Video encoding for archival

8

2

I would like to archive some home videos (DV). I don't need to save them losslessly, but I would like to encode them in something high-quality.

What format is both pretty indistinguishable from the original and will likely be readable 15 years from now?

WMA makes me nervous, because it's only one company that makes it, and they're constantly coming out with newer formats. (VLC couldn't open my WMAs that Windows Movie Maker made.)

Other things I've considered are h.264, Ogg Theora, DivX, and Xvid.

I don't mind paying for something, but usually that means the format is owned by only one vendor.

Jim Hunziker

Posted 2009-08-13T18:57:07.590

Reputation: 1 069

You probably realise this but you also need a policy to put the video onto appropriate media, keep backups (where needed and, though this shouldn't apply to home movies, allowed under licences) and to refresh copies and backups from time to time (including moving to current media before you don't have reader for the old media) – mas – 2009-08-13T19:25:10.317

Yup. My intention is to keep the original tapes off-site and keep the encoded videos on two hard drives in my home. If I'm disciplined, I'll check md5sums of the videos. – Jim Hunziker – 2009-08-13T19:27:43.867

"will likely be readable 15 years from now?"

15 years is a LONG time in the computer world. I mean really -- have you looked at the video codecs people were using in 1994? MPEG-2 wasn't even invented then. ;-) – Dan Esparza – 2009-08-14T00:05:44.853

@Dan: But they were using Cinepack, and you can still play Cinepack. – derobert – 2009-08-14T07:04:44.940

Answers

11

H.264 is the current state-of-the-art codec, will give you the best quality/size and will likely be in use for some time.

Ogg Theora is not as powerful as H.264, but is completely open and doesn't have the patent questions that H.264 has (link). Theora is also being used by Firefox for the HTML5 element, so it will probably stick around for a while.

Dirac is another codec you might consider. Like Theora it is an open standard, but aims to have quality comparable to H.264. It is being actively developed by the BBC, but is not yet in widespread use.

There's no one best answer, but for good quality and future use I would choose one of these three. At this point I would only use XviD if I needed to play the video on legacy devices.

dave

Posted 2009-08-13T18:57:07.590

Reputation: 641

the main factor for quality is "bitrate" anyway, and not "fileformat". the more bits you use to store the orginal bits the better. the more you throw away the less you can get later on, for whatever reasons you would may need them. – akira – 2010-05-12T14:15:21.147

@akira - within a given encoding format, bitrate does determine quality. But different codecs do more or less with a given bitrate, so you can't ignore encoding format when you need quality. Older codecs didn't have the CPU to play with that more modern ones have, and so had to use more bits per quality than a more modern codec. – Michael Kohne – 2011-12-29T14:08:31.557

4

I'd keep it in the original format as extracted from the camera. DV format is an IEC standard, and likely to be readable for quite some time to come. It's a pervasive format (it's use extends into industry, not just the home camcorder market), and thus it's likely to be something that can be extracted many years later.

If really want to convert now (to reduce storage requirements), then I'd say mpeg 4 might be a good choice (with your compression settings turned to 'minimum compression', of course). The reason I say that is that mpeg-4 is a VERY widely adopted standard, again making it likely to be readable for many years into the future.

Another good choice for likely future readability (though not necessarily for video quality or compression) is to store the video as a series of DVDs (playable on a television). The large number of DVDs in existence today makes it likely that DVD reading will be popular for many years into the future.

For absolute maximum future-proofing (while not going so far as to store the original computer and software!), I'd say you should store as many conversions of the video as you can, right along side the original video. This maximizes the chance that one of the formats you've chosen will survive the test of time.

You are also going to have to be careful to keep up with your media and make sure the thing you've store it all on isn't going bad, but you have to do that no matter what the format of the video.

Michael Kohne

Posted 2009-08-13T18:57:07.590

Reputation: 3 808

I'm not storing anything on DVDs because the shelf life of DVD-Rs is very short - often as little as 4 or 5 years. – Jim Hunziker – 2009-08-13T20:02:36.577

Agreed, I was more considering format than lifetime of the media. I don't know of any media currently available that I'd trust for more than a few years. HDs might do OK sitting on the shelf, but they might not. CDs are too damn small. SSD drives using USB might not be bad, if you thing the flash memories they are based on will last. – Michael Kohne – 2009-08-14T02:58:03.407

1

I wouldn't worry about the format too much. If all else fails, mplayer and VLC will be around in 15 years (being a very prevalent open source projects)

Obviously don't use an obscure, application specific format.. H.264, Theora, MPEG-4, XviD etc should be fine. You can always reencode the data should a better suited codec appear in the next decade or two..

The thing to worry about is the data storage, if your storage dies, or the files become corrupted, the codec is irrelevant. Every few years you should really transfer the files to new storage (which will be come easier as drives get bigger/cheaper), when you do that, check the files are still playable with current software/codecs.

If you're really paranoid, you could store a copy of the transcoding software (say mplayer/mencoder), and an operating system disc (a linux distribution?) - then if somehow h.264 becomes impossible to play, you create a virtual machine, install or compile the transcoding application and convert the files to something playable.. You can still emulate Windows 3.1 in VMWare, so you should be able to emulate the current Linux distros in a few decades (assuming no nuclear apocalypses)

dbr

Posted 2009-08-13T18:57:07.590

Reputation: 4 987