IETF Begins Standardization Process For Next-Generation 'NETVC' Video Codec (Daala)

[Guest]

IETF begins work on the standardization of the next-generation Daala codec, which will now be called NETVC.

IETF Begins Standardization Process For Next-Generation 'NETVC' Video Codec (Daala) : Read more

 

[zanny]

Just some nomenclature clarification: h264 is a patent encumbered protocol, it has many implementations. x264, the most popular implementation, is also free software - it just had to be developed outside the jurisdiction of US patent law, and US vendors cannot distribute their own compiled binaries of it Its fuzzy.

Same with x265. None of these video codecs are proprietary vs open the way Libre Office vs MS Office is. Even something like WMV has draft standard document implementations to write encoding / decoding software for them.

The only reason there can even exist a "patent encumbered video codec" is due to the continued existence of software patents in the US and pretty much no where else.

 

[m0r0nn]

IETF can't take over anything; better description wold be "Daala developers offer their codec as a base for collaborative development at IETF".

 

[photonboy]

I'm curious if EVOLUTIONARY ALGORITHMS have been utilized at all in codec creation.

By that i mean setting up the algorithm and have it create the codec within changeable parameters such as bandwidth, processing requirement etc.

 

[bit_user]

Just some nomenclature clarification: h264 is a patent encumbered protocol, it has many implementations. x264, the most popular implementation, is also free software - it just had to be developed outside the jurisdiction of US patent law, and US vendors cannot distribute their own compiled binaries of it Its fuzzy.

Your statement that you can't redistribute x264 is incorrect. If you pay them royalties, they'll grant you a proprietary-compatible license. I think they pass on the MPEG-LA's portion of the royalties, although perhaps licensees just pay them directly.

My company sells commercial products which include x264, and our lawyers make sure we strictly adhere to all open source licenses, patents, etc.

 

[bit_user]

I'm curious if EVOLUTIONARY ALGORITHMS have been utilized at all in codec creation.

By that i mean setting up the algorithm and have it create the codec within changeable parameters such as bandwidth, processing requirement etc.

I'm not sure exactly what you mean, but most codecs already have parameters for controlling those sorts of things. Read up on "Profiles and Levels" and you'll get a sense for how they do it.

However, if you mean the codec is actually a virtual machine, and the encoder generates a custom decoder along with the stream for it to decode, this is the same idea as Postscript, which is probably about 35 years old. The problem with using this for video compression is that it's hard to constrain the performance of the decoder so that it will run in realtime, on any client that supports it.

It's worth noting that one could already do something like that with Javascript.

 

[bit_user]

BTW, I don't believe any such thing as a patent-free video codec is possible. The genius of MPEG is that they established a process (the Licensing Authority) whereby everyone who has an IP stake in the technology is incentivized to get involved and negotiate generally reasonable fees that don't unduly hinder adoption.

And I'd rather deal with MPEG-LA, in whose self-interest it is that the technology is adopted and not tied to one vendor or service, than to hope that someone like Google remains benevolent towards users of VP8 & VP9.

 

[avada]

Just some nomenclature clarification: h264 is a patent encumbered protocol, it has many implementations. x264, the most popular implementation, is also free software - it just had to be developed outside the jurisdiction of US patent law, and US vendors cannot distribute their own compiled binaries of it Its fuzzy.

Same with x265. None of these video codecs are proprietary vs open the way Libre Office vs MS Office is. Even something like WMV has draft standard document implementations to write encoding / decoding software for them.

The only reason there can even exist a "patent encumbered video codec" is due to the continued existence of software patents in the US and pretty much no where else.

On the other hand these "uncencumbered" formats are retarded by design. They avoid many straightforward or efficient algorithms, because they're patented.

And yes, f*ck you US!

 

[avada]

I'm curious if EVOLUTIONARY ALGORITHMS have been utilized at all in codec creation.

By that i mean setting up the algorithm and have it create the codec within changeable parameters such as bandwidth, processing requirement etc.

I highly doubt it. It seems like to me that when it comes to codec/format creation, people lack imagination and and willingness to explore.
They just tread towards the path already taken. The only particularly significant change with dalaa is that they're using overlapping bocks instead of discrete blocks.