Choosing a Digital Video File Type

Last updated: 03 March 2009
Published in: Digitising analogue media | Creating new digital media
Tags: business & community engagement | codec | digital collections | digital preservation | digitisation | e-learning | file formats | video

Comments (2)

Summary

Choosing a file format can prove overwhelming for someone new to the world of digital video. The aims of this document are to discuss some of the key factors that should be considered before selecting a file format and to suggest suitable choices for specific applications

Introduction

A digital video file has consists of a ‘wrapper’ and a ‘codec’. The wrapper can be thought of as an envelope which contains elements such as video, audio and metadata. The video portion of this data usually encoded by a specific ‘codec’ (COmpression-DECompression algorithm). A codec can either be ‘lossy’- information is discarded in order to reduce file size and/or bandwidth or ‘lossless’- no information is irreversibly discarded. Common wrapper file formats are .avi (Audio Video Interleave), .wmv (Windows Media Video) or QuickTime (.mov) while common codecs include MPEG-4 ASP and Sorenson. Just to make things a little more complicated, not all wrappers are compatible with all codecs.

Both wrappers and codecs have been expanded and improved upon as digital video has replaced analogue video. Different software players and editors prefer different types of file, depending partly on proprietary alliances. The type of work being carried out by the software (for instance editing or restoration) also influences the choice of file format which is commonly used.

This document aims to offer a basic guide to the most suitable file types, for the purposes of:

• Capturing existing (analogue) video
• Archiving video files
• Creating new video content (with a video camera)
• Delivering digital video

For a more comprehensive introduction to data compression and the nature of digital media files, see the document File Formats and Compression.

Open vs. proprietary formats

An ‘open’ technology is not supported by a sole manufacturer but is instead supported by a community of users. A ‘proprietary’ technology is the property of a manufacturer or group of manufacturers and licences are typically put in place to restrict the way the technology can be legally used. Also, manufacturers are under no obligation to support a given technology in perpetuity and can withdraw it at any time. Any given video wrapper or video codec will be either open or proprietary in nature.

The need to use an open format will vary on the type of project being undertaken. Any digitisation project concerned with the long-term accessibility should consider a format which is both of suitable quality and non-proprietary.

A project more focused on the immediate delivery of digital video, with no strong interests in preservation, may not need to consider the long-term implications of using a proprietary format. However, even in these circumstances, open standard formats may provide greater accessibility for users, and it is therefore recommended to use open formats whenever practicable.

Realistically, a single project will often employ a number of different video file types, some open and some proprietary, each suited to different aims and objectives.

Capture file formats

Digital ‘capture’ refers to the process of importing (non-digital) video material into a digital format. On a desktop computer system this will be achieved via an analogue-to-digital converter such as a video capture card. Major capture cards will only capture analogue video in a small number of (usually) proprietary formats.

Similarly, ‘born digital’ video, whether on a digital videocassette or a hard drive, will already have an innate file format. This too, may be proprietary in nature.

Any digital video material held in a proprietary file format can be converted to an open format later, although attention should be paid to ensure the new open format is of equal or higher specifications to the existing proprietary one, if all data is to be retained.

Format requirements

Retains as much useful information as possible from the original material

Digitising analogue video means making considerable and inevitable changes to video materials. Consider the differences between a VHS tape and a DVD-video for instance. Before digitising, it will be helpful to decide on the ‘significant properties’ of the materials in a collection and ensure that they are preserved. Is colour important? Is sound required? What about interactivity? JISC have produced a useful report that suggests what these properties might be for moving image materials to be used in teaching learning and research.

Does not use ‘lossy’ compression

Quite simply, a video that is compressed with a lossy codec cannot regain its lost data. If image quality is amongst the significant properties you have defined, avoid lossy codecs. This problem is compounded if one compressed format is transcoded to another.

For delivery purposes, a copy of a high quality video file can be compressed later, using a lossy codec, to make the smaller files more suited to delivery. As lost data cannot be retrieved without re-capturing, the lossy/lossless dilemma must be decided upon early on in the digitisation process.

Suggested formats

Capture format choice is limited by the small number of formats any given hardware capture card can support, this may well be a single proprietary format which is closely associated with the manufacturer of the capture hardware. For mid-range desktop systems consider the losslessly compressed Blackmagic codec in an .avi or .mov wrapper. AJA are another popular capture card manufacturer and their losslessly compressed AJA codec in a .mov wrapper is another good choice for affordable desktop machines.

SAMMA machines are high-end, dedicated digitisation stations which can capture analogue material in JPEG2000 format in an MXF (Media Exchange Format) wrapper. Although not widely adopted this format has the advantage of being an open format, unlike the Blackmagic or AJA codecs which are proprietary. JPEG2000 is also capable of lossless compression.

Archive preservation file formats

The archive preservation file should be as true a copy of the source material as possible. It is important to remember that when video is captured at the highest available quality, the maximum amount of information is retained; even damage and distortion are forms of information.

If the video material does not look as good as you had hoped there may be a need for some restoration or re-mastering but original ‘un-restored’ copies should be retained. This ensures subjective improvements are not irreversible.

Format requirements

The requirements for an archival file format share the two requirements described in file formats for capture (above), it should also:

Be an open format, if practicable

Proprietary formats should not be used as an archival format, as there is no guarantee the format will not be made obsolete in the future. However, unlike other types of digital media, a consensus has not been reached on just how this requirement will be achieved for digital video with its large files sizes.

Suggested formats

Unlike digital audio files or digital images, there is no consensus among the archival community as to which file format (wrapper or codec) should be used for long-term storage. This situation is likely to change, but the community is understandably cautious and slow to adopt unproved technologies. In practice, many digitisation projects choose to archive files in the same (propriety) file format as they were captured in. This situation is not ideal, only a short term solution. The chosen codec (such as Blackmagic) may not be open but is lossless and is freely available from the manufacturer. The codec package will typically be archived along with the video files themselves in case it suddenly becomes unavailable. At some future point when a more ideal, non-lossy and open file format is decided upon, video files can be transcoded to this format without loss of quality.

At present the prime contender for a widely accepted archival format seems to be JPEG2000, a sill image codec which can be used to encode each frame of a video file as separate images. Putting these images back together into video is a complex task and the MXF wrapper, which is capable of retaining complex metadata, is therefore often used in conjunction with JPEG2000.

If resources are not in place to store huge, losslessly compressed video files a popular alternative is to encode as DV within an .avi or .mov wrapper. While DV is compressed format, it has very wide support as the DV codec is used within DV and mini-DV cameras. The DV format is also widely supported by editing software packages. This is not the case with JPEG2000.

Whichever codec and wrapper format is selected, a good general rule of thumb is to quantify how much (hard drive) storage will be available and then carry out some short trial captures. This will help determine which codec represents the best compromise between storage cost and quality. However, if preservation is the central aim of the project losslessly compressed or even uncompressed (using no compression at all) is highly recommended.

Example uncompressed archival video specifications

Video component: Standard definition in PAL format
Wrapper file format: QuickTime .mov
Format/codec: UYVY
Bit depth: 10 bit
Frame size horizontal: 720 pixels
Frame size vertical: 576 pixels
Frame rate: 25 frames per second
Frame type: progressive
Field order: N/A
Frame aspect ratio: 4:3
Pixel aspect ratio: 1:1
Colour space: YCrCb
Chroma sub sampling: 4:2:2

Audio component: Uncompressed stereo audio
Compressor: uncompressed PCM
Bit depth: 16bit / 24bit
Sample rate: 48KHz
Number of channels: 2
Audio interleave: 1 sec

Resultant file size: 93 GB/hour (approx.)

File formats and the creation of new video content

File formats for recording new video material are closely associated with the type of camera used. It is unusual for any given camera to support more than a handful of file types. Before choosing a format to record in, all future plans for the footage should be defined.

You may be creating new video resources for one of many reasons: for teaching purposes, creative output or for starting or adding to an existing institutional collection. It may be the case that your recordings are only for immediate delivery to a defined audience and high quality is not important. You may not wish to archive the recordings or even keep them to be used again. In these circumstances it may be best to record at a lower quality suited to delivery.

If, on the other hand you may wish to re-purpose videos in the future, the same approach should be taken as when capturing video: use equipment which preserves as much information as possible. Lower resolution copies can still be generated for delivery and higher quality versions kept in reserve.

For these reasons it is always recommended that recordings are made at the highest possible quality. However, maximum quality often equates to larger file size and so available storage may be a defining factor.

For consumer and pro-sumer video camera equipment, only lossy recoding will be possible, typically in DV format (for standard definition) or HDV (for high definition). Production of a losslessly compressed archival version will not re-introduce quality lost at the time of recording, however it may help by moving from a proprietary format into an open one. However, this is rarely done in practice due to the extra storage space required with no perceptible increase in quality. Born digital video then, is generally archived in the original lossy recording format.

Format requirements

Is fit for purpose and of suitable quality

The purpose or purposes of the video material should be clear before recording any footage. Only then can decisions regarding image and sound quality be made.

Is compatible with editing software

This is only a requirement if post-production editing is planned.

Suggested formats

HDCAM or HDV for high definition materials, DV or Digibeta’s native DCT encoded format for standard definition material. As mentioned above, choice of camera is likely to dictate choice of file format.

File formats suited to delivery

When choosing a file format for delivery, the main factor is choosing a wrapper format and codec with the widest accessibility to your audience. You may find that you need to compromise file size and quality (through lossy compression) to effectively deliver your files. The open / propriety nature of files is not necessarily a large concern as future delivery files can be generated from high quality archival copies as new delivery formats become popular.

File requirements

Compatibility

The file types need to be compatible with the playback facilities of your users.

Is of suitable file size and bit rate

Whether the video material is to delivered on an optical disc or the Internet, maximum file size and bit rate restrictions will apply.

Is fit for purpose and of suitable quality

The final delivered video files will be a compromise between size and quality, but no important visual and auditory information should be sacrificed.

Suggested formats for Internet delivery

When video files are hosted online, they are often ‘streamed’ (a copy of the file is not saved as it plays). Alternatively they may be simply downloaded (a copy of the video is made on the computer).

Streaming files are accessed through web browsers, and where proprietary file formats are used a third party browser plug-in may be required to play the file.

There are various file formats that can be streamed but two of the most common two, both often used to deliver relatively low quality video, are the proprietary formats Flash video and Windows Media Video.

Suggested formats for delivery on mobile devices

There is an increasing need for video, and digital media in general, for playback on portable media devices such as mobile phones or a dedicated media players. These files are usually downloaded from the Internet first and so considerations for Internet delivery apply here too. File formats chosen are the same as for Internet delivery with the additional dominance of MPEG-4 AVC based files.

Suggested formats for delivery on optical media

Video files can be stored on optical discs, usually as either data files or in the DVD-video format.

If video is stored as simple video files on a CD or DVD, the maximum storage space and the maximum playback speed will be the only influences upon deciding a file type. If real time playback is not required (files will only be played back after being copied to computer’s hard drive) only available optical disc space need be taken into account.

DVD-video, on the other hand is a tightly regulated specification that can be read by both set-top DVD players and accessed via computer drives. DVD-video authoring packages are the best choice for making such tightly specified data structures. Video content will have to be in MPEG-2 (very occasionally MPEG-1) format, which is highly compressed to meet relatively slow maximum read speeds.

Conclusion

The correct file type for the job can be discovered by trying out different formats: from creation to end-use. If video materials are to be used solely within a single collection, this process may be enough to determine which type of file to choose. If however, videos are to be used across collections, or if future plans might make this desirable, then some attention should be paid to interoperability and different types of standards drawn upon. Creating files which are high quality, popular and in open formats can save a lot of work later on.

JISC Digital Media offers advice on creating effective and sustainable digital media resources.

Last updated: 03 March 2009
Published in: Digitising analogue media | Creating new digital media
Tags: business & community engagement | codec | digital collections | digital preservation | digitisation | e-learning | file formats | video

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Comments (2)