Canon's new 50Mbps MPEG-2 Full HD (4:2:2) codec

[Robert M Wright]

I dislike the term "prosumer" as well - especially the way it gets used to diminish some nice tools, simply because they are not high end tools, sort of akin to looking down on a Chevy merely because it's not a Cadillac.

To my mind, cameras like the XH-A1/G1 and XL-H1 are bona fide "professional" cameras - primarily functionally designed to be used for professional purposes, and indeed they generally are used as such. Those cameras aren't VariCams, but that sure as heck doesn't make them toys.

In turn, to my mind, the term "prosumer" would be far more appropriate to use in describing a camera more along the lines of an HD1000U or an HMC70. It's (more than) a bit of a stretch to claim that they were functionally designed to be used for professional purposes. It's not the engineers designing those cameras that have professional use in mind, so much as the marketing departments of the companies manufacturing them. The marketing departments of those companies have succeeded in selling those cameras to folks that indeed do use them for professional purposes, but the engineers basically just took the guts of what are very much consumer cameras and wrapped them in big shells with a few flourishes (like XLR connectors) - hence a consumer product being used for professional purposes, sort of like attaching a ball to the back of a little Hyundai and hitching a trailer to it for delivering washing machines or something.

[Tania Ratu]

Rick young from macvideo.tv has an interview with the canon rep from the supermeet.

MacVideo - The ultimate resource for video on the Mac

Ratu

[Chris Hurd]

That Canon rep has a name -- he is Joseph Bogacz of Canon USA.

[Tania Ratu]

Yes, sorry about that.
Ratu

[Craig Bellaire]

If these chips are anything like the XHG-1, then this is gonna be nice... Been nicer with 1/2 but lets see the price before we jump...

I'm shooting the XHG-1 with FFV's 100 megabit J2K, 4:2:2 through HD-SDI and I have to say this little camera and J2K codec is REAL nice...

Raise your hands in the air and say.... NO JAGGIES!!!!!!!!!!!!!!!!!

NUF Said!

[Tom Roper]

Originally Posted by Tim Polster

Tom,

Are the time savings you are speaking of with HDV and the EX codec in taking the raw footage and putting it directly to Blu-ray without editing?

This would be to watch the raw footage through a Blu-ray player and a television?

I mean WITH editing, but not with grading.

In other words, you can arrange your clips, combine them on the timeline, make cuts and splices and insertions, smart render with very little time and no quality loss, but if you make color corrections it will have to re-encode.

This can go straight to Blu-ray because XDCAM-EX 35mbps HQ 4:2:0 mpeg-2 is about ideal for the format.

50 mbps 4:2:2 is superior, but has to be resampled costing time, because the Blu-ray format doesn't support 50 mbps bitrate. By the time you've resampled it, it's gone through a generational loss (and gone to 4:2:0) that may not in the end be much better than native 35 mbps 4:2:0 would have been.

[Robert M Wright]

There's time savings with smart rendering, but MPEG-2 isn't real tough to encode with a modern CPU anymore. As far as time savings, with smart rendering HDV or XDCAM-EX, the faster CPUs get, the less real-world benefit there is (image quality degradation issues, from from an extra generation of compression, aside).

[Tom Roper]

Robert,

Fast modern CPUs will also do the job of smart rendering faster, so the real world benefit of smart rendering will remain 250-300% over re-encoding to mpeg-2 no matter what processor you use.

As an aside, if I am going to go to the bother of a re-encode in the first place, I will go the full step to AVC h.264 which is orders of magnitude slower than simply re-encoding to mpeg-2.

It's the convenience that native 35 mbps 4:2:0 from the camera can be about as good as a first generation down sample of 50 mbps 4:2:2 to something that goes on Blu-ray, while saving time. That still holds (for mpeg-2). The AVC h.264 would look better but take lots of time.

[Tim Polster]

Originally Posted by Tom Roper

I mean WITH editing, but not with grading.

In other words, you can arrange your clips, combine them on the timeline, make cuts and splices and insertions, smart render with very little time and no quality loss, but if you make color corrections it will have to re-encode.

This can go straight to Blu-ray because XDCAM-EX 35mbps HQ 4:2:0 mpeg-2 is about ideal for the format.

50 mbps 4:2:2 is superior, but has to be resampled costing time, because the Blu-ray format doesn't support 50 mbps bitrate. By the time you've resampled it, it's gone through a generational loss (and gone to 4:2:0) that may not in the end be much better than native 35 mbps 4:2:0 would have been.

Hey Tom,

I was basing my observations on my workflow. I use Edius and my workflow is the same no matter what footage is present. I get a 2:1 (2x realtime) encoding output to Mpeg-2 for Blu-ray and Edus does not need to render not matter how much color correction or pretty much most anything else you can do to the footage. This is with an Intel i7 920 processor.

I have not used any other NLE's outside of Premiere a little.

So from my point of view, I want the 4:2:2 with as high a bitrate as I want to use up card space with...

[Tom Roper]

That's fine Tim but I think you are wrong if you are saying your footage is not being re-encoded when you do color grading. And for mpeg-2, it's also being downsampled to a lower bit rate for compliancy with Blu-ray which also should include going to 4:2:0 chroma. If it didn't, then there would be even more compression of the luminance samples in order to preserve the 4:2:2 chroma resolution for a given file size.

[Tim Polster]

Hey Tom,

I am saying that putting 4:2:2 high bitrate footage into my system or 4:2:0 lower bitrate footage will yield the same output time weather I put on correction filters, sharpening etc... or not. Drop the clips on the timeline from the CF card, edit & encode.

Staying with as high of a an aquisition format/bitrate as possible does not translate to any more encoding time in post while allowing more lattitude.

If I use AVC footage I transcode to Canopus HQ which adds an entire step before editing.

Just trying to point out that my use of the Nano flash has shown me that high bitrate Mpeg-2 is not a strain on system resources.

[Tom Roper]

Tim,
I did not say anything about a strain on system resources. I did in fact mention that many people make parallel investments in computer technology to speed the process while missing the benefit of smart rendering. You actually fall into that, because your workflow describes that smart rendering, i.e. no-recompression lossless rendering is not taking place. Your i7 processor should do smart rendering in less than realtime. I use Nanoflash too BTW, It's not relevant to the point, that for speed, native XDCAM-EX or HDV ports straight to Blu-ray without conversion. No matter what processor you use, the savings in time will be 250-300 % faster than a re-encode as you are doing now. So basically what you can state, is that you can afford the rendering time with your NLE, i7 processor and workload. That's great! (But it still takes longer than staying native, by 250-300%.) If you have 6 hours of video, that could be significant, or perhaps not.

[Robert M Wright]

No disrespect intended, but if someone is running a video production business so tight that color grading (and essentially anything but digitally cutting and splicing raw footage) is skipped in post, in order to be able to gain a little speed via smart rendering, that's essentially what most folks would probably consider akin to a Jiffy-Lube style in-and-out-the-door-quick video production mill, and I've got to think that the difference in quality between HDV acquisition (much less XDCAM EX) and acquisition using 50Mbps MPEG-2 at 4:2:2 just wouldn't be a source of any concern whatsoever anyway. I mean really, it sounds a lot like a production house that might re-use tapes, shooting with b-stock HD1000Us purchased used at auction, perhaps until they literally fall apart, simply because there's nothing else cheaper that shoots HDV and doesn't obviously look like a consumer camera while holding it. Why would 50Mbps MPEG-2 and 4:2:2 color enter into consideration, even remotely for a moment, in the first place? (I've got nothing against folks running a bare-bones video mill, nor have I've got any objections to folks operating a Jiffy-Lube either, and I know it can be awfully tough for wedding shooters to turn a profit that yields a living wage.)

[Chris Hurd]

Originally Posted by Robert M Wright

No disrespect intended, but... that's essentially what most folks would probably consider akin to a Jiffy-Lube style in-and-out-the-door-quick video production mill

No disrespect back, but this is indeed critical to wedding videographers, however not the ones you're talking about. Instead, it's the lucrative higher-end wedding videographers who specialize in same-day edits and they are the perfect example of why this is in fact a very important time-saving issue. The difference in quality plus the difference in time is perfect for that particular market.

[Robert M Wright]

Perhaps I was mistaken, but I didn't get the impression that the intent of presenting smart-rendering as a time saver had anything to do with the benefit of faster turn-around doing SDEs. I was under the impression that it was being presented more or less for general labor/cost effectiveness purposes in more traditional post production.

That's an interesting point though, and made me stop and think. Are SDEs actually being burned to Blu-Ray disk, for playback nowadays? I was under the impression that typically they were played back from a laptop computer, which shouldn't really preclude the use of smart rendering with 50Mbps MPEG 2.