Video Archival Methods

Accession

The Video Archival process begins with Accession of a set of media when it arrives in the Macaulay Library. In this case let's say it is 12 HDCam tapes from Hawaii. A record for each tapes is entered into an Accessions database, all 12 are entered as a single Accession. The recordist, locality, and information about the contents are stored here and a barcode label is affixed to each tape.

Duplication

Next each Hawaii tape is duplicated for two important reasons:
(1) Security. A safety copy is needed in case anything were to happen to the original, and to assure that wear on the original is kept to a minimum. The original is eventually stored in a secure off-site location.
(2) Time code breaks. An important problem involving time code is solved in our duplication process. The time code of a tape, represented as a string of numbers like this, 02:24:15:29, (hours:minutes:seconds:frames) is very important in our archival process. The original time code is carried through the whole process all the way to digitization and storage. This makes it easy to link each clip back to the original source in exact fashion. The problem: sometimes there are breaks in this time code that show up on the original tapes for unexplainable reasons. Solution: Through a sophisticated process, time code breaks are found and clean time code is produced on the duplicate.

Logging

A video archivist begins by loading a duplicate Hawaii tape into the deck and beginning a project in a video production program called Final Cut Pro. The goal of the logging process is to organize the tape into individual clips, determining the start (in-point) and end (out-point) of each clip, organizing into species and behavior specific clips. Final Cut Pro organizes each individual clip created, numbering them sequentially, Iiwi_1, Akiapolaau_2, Iiwi_3. When the entire tape has been logged, an EDL (Edit Decision List) is exported from Final Cut Pro. This is an important document that represents all of the in and out points determined during the logging process.

View a detailed description of the Tape Logging workflow.

Digitization

Encoding
The next stage is carried out on a different computer called the VideoBank Encoder. Digitization begins here with the creation of a large Mpeg2 file. The first step is digitization of the entire tape. The Optibase Mpeg encoder card in the computer takes control of the Video deck via a serial digital interface connection, reads the video signal and creates the large Mpeg2 in real time. Next, the EDL is imported and applied to the Mpeg2 file. The in and out points for each clip determined during logging are applied and each clip appears as an independent asset.

Transcoding

The final stage takes place on a VideoBank Transcoder. This machine carries each asset through the final stage in which medium resolution (Mpeg1) and low resolution (Mpeg4) versions of the same clip are created.

Annotation

Metadata for each clip are entered by the archivist and stored in an Oracle database that handles all media asset information, including data on the recordings, their contents, conditions under which they were made, location data coupling this information with the Accession data enteted upon arrival in ML. The same database handles behavioral analyses, systematic information, and habitat descriptions. The data-model is very powerful, holding nearly 600 attributes in 71 related tables. Twenty-two of these tables, for example, are used just to describe behavior with sufficient flexibility to allow for their functional and descriptive categorization while accommodating discordant opinions on those categorizations.

Storage

Each clip from the first Hawaii tape is then represented by a high resolution Mpeg2, an Mpeg1 which can be downloaded for use in presentations and an Mpeg4 for future streaming on the Web. All assets are stored on a 25 TB Apple Xserve/Xserve RAID array, where they are readily available on an internal network, and will soon be available to the external world on the Web.