Canon adopts MPEG-2 Full HD (4:2:2) file-based recording codec for upcoming cam

4:2:2 50Mbps long GOP MPEG-2 should pretty much smoke AVC-I
Sorry, no -- and please stop making uninformed grand sweeping statements like this.

AVC-I and 50mbps long-GOP MPEG-2 have already been tested extensively by the BBC, with AVC-I being chosen as the primary codec for their Digital Media Initiative. Especially because it's 10-bit, vs. the 8-bit of MPEG-2.
 
XDCAM and XDCAM EX are already production standards, acceptable by networks.
By some networks, yes -- and rejected by others. The BBC, for example, does not consider either flavor of XDCAM to count as "HD". It does, however, consider MPEG-2 at 50mbps to count as HD.
 
I doubt it. it will perform perfectly under norml conditions but not on all situations. AVCI will kick its mpeg2 ass on straining situations :D

If you only shoot non-stop car chases, maybe, just maybe, 50Mbps AVC-I will yield slightly better images. For well over 95% of typical footage, I'd bank on Canon's 50Mbps 4:2:2 MPEG-2 encoder yielding better images where it really counts.
 
Sorry, no -- and please stop making uninformed grand sweeping statements like this.

AVC-I and 50mbps long-GOP MPEG-2 have already been tested extensively by the BBC, with AVC-I being chosen as the primary codec for their Digital Media Initiative. Especially because it's 10-bit, vs. the 8-bit of MPEG-2.

Uninformed? I'm sure there are folks that are more informed, but sorry, I didn't just fall off a turnip truck when it comes to understanding video compression.

The 50Mbps flavor of AVC-I tests better than 50Mbps 4:2:2 long GOP MPEG-2??? I'd like to read that white paper!

10-bit precision would be the only distinct advantage - and that is 10-bits at 4:2:0, not 4:2:2. Intraframe only compression would only be an advantage for quite unusually high motion sequences.
 
EBU tested long-GOP and intra and the verdict was that either of 100 Mbit/s intra or 50 Mbit/s long-GOP were acceptable. But AFAIK they tested DVCPRO HD, not AVC-I as an intra codec.
 
Uninformed? I'm sure there are folks that are more informed, but sorry, I didn't just fall off a turnip truck when it comes to understanding video compression.

The 50Mbps flavor of AVC-I tests better than 50Mbps 4:2:2 long GOP MPEG-2??? I'd like to read that white paper!

10-bit precision would be the only distinct advantage - and that is 10-bits at 4:2:0, not 4:2:2. Intraframe only compression would only be an advantage for quite unusually high motion sequences.
Where did 50mbps AVC-I enter the equation? You said AVC-I, not 50mbps AVC-I.

Seems like you were assuming people would know you were talking about 50mbps AVC-I, and I was assuming that AVC-I means 100mbps because -- heck, everyone, everywhere, at all times, uses AVC-I at 100mbps whenever they can. 50mbps is for news or desperation cases.

And yes, I agree that 50mbps MPEG-2 would deliver a better image than AVCI-50.

And "uninformed" was aimed at the specific situation of the BBC having tested both codecs and choosing AVC-I.
 
This is what I was responding to (and exactly how I quoted it in my response, aside from bolding the key portion here):

...with 50 Mbit/s and 4:2:2 color it will be a good alternative to AVCHD and to 50 Mbit/s AVC-I. Bravo, Canon...

Perhaps I could have been more clear in my response.

I sort of wonder how well the image quality from Canon's new 50Mbps 4:2:2 MPEG-2 encoder will compare with the image quality from encoding with Panasonic's 100Mbps flavor of AVC-I - while perhaps not quite as good, I bet it's pretty close (probably the sort of thing you really have to blow up bigtime to see any difference at all).
 
I sort of wonder how well the image quality from Canon's new 50Mbps 4:2:2 MPEG-2 encoder will compare with the image quality from encoding with Panasonic's 100Mbps flavor of AVC-I - while perhaps not quite as good, I bet it's pretty close (probably the sort of thing you really have to blow up bigtime to see any difference at all).
This is what Sony had to say in December 2008:

An equivalent picture quality to an MPEG-2 Long GoP compression of full 108050i/60i 422 signals at 50 Mbps could be matched by an AVC Long GoP scheme operating at rates between 32-36 Mbps (depending on the complexity of picture content).
...
The picture quality level of MPEG-2 Long GoP at 50 Mbps and AVCI at 100 Mbps with 4:2:2 HD signals are comparable.

And this is what Sony said on 1080-line 50p/60p:

The capabilities of the bit-rate reduction techniques designed for the ‘sweet spot’ of MPEG-2 have been
exceeded and will make its use for compression of 1080 50p/60p, most likely, unacceptable. A solution would be to increase the compressed data rates to 100 Mbps of higher in order to maintain reasonable compression ratios.

An alternative is to use the more efficient coding scheme of AVC Long GoP. Instead of using MPEG-2 Long GoP for
the compression of 1080 50p/60p at rates approaching 100 Mbps, one could use AVC Long GoP with all of its advanced data reduction techniques to produced robust, compressed video streams at 50-60 Mbps for 1080 50p/60p content.

Sony is developing a highly complex IC which will implement AVC Long & Intra GoP to be deployed in future products, handling 1080 50p/60p signals for a variety of mastering, and …eventually mainstream HD production applications.

I suppose, we should expect something like AVCHD422 with 32 Mbit/s and 4:2:2 color sampling as a cost-effective solution for film-like frame rates and 720-line 50p/60p, and a 60-something Mbit/s variant for 1080-line 50p/60p in the near future.
 
Last edited:
Sony is being pretty clever with their NX5U memory rebate offer. It's valid until the end of March . . . right before NAB, when Canon is likely to unveil details.
 
It's my understanding that Canon is going to be offering more detail, at least about the new codec, in a few days in SF.
 
Guys, you´re only talking about the codec here? I just can´t stop thinking of how the new "file-based professional video camera" will look like. I´m pretty sure it won´t be the 5DXLH2 a lot of us would love to see - but what about taking the game to 2/3 shouldermount? Is it in times of HDSLRs and tapless workhorses like the HPX170 or EX1 really likely to just be a codec upgrade on the XL/A1 series?
I´m just saying canons current HDV cameras are as good as it gets, the new codec would be beneficial in post, but would it otherwise challenge the curent lineup of cameras? I got high hopes the new codec will be the backend of some serious glass and a brandnew sensor/dsp combo...
 
Guys, you´re only talking about the codec here? I just can´t stop thinking of how the new "file-based professional video camera" will look like. I´m pretty sure it won´t be the 5DXLH2 a lot of us would love to see - but what about taking the game to 2/3 shouldermount? Is it in times of HDSLRs and tapless workhorses like the HPX170 or EX1 really likely to just be a codec upgrade on the XL/A1 series?
I´m just saying canons current HDV cameras are as good as it gets, the new codec would be beneficial in post, but would it otherwise challenge the curent lineup of cameras? I got high hopes the new codecwill be the backend of some serious glass and a brandnew sensor/dsp combo...

The codec is indeed the subject of the thread (perhaps started by mere coincidence, shortly after Canon made the announcement about it), so it does follow that it is what is being discussed!
 
Just got this email for the Ninth Annual San Francisco SuperMeet:

Canon USA is going to make the first US public announcement that they are going to introduce a new Canon MPEG-2 codec and workflow to the industry at the San Francisco SuperMeet. They are also going to have the first US public preview of their new tapeless HD video camera at the San Francisco SuperMeet. So be the first on your block to see this new camera as the SuperMeet will be the first place in the USA to see it.

Exciting :)
 
Back
Top