July 26th, 2012, 10:41 PM
Here's a great video from the BBC highlighting their 'new' DIRAC codec - from 1.5Gbps to 8Mbps:
The DIRAC Codec (BBC Promo) - YouTube
Here are some advantages as claimed by them:
Visually lossless at 8Mbps
Can be used on existing networks
Saves them $20 million a year
Intermediate codec for films, and archival
No licensing/royalties fees for software or hardware - fully free for commercial use
Certified by SMPTE VC-2
Is going to the codec for UHDTV (8K), developed by NHK, Japan
July 26th, 2012, 11:36 PM
Thanks for the video. I never knew this thing existed and even now why it's not widely adopted.
July 27th, 2012, 12:47 AM
I suspect one reason is that current codecs are earning money rather nicely for their commercial owners through license payments. Also, consumers can be extremely brand loyal and unless their favoured brand uses it, they won't.
July 27th, 2012, 04:04 AM
And I suspect that there is a lot of 'not invented here' both inside and outside the UK.
The primary consumer video compression standards have always been set by Hollywood approved distribution standards, e.g. MPEG1 for VCD, MPEG2 for DVD and DVB, and now the bandwagon has rolled on to give us MPEG4 and all its derivatives, H264, AVCHD, etc. for the various HD channels. With Sony and Panasonic having feet in both media distribution and equipment camps, it is inevitable that Blu-Ray and camcorders would go that way.
DIRAC has been around a few years now, and as the video says, the BBC have more than recouped their development costs. Its a shame that it isn't accepted much elsewhere unlike Sound-in-Sync, NICAM, teletext, and many other creations.
July 29th, 2012, 06:16 PM
Pretty cool stuff and thanks for posting, I've never heard of it before.
July 29th, 2012, 09:03 PM
Very interesting. However it looks like this video is probably about 4 years old. I googled and read where the BBC did use it internally for the 2008 Olympics and that it did indeed get approved as VC-2 format.
However after that there just does not seem to be much of anything about it. I am rather surprised.
July 30th, 2012, 12:27 AM
However after that there just does not seem to be much of anything about it.
The codec did not catch on because there is no low-power encoding and decoding hardware and because there is no high-quality psychovisually-optimized software encoder. Since a wavelet based codec with motion estimation should theoretically outperform DCT based codecs, then Dirac or something similar to Dirac may eventually become widespread. However, the reference implementation of Dirac is not as efficient as the x264 implementation of H264. Moreover, there does not appear to be any activity focused on creating a Dirac encoder of similar quality.
The BBC intentionally did not patent Dirac. It is plausible that the existence of patent free codecs such as Dirac, Theora and VP8 are the main reasons H264 licensing for internet use is still so liberal. Therefore, even if Dirac doesn't catch on, it has already served a very useful purpose.