The Media Fragments specification currently only envisages cropping rather than up-scaling of a selected region (See relevant section in Media Fragments Spec)

The drawbacks to using mechanisms that manipulate the <video> tag, such as the <canvas> tag are:

  1. Accessing potentially DRM protected pixels within the <video> tag
  2. An inherent computational overhead is required even if the mechanism can be hardware accelerated.
  3. No mechanism to time sync the tags, which can lead to aliasing
  4. Cross-origin-restrictions can prevent <canvas> tags being accessed by JavaScript from other sources

Dependencies:

  1. Possibly Media Fragments


What needs to be standardized:

  1. A mechanism to define the horizontal and vertical offset of the View-Port in pixels within the <video> tag
  2. A method for defining the width and height of the View-Port in pixels within the <video> tag
  3. A switching mechanism that ensures the video pipeline is not affected