Modeling Time-Based Media
Modeling Issues
- Differences between traditional data modeling and time-based data modeling:
- Data Flow
- Timing
- Temporal Composition
- Synchronization
Support for a BLOB (Binary Large Object) data type has been added to most commercial database systems. Although a BLOB does allow the user to reference large data objects, a BLOB does not provide enough support for complex data types.
Key Problem: The description of the structure of time-based media in a form appropriate for querying, updating, and presentation.
- Interpretation
- Traditional data modeling does not consider the interpretation of the data; mainly because the data is of a simple type.
- Heterogeneity
- Different media types may be contained within a single storage unit.
- Interleaving
- Elements of different media types may be interleaved in a single storage unit.
- Padding
- Unused (or blank) data may be added to account for differences in transfer rates.
- Out-of-order Elements
- Key elements may be placed ahead of elements that precede it in time.
- Scalability
- Only the necessary parts of the data need to be sent to the client. Example, scaling the video stream to a lower resolution before sending it.
- Quality Factors
- The encoding scheme used determines the quality of the stored information. Some encoding schemes, such as JPEG and MPEG, are lossy.
- The low level parameters should not be viewable due to the specific nature of the parameters. Only generic quality factors should be viewable.
- Timing
- There are two timing issues: real-time constraints (display or playback rates) and temporal correlations (relationships between media objects).
- Real-time constraints are concerned with the sampling rates of record and playback.
- Temporal correlations are concerned with the timing between media objects, such as synchronization.
- Derivation
- Derivation is the construction of a media object from another (possible more than one) media object. All that is needed to construct the derived object is the non-derived objects and various parameters needed in the transformation.
- An example is a "wipe" video transition where one video scene is wiped away to reveal the following scene. The "wipe" video transition can be constructed from the first and second video sequences.
- The derived object may or may not be stored in the database depending upon the available space and the resources required to construct the derived object.
- Composition
- Composition refers to the ability to combine two or more "simpler" media types to construct a more complicated media object. This is usually in the form of spatial or temporal relationships.