:warning: This file is part of the "archive" – it could contain statements or terms which are incompatible with the main body of the specification and the identity and timing model. :warning:

Notes on TVIs

TVIs are deliberately defined in quite abstract terms. A TVI can be thought of as simply a mechanism for identifying a number of TDSs that are conveying the same "message".

Given relatively few constraints are applied to TVIs, there is the prospect for a number of 'grey areas' in what can be viewed to represent the same TVI. This is intended to allow the user to decide the most appropriate degree of 'similarity' between TDSs which are permitted to be identified as being approximations of the same TVI. In a practical and interoperable implementation it is likely that some additional constraints will need to be applied to how TVIs are used in order for them to be useful.

In practical systems it is normally useful to define a TVI for each "element" of content (such as a TVI for each of the audio and the video that form a programme). These "elements" are chosen to be at a practically "useful" level of granularity (such as defining a TVI for the multi-channel audio created for a programme). However, these "elements" can always be decomposed as required (such as breaking down the multi-channel programme audio into its constituent audio channels, with one TVI defined for each audio channel – note that decomposing a TVI / extracting a subset of the content logically results in different TVIs).

It would not normally be practical to define an entire "scene" (e.g. in a TV studio) as a TVI. For most practical purposes this is not "granular" enough, and establishing which items of content relate to this TVI would be very difficult to automate. Other mechanisms should be used to describe "scenes" and other high-level concepts such as actors, objects in the scene, etc.