I think it depends which XDCAM codec. AVCHD and XDCAM EX are both 8bit 4:2:0, so I'm not sure there's much difference there. But XDCAM HD422 is 4:2:2 (also 8-bit, I think) so it would be better for grading.

That all depends. The 35 Mb/s from the Sony EX is for the picture only. Sound is extra on top of that. While I seem to recall that the 24Mb/s rating of AVCHD includes the sound datarate in that total.

Even uncompressed audio is well under 10% of total bandwidth - and compressed audio, far less.

AVC, at roughly 2/3 the bandwidth of MPEG-2, can produce roughly similar quality, if the codec is efficient. AVC is still fairly young, so some modest improvements on in-camera compression efficiency are likely over the next few years, but right now it looks like codec efficiency in prosumer level AVCHD camcorders is already quite respectable. The major camcorder manufactures have had more than a couple years of experience, producing consumer AVCHD camcorders. In this industry, that's a whale of a long time really.

To put numbers on the data rate,
48 KHz 16 bit PCM stereo sound requires 1.536 Mbps.
The compressed MP3 sound that normally piggybacks on HDV and AVCHD uses ~376 Kbps.

A lot has been said in the UK recently about the BBC HD channel and bitrates. The coding system is H264, and until a few months ago the bitrate was around 16Mbs, then it was announced (to howls of protest) that it would be reduced down to 9.7Mbs. The argument was that they had got new coders, much more efficient, and this is what allowed them to reduce the bitrate for comparable quality.

I was sceptical when I heard what was proposed, but my own subsequent experience is that what they say seems accurate - I'm noticing virtually no difference in picture quality. This is all referenced at BBC - BBC Internet Blog: The Hitchhiker's Guide to Encoding: Life, Encoders and Everything (Or a brief history of HD encoding)

I'll quote just a bit of that:
Yes - a couple of years ago even expensive broadcast real time encoders weren't that much more efficient than MPEG2. It's only relatively recently that the theoretical promises of AVC have started to be realised in practice, at least for real time encoding.

(Blu-Ray is different, when the coding is done in non-real time, and two-pass becomes possible, for camcorders you obviously need a real time encoder.)

So if that was the situation for real time broadcast a couple of years ago, and I don't find it credible that the AVC-HD encoder in a prosumer camera such as the HMC150 was substantially better.

That is not to say the situation won't change, so my feeling is that future improvements will be more than modest, but AVC-HD in cameras whose design is a year or two old is not that much better than MPEG2.

Very interesting. This stuff is complex, but it is important to stay on top of it all.

Hang in there.......you may be very surprised by the end of 2010. That's all I can say.

That sounds interesting!

It would be uncharacteristic for Canon to introduce a new model, and then top it in the same year. If there is something else coming out, it probably is different enough not to threaten sales on this prototype.

Thinking about this and reading what everyone has posted, if I were Canon I'd continue w/ the 1/3" for as long as I could get away with... maybe move up to 1/2" or at the most 2/3" in another pro video line.

And I'd let the DSLR continue to merge w/ digital cinema camera needs. But I would not create a specifically designed "Red killer" camera, as I don't want to get into a fight over such a small market.

Nothing wrong with a little dust-up between companys......As Chris pointed out, Canon will first make something that will sell to the most people and then after that, we'll see.....The guys in Japan know there is a desire for a DLSR in a video configuration....they've seen it posted here and have heard it from CanonUSA people as well as, people attending trade shows. Its just a matter of where it fits in with Japan's plans and goals.

Jim

If Canon were to take the guts of a video DSLR and redesign it as a camcorder, don't expect it to be even close to cheap. With as limited a market as the camcorder would have, Canon would have to price it pretty high to recover the product development costs.

If Canon insists on only keeping 35mm sized sensors in DSLR's then expect the video side of these cameras to NEVER get better.

It's been proven time and time again that the video processing side of these cameras really suffers and without a proper body to house larger and more robust boards then they will never be true cinema cameras and will remain a novelty amongst the low-end of filmmakers with zero budget productions.

As I've explained in my article, there are two very important reasons why this "true cinema camera with 35mm sensors" concept won't materialize anytime soon... and they are highly significant roadblocks:

The AF System (must use the phase detection process, which isn't cheap)

The Lens (must be motorized with at least an 8x to 10x zoom, which isn't cheap).

Currently there is no suitable Canon EF lens adaptable for this purpose... certainly not the EF 28-300mm f/3.5-5.6 L IS USM, which is a push-pull "slider" type zoom. The missing glass is the real reason why Canon won't go in this direction, at least not until after RED has sounded the market first (and maybe not even then). It's a very tight niche which possibly won't recover the R&D cost of the lens -- which can run into millions of dollars -- and might not represent enough of a profit margin to pursue.

The difference with RED is that their target niche isn't just one of a wide variety of corporate interests; instead its *only* interest is indeed the digital cinema market, in which Canon has only a very small vested interest relative to everything else they do. RED has tremendous financial resources readily available to pour into this one specific thing, while Canon does not, despite the fact that it too is a multi-billion-dollar company.

I would play it a bit differently. Clearly, videographers like large sensor video, and many of us buy Canon lenses for our DvSLRs. I would put the DvSLR guts in a video camera body with very few changes. Yes, it will have rolling shutter, and be difficult to focus. But it will look like a video camera.

I would add software features like controllable audio gain, zebras, etc. In other words, I'd integrate Magic Lantern-type features. I'd also articulate the LCD screen. And I'd add XLR audio inputs. About the only other change I would make would be to add a slower anti-aliasing filter for its video-centric role.

The R&D costs for such a camera would be minimal. It would be the guts of a 7D with the LCD screen and audio input of an existing camcorder. The sensor would be a 7D unit with a video AA filter applied after the fact.

It wouldn't be a RED killer. It would be for the DvSLR videophile. And it would exist to sell more lenses and get people into the Canon fold.

As Canon's DvSLR video quality improves, I'd keep leveraging the technology for this hybrid cam. Maybe someday it gets electronic T/W lens control and autofocus, but not in the first round, and maybe not in the second.

And I'd price it below Scarlet. RED followers will bash its quality, but if the costs are managed, it will own the next tier down in the market. With most people delivering on the Web, not everybody needs RED-level quality.

With this approach, it's mostly a mechanical exercise. The parts are already in the bins and the heavy R&D cost have already been paid off.

Given that the "DvSLR sensor in a video cam" question is the most asked item of Canon's pro video reps (according to one rep at CES), why not answer the market?

What you say is true, Chris, but Canon certainly has the ability to make a world-beating 2/3" cinema camera (in a much different price bracket than is under discussion) yet they won't do that, either, I'm afraid. Sigh.

I'd love a 2/3" all-manual shoulder mount camera with great color and low light capability. They would take over the market from Sony, JVC, and Panasonic while using lenses they're already producing. Seems like if they can make a 7D for under $2k and a prosumer body with all the switches and knobs for under $4k, they could make a pro body that took AB batteries and B4 lenses.

I think Jon is on the right track here, but it would seem a very limited marketing segment.

"Buy this vew vDSLR. It acts like a true film camera...No, we mean it!"

I think a fair amount of people might get such a camera and be disappointed that it would not act like a video camera. Might be a customer service/relations issue.

It would be a pretty cost efficient way for Canon to give the (small) market what it wants.

The next few years will probably heat up in this space.

Canon would also have to deal with the cry for resolutions higher than 1080 w/ 8-bit 4:2:2 color. Red has a whole set of custom software made to deal with this need. There is no way, IMHO, that Canon will make its own version of RedCine-X.

And as soon as Canon markets a "real" digital cinema camera, 4K, 4:4:4, 12-bit Log will all be on everyones wish/need list. What codec do they use? It becomes a big mess.

If I were Canon, I'd let Red be Red and Canon stay Canon. I'd continue selling huge quantities of cameras and not do battle with a business model that only seems to work b/c a genius iconic businessman w/ $billions and a love for cinematography is running the show.

But the second Red seriously tries to follow through on its once mention goal of selling to soccer moms, I'd try my best to vanquish them. Until that happens, I'd be content fight only border skirmishes w/ Red, using #D's and MarkRomannumerals as my weapons.

Fascinating though it is, the debate about 35mm detector cinema cameras is over my head, and not very relevant to the original topic of this thread - the prototype SDHC-card cam in the pictures that looks like an XH-A1s replacement.

I moved up from an XM1 (which I loved but became unreliable) to an XH-A1 in 2007, and have been pleased and impressed with the pictures the HDV cam produces. I often incorporate footage from a friend's Sony FX1 in my productions, and the Canon is noticeably better in the same conditions, except in very low light.

My only gripe with the XH-A1 is its bulk. Lugging it and a suitable tripod by myself across moorland and around city streets is a pain, and if I fly, it means instant excess baggage charges. For me, the XM1 is the ideal size. The Panasonic HMC40 looks very interesting, as does the JVC HMC-100. I'm hoping that Canon will follow the XH-sized cam with a slimmer version that will fill the gap that currently yawns between the HV40/HF-series and the XH.

I'm in no hurry to move to tapeless shooting. I have an archive of a couple of hundred DV and HDV tapes, so I will need a means to at least play them for many years to come. However, I've always known it was a matter of "when" not "if". Hopefully, my present cam and editing PC will keep going for at least another year. Looks like next year's budget ought to include a Canon AVCHD/SDHC cam and a PC capable of editing the material. (I wonder if Avid Liquid's replacement will be ready by then? But that's another story...)

The problem at the moment is sitting on one's hands until some working models appear at NAB and similar events. I wonder when the first 25p/50i versions will show up?

While everyone is asking and waiting for the next camera in the video line to align with the DSLR camera line, I like this idea that Canon go in reverse. They can just continue improving the video capabilities of its DSLR, alleviating some of the R&D burden from the video division. (I know Chris pointed out in the article that photo and video fall under the same group, but as far as camera lines go, we view photo and video as separate)

As someone pointed out earlier, the Digital Cinema market is too small for Canon to really target effectively. I can see them using the guise of the prosumer photo market that sells to a larger market to build the bridge to the video line, not the other way around. That way, we can slowly see the video features improve (such as zoom lenses, audio recording, etc.) while still making cameras that sell really well.

But from the other perspective, a company can only go with a line of cameras for so long before they are well behind everyone else... and these 1/3" chips are eating dust right now. And I doubt these predicted changes to the XH line will suffice what we wanted.

Well its hard to say that 1/3" chips are eating dust right now. Every manufacturer makes them and will probably continue to make them for some time coming.

1/3" does serve the greater whole more than a larger APS-C sensor would. And that's what Canon would be interested in most.

Again, awaiting what they will ultimately do.

Here's an interesting thought, and not to sway the conversation too far.

I am most likely going to upgrade to the next version of the Canon XL or whatever that equivalent would be, but have been toying around with keeping my XL H1S and using the money to buy a NanoFlash for it. What would you guys do?

Sorry, but 1/3" chips are the key to getting a 20x zoom under $4K.

The XLH1s w/ the Nano (100mbs or higher) is a great way to go.....In fact, you go to the head of the class.....so to speak.

Jim Martin

The suspense is killing me! ... as far as the specs for what Canon's new camera will have.

I'm still faithfully using my GL1 and XL1s in the meantime.

Based on what Chris said above, if the chips are 1/2" we sacrifice some zoom capability to fit in a simliar form factor... what would it go to? 14x or something? I could probably live with that if it meant better low light performance and better picture performance overall.

Right now, I've been distracted by this notion of building a DIY dolly system and also a DIY slider system. I figure I can go ahead an invest some $ in these options and help take my videos to a completely new level, and by the time I get that all ironed out, maybe Canon's new cam will be out! (Side note, also impressed with that kessler crane-light product!.. might get one of those too a little later.)

What I'm not looking forward to, is having to spend $$$$ on a computer system. Mine is about 4 years old! But by the time I get ready to upgrade to a new one to handle the new Canon camera, the Intel 6-core processor should be out, which may help things a bit.

.... waiting....

Sorry, but I don't see any significant road block there.

True Cinema camera = manual focus and manual zoom.

So there goes the AF and the motorized lens. Problem solved.

People aren't using motorized zooms and AF on the 5D/7D, and they darn sure aren't doing it on RED, ARRI, Viper, SI2K, Genesis, DALSA, or any other True Cinema camera.

So get on with it...

Not true at all, sorry -- motorized ZIFs are used *all the time* in high-end digital cinema.

Take out the AF and a motorized lens, and you've taken Canon out of the equation.

I thought we were clear about the target price range: around $5K or so. Not the high-end stuff like Viper, et al.

If it's Canon, then it'll have AF and a motorized zoom... just like the forthcoming 2/3" Scarlet with integrated 8x AF lens for less than $5K. That's what we're talking about here. Not the $15-$20K+ market. Canon won't ever go there.

Ok, as you've alluded to later, we are talking different markets.

Right. Which is sad and totally unnecessary.

So, putting a manual SLR lens on a A1/H1 replacement with a 7D sensor in it is undoable because the development of an AF and motorized zoom is not feasible. Even though the market clamoring for that camera is using non AF and non-zoom lenses on the 5D/7D to begin with... Does this strike you as odd?

This I cannot doubt at all. Frankly, I'd rather see RED leave this market altogether. I have no doubt that in 1-2 years another vendor will be offering exactly what people are asking for, and doing it in the $6k-$8k range.

I think the days of 2/3" shoulder mount cameras is coming to a rapid end. Every time I see a local affiliate out these days it's with a HVX200 or an EX3. I don't see anyone carrying the shoulder mounts any more unless its the rookie who doesn't yet rate enough to get a new camera.

No. Not in the slightest. They're using those lenses because currently there is no other alternative. And they've been complaining rather loudly about the lack of AF in video mode ever since the 5D Mk. II came out. You don't hear so much about motorized zoom because the folks that really need it aren't about to abandon their camcorders (because there is no kludge or workaround for that, or for the lack of long zoom ratios).

It might be a difficult realization for some of our readers to accept, but the fact is that the market clamoring for that camera simply is not that large. Vocal, maybe, but not that large. And they are grossly outnumbered, as I have previously mentioned, by the corporate / event video market that needs AF and long zoom ratios and an MSRP under $5K. That's the market that will be served by this new Canon replacement for the XH series.

I don't doubt the major money market is the corp./event video market. I'm firmly in that market. By day. But the 7D side of me (my weekend/night work) is digital cinema. And that part of me could care less about AF or digital zoom. That is the part of me that is buying the 7D, rails, follow focus, remote iris control, and steadicam.

Maybe it's not economically feasible for Canon to address that market in their video division. That's ok. I think that's why RED decided to step in there anyway. And do what the others wouldn't. I didn't even realize the fixed lens Scarlet was supposed to have a motorized lens. Just never cared.

In ant event, I wish Canon well with whatever video camera they decide to bring out. I just hope it brings something more to the table than the "me-too" stuff that's been speculated about here...

Well other than the post by some who reckons they have insider knowledge I would be surprised if Canon release something with either a 35mm or AP sensor. They are a conservative company and I doubt that they are suddenly going to start competing with RED which runs as the completely the other end of the spectrum. Even just using the same guts as a 7D and making it more video cam like would still be expensive to put together the production line......

I'm still very interested to see what RED do with Scarlet and AF, it's gonna be a tough gig to get it right, some companies have spent years on it and still not cracked it.

Is it possible to split this thread in two? One for discussion of the Canon prototype and another for those who want to play "fantasy digital cinema cam"?

Sorry, that's been my fault primarily. We'll push this back to discussing the prototype. I'm starting to second-guess a couple of my predictions, so I'm going to update that article (today, hopefully).

The ergonomics of the EX3 aren't great and it would be better as a shoulder mounted camera. Above a certain size and weight shoulder mounting makes sense and the EX series are entering into that territory, plus the much loved feature of the JVC HD series is the shoulder mount.

The BBC has 2/3" shoulder mount cameras and there continues to be a demand for them. HD 1/3" chips (at least currently) have problems in low light. Videojournalists tend to be the news people shooting with these cameras rather than the BBC camera people. However, they do get used on documentaries and general programmes, often by directors and researchers.

The 2/3" Scarlet is sitting at the cross over point and how it's used my depend on how much hand held work you use it for. However, being hip and having a sore arm at the end of the working day does seem a pointless exercise.

The requirement for 1/2" and above sensor size by HD broadcasters makes certain requirements size wise in a camera and this will effect how you operate it. Body braces for cameras is so a 1950s concept, the great camera designers of the 1960s, 70s and 80s did away with them.

Chris-

I knew you had a crystal ball !

...Jim

ps- give me a call today at FT

Apart from shooting the kids at a birthday party who would use an AF lens on a video camera? Ever? I can't imagine anyone wanting one a even a semi-pro camera. In fact it makes me shudder just thinking about it hunting focus and ruining shots!
Steve

There's another feature that could be added with a DSLR sensor in a video cam body that does not exist in the DvSLRs - a KILLER digital zoom.

Consider that we have aliasing problems with the pixel skipping. A 3x digital zoom switch would get rid of the aliasing, and would extend the zoom range. A 5x optical zoom, like a 28-135, coupled with a 3x digital zoom gives an effective 15x range.

A digital zoom just doubles up the pixels though, so you'd get worse quality not better.
Steve

Normally, a digital zoom doubles up pixels. In the case of the 7D/5D2/1D4, Canon skips roughly two of three pixels. For the digital zoom, I propose that Canon window the sensor. They could do that without line skipping, and quality would increase.

OK. But that wouldn't be a digital zoom, that would be windowing, that's why I got confused.
Presumably there must be some reason why they didn't do that? Just to keep proper coverage with their lenses maybe? I suppose the other problem with windowing may be that you'd lose too much resolution?
I'm sure you'd find that it's not as simple as it seems.
Steve