EBU HD acquisition codec tests

[David Heath]

Some interesting accounts of new generation codec (DNxHD, JPEG2000, AVC-Intra, and XDCAM-HD422) tests by the EBU at EBU Technical Review , (click on "HDTV production codec tests")

If you're looking for "A is better than B" type conclusions, you'll be disappointed, but it may be worth quoting from the EBU conclusions:

For acquisition applications an HDTV format with 4:2:2 sampling, no further horizontal or vertical sub-sampling should be applied. The 8-bit bit-depth is sufficient for mainstream programmes, but 10-bit bit-depth is preferred for high-end acquisition. .......All tested codecs have shown quasitransparent quality up to at least 4 to 5 multi-generations,.........

The EBU recommends in document R124-2008 that:
�� If the production/archiving format is to be based on I-frames only, the bitrate should not be less than 100 Mbit/s.
�� If the production/archiving format is to be based on long-GoP MPEG-2, the bitrate should not be less than 50 Mbit/s.

[Alister Chapman]

An interesting document. In parts of the full report there are some interesting arguments against 10 bit compressed codecs is and the importance of avoiding scaling gets a lot of attention.

Makes me chuckle that the tests center around 1080 production when the EBU have been pushing 720P production.

[David Heath]

Originally Posted by Alister Chapman

Makes me chuckle that the tests center around 1080 production when the EBU have been pushing 720P production.

I have to disagree Alister - the EBU stance for the last few years has been in favour of progressive video, not interlace, but their big emphasis has been regarding TRANSMISSION, not production, and very clearly 720p/50 v 1080i/25.

I'm pretty sure that when this first became a hot issue (IBC 3 years ago?), they clearly stated that whilst they favoured 720p/50 for transmission, they also felt that future production should ideally be 1080p/50, when the kit became freely available.

What also came clearly across at their presentation two years ago was that they were basing their case on comparisons of how 720p/50 and 1080i/25 stood up to heavy compression for transmission. A question was asked from the floor regarding 1080p/25, and they didn't seem to have an answer, it just didn't seem to have been considered for the trials - which may be seen as surprising as so much currently produced material is precisely that.

My understanding is that broadcasters are tending to currently side with 1080i transmissions as much because they enable 1080p/25 to be transmitted in full resolution (as 1080psf/25), as for actual 1080i/25. For such as a sports only broadcaster, (where 50Hz motion is essential) when the choice really is 720p/50 or 1080i/25, the EBUs stance makes sense. For channels such as BBC-HD, with a large drama content as well as sport, a basic 1080i/25 system able to carry 1080psf/25 makes sense.

But let's look forward to 1080p/50..........

[Christopher Drews]

Did I miss something? Didn't they say which codec yielded the best results?
Or is it just this blanket statement: Use a codec above 100 megabits per second.
-C

[David Heath]

Originally Posted by Christopher Drews

Did I miss something? Didn't they say which codec yielded the best results?

No, they didn't say, and I believe quite deliberately:

The reader should be aware that, due to the complexity and framework of the tests, only a deep analysis of these documents can provide a complete appreciation of the results. ..........
It was agreed between the EBU project group and the vendors to make the reports about the test details available to EBU Members only.

I suspect it's because it's just not possible to say that "codec A yielded the best results, period" - codec A may be better on some test images, codec B on others, and codec C on a third set.

Originally Posted by Christopher Drews

Or is it just this blanket statement: Use a codec above 100 megabits per second.

No, the advice is 50Mbs and above for long-GOP, 100Mbs or above for I-frame only codecs. They also say "All tested codecs have shown quasitransparent quality up to at least 4 to 5 multi-generations..... ", which is quite a vote of confidence in all four. It should also remove any vestiges of doubt about the use of long-GOP codecs for critical work - at least if 50Mbs XDCAM-HD422 is used.

[Joe Yaggi]

Hi there,

I've been lurking but now facing the possibility of being able to upgrade with a project that may be signed this week.

So I'm wondering if any of you think there'll be a firmware upgrade for the PDW-700 in the near future to bring it into the 100 mb/s realm?

Thanks very much,

Joe

Originally Posted by David Heath

No, they didn't say, and I believe quite deliberately:

I suspect it's because it's just not possible to say that "codec A yielded the best results, period" - codec A may be better on some test images, codec B on others, and codec C on a third set.

No, the advice is 50Mbs and above for long-GOP, 100Mbs or above for I-frame only codecs. They also say "All tested codecs have shown quasitransparent quality up to at least 4 to 5 multi-generations..... ", which is quite a vote of confidence in all four. It should also remove any vestiges of doubt about the use of long-GOP codecs for critical work - at least if 50Mbs XDCAM-HD422 is used.

[Perrone Ford]

And people thought I was nuts for mastering to DNxHD 175x/220x or 145 for 8bit. Confirms my findings that the codec was essentially lossless, for as many generations as my projects get rendered. Not bad for free.

For those of you with FCP, Edius, and CS3/4, do you have a JPEG2k codec installed stock? I know I can't export in AVC-I, or XDCam-HD from my NLE, but I do have JPEG2000.

[Tripp Woelfel]

Originally Posted by Perrone Ford

And people thought I was nuts...

Never! (grin)

Seriously, I stumbled across your thread about the Avid codec which prompted me to do some investigation last night about it as a post and mastering format. At first look, after only two cups o' joe, it appears to open up some new horizons. Those of us on PCs not using Cineform were stuffed when it came to archiving HD masters. This could all work out quite well. I'm going to do some testing with it today.

Thanks for doing the research on this. It might just be the first step in bringing me over to your camp of going tapeless.