Formats and codecs for digital video preservation (Guideline 22)
Some formats that are used for preservation include:
|A multimedia file format developed by Microsoft and IBM. A derivative of the RIFF format. It is well understood by many editing and transcoding systems. This is now virtually obsolete and has limited metadata capacity but is widely supported.|
|An object based file format developed by the Society of Motion Picture and Television Engineers (SMPTE) that wraps video, audio and other bitstreams. The MXF standard has been increasingly adopted in broadcasting. See the European Broadcasting Union website (http://www.ebu.ch/) for technical reviews.|
|A multimedia file format developed by Apple Computer Inc that wraps audio, video, effect or text. It is well understood by many editing and transcoding systems. Note: While widely supported with good metadata capacity, this is proprietary.|
|An object oriented file wrapper developed by the Motion Pictures Experts Group (MPEG). It specifies the use of JPEG 2000 for timed sequences of images (motion sequences) possibly combined with audio.|
|A still image codec created by the Joint Photographic Experts Group in 2000 and published as an ISO standard, ISO/IEC 15444. JPEG2000 has been increasingly adopted in moving pictures sectors.|
|An open, royalty free video compression format, specification and system developed by the BBC.|
FF video codec 1 (FFV1)
|A lossless, intra-frame video format. The encoder/decoder are part of the free, open-source library libavcodec in the project FFmpeg.|
Only linear uncompressed codecs or mathematically reversible lossless codecs should be used for preservation. 
For more information on formats, including their sustainability, you could start by examining their profiles on the Library of Congress Digital formats website or their individual descriptions on Wikipedia. See also the JISC fact sheet Choosing a digital video file type.
The National Film and Sound Archive, Australia has announced that it has also purchased SAMMA digitisation systems to use JPEG 2000 frame images, losslessly compressed with an MXF wrapper for the digitisation of more than 40,000 hours of video. See Case studies of digitisation of analogue audio and video for more information.
Note: The above solutions are for large scale projects. At present it is not easy to use standard video workstations for the efficient preservation of digital video content. For example, the National Film and Sound Archive advise that almost all workstations on the market do not provide capture to lossless JPEG2000 and can only provide archival acceptable formats that are uncompressed. If organisations have a small collection of valuable video material, they may need to contact an outsourced provider to digitise them. Appendix 2: Outsourcing the digitisation of audio or video includes a list of considerations when using outsourced providers.
When creating digital video or digitising analogue formats, it is generally recommended that:
- the audio tracks of digital video recordings should be at a minimum of 48KHz (the broadcast sample rate is always 48 kHz and 16, 20 or 24 bits depending on the age of the format and equipment)
- a minimum of 720x486 pixels at 25 frames per second is achieved
- the colour depth match the number of colours as well as the colour encoding and luminance of the original material. Black and white originals should be recorded in grayscale; colour originals should be recorded in full-range colour (8 bits per channel with 10 bits per channel preferred).
Formats that do not have sufficient quality for archival retention
While they may be useful for short term records, or to meet particular needs, the following digital audio files do not have sufficient quality for preservation purposes (although they may be used for short term records or copies if required):
- Files created for streaming broadcast (e.g. RealAudio, Windows Media) as they are too compressed and sacrifice quality
- Reference files of lower quality than the original (e.g. for use on the website) as they are generally too compressed and sacrifice quality
- Files that have been through two or more codecs. 
Quick tips regarding digital file formats
Choose a format that is suitable to the project aims and the longevity of the recording.
If formats are required in the long term or as archives:
- Use the best quality format you can afford for creation and/or digitisation.
- Use no compression (or lossless compression if compression is unavoidable) for creating a digital preservation master. Lossy compression irretrievably reduces quality.
- Use an open format or a format which has good ‘sustainability factors’ for the preservation master. If you can’t create in such a format, copy to one when possible.
- Define your metadata requirements carefully and ensure metadata is collected from creation.
- If enhancements are to be made, make these from a copy. Save a ‘master’ of the original and enhanced version.
- Create copies for delivery/access in formats that meet your needs (e.g. lower resolution formats).
- If possible, store a copy of your preservation master (and enhanced master, if any) at a different location for disaster management purposes.
- Regularly monitor your masters for obsolescence or degradation.
 Information derived from Wikipedia and the Library of Congress Digital formats website at http://www.digitalpreservation.gov/formats/
 NFSA, Op.cit.
 Library of Congress, Sustainability of digital formats: Planning for Library of Congress collections, Material eXchange Format (MXF), available at: http://www.digitalpreservation.gov/formats/fdd/fdd000013.shtml
 NARA, Op.cit. In general component colour is used Y/U/V. There are many standards for encoding YUV 4:4:4, 4:2:2, 4:2:0 and all affect the quality. 4:4:4 is the ultimate, but as there is too much data in a 4:4:4 encoding, most archives use 4:2:2.
 NARA, Ibid.