When to normalize born digital video for preservation and when to keep heterogeneous original file formats

With the long term in mind, (50-100 years) what situations is it good to normalize video for preservation to particular formats that have less preservation risks and when is it better to just stick with the originals?

Trevor Owens

• file-formats

Comments

Answer by wizzard0

It is hard to predict which formats will stick, but there are general rules to consider:

1. The format should be simple (i.e. do not require setting up complex streaming servers etc). For example, MJPEG is simpler than MPEG which is simpler than full-blown DVD container with menus etc.
2. The format should not require cumbersome licensing or obtaining patent rights to decoding/playing.
3. Source code for the decoders should be available, otherwise we risk decoders dying together with companies that maintain them
4. Since different codecs introduce different kinds of artifacts, give plenty of headroom when selecting bit rate. If possible, use PSNR or some other metrics to check re-encoded video for artifacts. If the video can be encoded to lossless formats like huffyuv with enough compression (e.g. screencasts or cartoons), use them.
5. Keep the originals too.
6. If possible, keep 2 formats - one with inter-frame compression (e.g. H264) and one with intra-frame only (e.g. MJPEG). Use the latter for creating new re-encodes when needed.

Comments

• kvanmalssen: Keep in mind that an audiovisual file is a complex digital object, with multiple "formats" to consider. There is the container format, then the encoding formats of the video, (optional) audio and other tracks. It is important to look files in their entirety when thinking about migration. While migration will likely have to occur frequently for access purposes, depending on your users needs, a high quality preservation master whose file and encoding formats score high on the Sustainability of Digital Formats criteria should have staying power.