OpenMPEG: a project to create GPLed MPEG4 implementation

The MPEG4IP project is more focused on streaming than codecs, but maybe there are some possibilities for code reuse/sharing there.

Unfortunately, I predict that the MPEG-4 patent holders will have the same response to this as Fraunhofer did to LAME, and the result will be the same: Code only available from non-US sites, not bundled with any distributions, etc.

I'm really happy to see this. An open source implementation of MPEG-4 will be an valuable reference and research tool, as lame has been for audio encoding. The paucity of open mpeg software has been worrying; I hope this project succeeds.

We need better mpeg-1/2 encoders as well.

For those not aware, there is another project working on an open implementation, started by the people behind the "DiVX ;-)" hacked binary codec. Their license is non-free, iirc because they tried to make some of the patent and use control issues explicit, but they do develop in the open and have working code available. And their intentions are good.

Unfortunately, MPEG-4 is heavily patented and so while these projects might be libre in many parts of the world they will never be gratuit.

I mean in no way to discourage people from participating in the OpenMPEG effort, but I did want to point out another way forward. To create something the whole world can use we need an open standard that avoids any patent encumberence. That means developing it ourselves.

Perhaps the most famous effort in this direction is Ullrich Hafner's FIASCO codec based on weighted finite automata. It works, but is designed for low-bitrate applications, and isn't really a replacement for mpeg. (The quality of the examples is similar to realvideo, albeit as a lower bitrate.) To my knowledge no one's done a patent search of the algorithms involved.

One of the most promising efforts in the quest for a free codec (IMHO of course) centers around the Xiphophorus group. Their work so far has concentrated on the Vorbis perceptual audio codec, but a complementary video codec ("Tarkin") has always been a part of the plan. Discussions are just beginning, but there is now some sample code, e.g. here and here. These two are 3D (or 2+1D) wavelet codecs, which seems the most promising direction to go in, though other approaches are being considered and there's by no means consensus on that. But the most important thing is that it's a group of very smart developers who've demonstrated a willingness to wade through the patent literature and work work around it as necessary.

If you're interested in video, I suggest you check out all of these alternatives.

I definitely think that while open implementations of codecs is good and useful, I don't think that it is a long-term solution. I am a firm believer in what the Xiphorous people are trying to do: develop free replacements for media formats that are patent-free. This is the only way to really gain freedom in the long run. Hopefully, these formats will be so good that they can challenge the proprietary formats for market acceptance. One thing that is on their side (and I think Xiphorous is doing a good job of pointing out) is that companies that want to develop products and services around multimedia formats save a great deal of money by not having to pay for licenses. Hopefully this will become a compelling argument as patent holders become increasingly greedy. But we should focus on the fact that we need a fully free computing infrastructure - that has to include file formats, not just protocols and OSes.

I definitely think that while open implementations of codecs is good and useful, I don't think that it is a long-term solution. I am a firm believer in what the Xiphorous people are trying to do: develop free replacements for media formats that are patent-free. This is the only way to really gain freedom in the long run. Hopefully, these formats will be so good that they can challenge the proprietary formats for market acceptance. One thing that is on their side (and I think Xiphorous is doing a good job of pointing out) is that companies that want to develop products and services around multimedia formats save a great deal of money by not having to pay for licenses. Hopefully this will become a compelling argument as patent holders become increasingly greedy. But we should focus on the fact that we need a fully free computing infrastructure - that has to include file formats, not just protocols and OSes.

If you really fear that Microsoft will get a monopoly on this considering a BSD style license might be a better way to get sponsors and companies to commit people to distribute code to this project.

Potentially LGPL could work as well, but having a GPL codec seems like it only would be incorporated by GPL'd players, encoders and such. While putting under the BSD style license a lot more companies would be willing to put it in their products and makes it much more wide spread and provide alternatives on all platforms.