Windows Media Photo

Windows Media Photo was announced last week. I'm convinced that a new proprietary file format is a really awful idea.

As a prelude, let's review what has happened with video formats:
  • Patent holders for MPEG-2 and MPEG-4 formed MPEG-LA, which is an organization to aggregate fees to the multitudes of patent-holders for these technologies.
  • For an example of MPEG-4 licensing terms, see this PDF. You can pay up to $1M a year for your organization to provide MPEG-4 content or encoders/decoders to customers.
  • After a huge amount of effort and marketing and licensing and legal wrangling, all this hassle means that MPEG-4 is not really a practical format for personal content. Yes, there's Quicktime, DivX, and XViD (which is open-source), but generally you can't expect the recipient of a video to be able to decode it more than 1/3 the time. This is the reality of sharing MPEG-4 today.
  • Of course, applications where there's service-level vertical integration are able to use MPEG-4 technologies today: satellite, iTunes Video, etc.
  • Microsoft was able to slip WMV into this market (because of this stupendous lack of a great open standard), and people generally thought it was fine. At least WMV delivery works some of the time now.
But I'm just getting warmed formats are easy. They only require that software is popular at a given time. Flash is a delivery format, and it works great.

Archival is harder. Archival means that a format will be available when you want to see it in the future. This might be 80 years from now, or it might be two years in the future. The dominant platform might be a bio-computer running a new OS that has nothing at all to do with Windows.

So let's review what happened with JPEG:
  • Tom Lane's Independent JPEG Group and their free implementation made JPEG a de-facto standard. I understand the standards process, but it is not responsible for the success of a format. I believe it is almost solely Tom's implementation that has made JPEG successful.
  • The IJG JPEG implementation compiles to about 30KB of code, so it can be distributed in almost any embedded device or application, and it's fast, so it can decode and encode quickly almost anywhere.
But despite these differences, video (and audio) and photos are apples and oranges today.

Why? MPEG-4 is a delivery format, and it's not designed as an archival format at all. The licensing terms are per-year, and a majority of the licensing language concerns broadcasters. The situation is similar with audio formats - who cares how compressed your audio is, when some audio engineer has an uncompressed AIFF or WAV file of the data back in the studio?

The same rules do not apply (and they must not apply) to people's personal photos. If you want to be able to view your photos in digital form in 80 years, the only reasonable way to do it is to use a format that is open, with freely available implementations, and without silly licensing terms.

I'm a commercial software developer, and you won't hear me say that often. But in this case, formats really have to be kept alive, and the stakes are too high otherwise.

Don't go encoding your photos in a proprietary image format, and especially don't archive them that way. You will very simply regret it in the future.

1 comment: