Join 3,425 readers in helping fund MetaFilter (Hide)


A new market for High Definition surveillance cameras?
September 24, 2006 7:06 PM   Subscribe

RED ONE is a 12.6 megapixel digital film/HD camcorder developed by Jim Jannard, founder of the Oakley sunglasses company. The camera will retail for $17,500, and is alleged to outperform HD and digital film cameras from established companies like Sony, Arri, Panavision and Dalsa (whose offerings all cost well in excess of $100,000). The general consensus among pundits in media production circles is that Jannard's camera will be a true disruptive technology. Last night, no less than 24 hours after the very first publically available sample images from the camera's "Mysterium" sensor were posted to the RED Digital Cinema website, the company's development offices were broken into. According to Jannard, "Everything they took was camera and camera file related...there is no question all they came for was RED camera stuff." (Additional obligatory and annoying YouTube links: First public demonstration of the RED camera at the IBC convention in Amsterdam and the RED Q & A session that followed.)
posted by melorama (79 comments total) 4 users marked this as a favorite

 
A part of me immediately assumes this was your everyday case of industrial espionage, but the cynical, other side of me thinks that it could also be a PR stunt.

It seems oddly coincidental that the break-in happened on the very same day that they publically posted images from the heretofore vaporware "Mysterium" CCD for the first time ever.
posted by melorama at 7:27 PM on September 24, 2006


that's quite a long post for something not immediately interesting.
posted by wilful at 7:45 PM on September 24, 2006


humorously enough, that youtube video is refusing to load for me.

the whole red thing really does reek of vaporware, though.
posted by tumult at 7:46 PM on September 24, 2006


and thanks for the snark, wilful
posted by tumult at 7:48 PM on September 24, 2006


They certanly have an ugly-ass website, though.
posted by delmoi at 7:50 PM on September 24, 2006


If I had this amazing new "disruptive" tech, I imagine that I would do what was necessary to safeguard anything and everything related to said tech, including security guys, what, maybe a SAFE!?

Either it's a PR stunt or they're idiots that are unable to secure their shit.
posted by exlotuseater at 7:58 PM on September 24, 2006


I'm not sure the sample pics on their website do anything to sell the camera. Interesting camera design though. Seems terribly familiar somehow...
posted by Thorzdad at 8:03 PM on September 24, 2006


wilful: "that's quite a long post for something not immediately interesting."

I suppose one could say the same for yawn-worthy posts about the Commonwealth Games, Worldwatch Institute reports and George Monbiot rants too, eh?

Although I think those subjects are boring with a capital-B doesn't necessarily negate the fact that they may be relevant and of interest to others, now does it?

Now move along and snark in another boring FPP, if you don't mind, please...
posted by melorama at 8:06 PM on September 24, 2006


Okay, call me even more cynical but I actually don't even believe this company or camera exists. "Mysterium" sensor?

Everything about this company seems extremely amateurish, it's hard for me to imagine that a company that could really raise the kind of money needed to build something like this would really be so incapable of hiring a web designer.

The cincher for me though is the "still image gallery". The photos look computer generated especially the car and watch, but the pictures of the women look a lot like some of the really high end 3d stuff shown at the CG choice gallery.

And look at this picture (4096x2304 pixels) The girls face looks realistic, but still slightly plasticky. But what about the "smoke" It looks totally bizarre and not at all like cigarette smoke.

I say the so-called still images are actually CG and that this is viral marketing for some CG product, or who knows what. Maybe even an investment scam.

I mean if you really had this amazing camera, wouldn't you take pictures outdoors in the sunlight? Why would you take such contrived compositions and where would you get such smooth and flawless inanimate objects to photograph?
posted by delmoi at 8:07 PM on September 24, 2006


I like how the sample images are obsessed with beautiful women smoking....the all-seeing lens of a revolutionary new technology can certainly do worse than seek out blondes with cigars.
posted by Bistle at 8:07 PM on September 24, 2006


I wonder if this is an attempt to scam pre-orders.
posted by delmoi at 8:08 PM on September 24, 2006


exlotuseater: "If I had this amazing new "disruptive" tech, I imagine that I would do what was necessary to safeguard anything and everything related to said tech, including security guys, what, maybe a SAFE!?

Either it's a PR stunt or they're idiots that are unable to secure their shit.
"


That's why I'm inclined to think that it's some wierd PR stunt. You'd think a guy who ran a company as huge as Oakley would know enough to have decent security, let alone not posting a direct link to a satellite photo and street address of the office holding all this top secret technology on their website
posted by melorama at 8:09 PM on September 24, 2006


As someone who is following the Red Camera developments, it is odd to see they get broken into the day after. But then agian, it does seem like a good time to do it. If I had just finished one of the most distruptive technologies the industry has seen, I'd be out partying and getting drunk, celebrating my accomplishments.
posted by SirOmega at 8:11 PM on September 24, 2006


Pepsi Red?

Sorry, I had to.
posted by knave at 8:15 PM on September 24, 2006


...and sleeping with a gun under my pillow.
posted by exlotuseater at 8:16 PM on September 24, 2006


Ah, looking at the youtube video, it's that the girl was blowing a bubblegum bubble, not smoking. If that was animation, it was very good. But I still imagine it's possible it could be CG.

The whole thing just seems fake to me, though. Filming scenes outdoors would show off the camera much better then these contrived settings, and getting the CG produced would be a lot cheaper then developing a sensor that outperforms anything made by huge companies like sony, cannon, and other companies that spend millions on sensor research.
posted by delmoi at 8:25 PM on September 24, 2006


It looks totally bizarre and not at all like cigarette smoke.

That's because it's a gum bubble just after bursting. And if you go to DVInfo.net, dvxuser.com, hdforindies.com or cinematography.com, I think you'll see that it's not some viral campaign for CG or a scam. They also showed a prototype and footage at IBC. And the main guy is Jim Jannard who founded Oakley. And their team has several well respected people in the video community.
posted by chris24 at 8:25 PM on September 24, 2006


You guys don't think the company exists at all? This company is fairly well known in digital video circles. They sell things that real people own. They showed off a real camera at a real event. The smoke is strange looking because it is out of focus.

And Thorzdad: Those aren't pictures, they're captured frames from videos. If you're not impressed by this then you must be pretty hard to impress.
posted by floam at 8:25 PM on September 24, 2006


delmoi: "Okay, call me even more cynical but I actually don't even believe this company or camera exists. "Mysterium" sensor?

Everything about this company seems extremely amateurish, it's hard for me to imagine that a company that could really raise the kind of money needed to build something like this would really be so incapable of hiring a web designer.


You're right to be cynical. Pretty much everybody in the film and video production industry was (and still is, to some extent) when RED was first announced.

But since you don't really follow this industry closely, I'm probably in a more suitable position to state that if there's one thing that is undeniable about RED, it's that this is indeed a "real" company, with real, well-established , highly respected people in the industry working on this product (guys like Graeme Nattress and Ted Schilowitz, who are practically household names in the production industry are key engineers who quit their previous, secure day jobs to go work for RED). Whether or not they can actually ship this product at the stated price, however, has yet to be seen.

RED doesn't need to "raise money" to build this product. Jim Jannard is self-made billionaire. He originally founded Oakley as an optical research company, not a sunglass/fashion company. His motivation to form RED was purely out of his own dissatisfaction with all of the digital cinema and HD cameras that are currently on the market (he's a total camera and film geek aparantly), and set out to make a camera that he would be happy with...it turned out that such a camera was, by necessity, inherently "revolutionary".

But the guy certainly has money to burn, and none of the existing burdens that big companies like Sony and Panasonic have to protect in order to ship a product like this. They don't have an existing customer base to worry about alienating by releasing a product that good, at such a low price. I can imagine a LOT of Sony F-950 owners are a bit peeved at the mere idea of RED.


The cincher for me though is the "still image gallery". The photos look computer generated especially the car and watch, but the pictures of the women look a lot like some of the really high end 3d stuff shown at the CG choice gallery.

I say the so-called still images are actually CG and that this is viral marketing for some CG product, or who knows what. Maybe even an investment scam.


Go have a look at those YouTube links once the site starts working again...as crappy as the video is on YouTube, you'll quickly see that if the footage of the women are CG (and I'll grant you your skepticism about the watch and car images), then they will have probably created the most amazing, stunningly lifelike CG of human forms ever seen.
posted by melorama at 8:28 PM on September 24, 2006 [1 favorite]


Oh, and regarding the hilariously named "Mysterium" sensor:

"I've taken a lot of ribbing for some of the names I have come up with for our materials and products over the years. Unobtainium was the 1st back in 1976. "Thermonuclear Protection" was another. The "Mysterium" sensor is just that. A mystery because we won't talk about the process or the nature of how we are getting 4k @ 60fps from this miracle sensor. Trust me, many companies have tried every trick in the book to find out."

(from an interview with Jannard on HDForIndies)
posted by melorama at 8:34 PM on September 24, 2006


I'm thinking at least a bit of marketing smoke if not outright vapourware. Isn't 12MP and 60 frames per second somewhere around a GB per second of transfer? What the heck bus are they using and what are they writing to? Firewire 800 is 786 Mb/s and SATA 3 is 3Gb/s. That or some serious on the fly compression.
posted by Mitheral at 8:34 PM on September 24, 2006


As modular as it is claimed to be, it would seem like a very high-end product could be created from this for the prosumer market, that would easily blow away entry- and mid-level digital SLRs.
posted by Blazecock Pileon at 8:35 PM on September 24, 2006


Peoples...we're not talking about a Digital SLR STILL camera. This is a video/cinema camera.
posted by melorama at 8:38 PM on September 24, 2006


Sorry Blazecock...i misunderstood your post.

That is actually an angle to this that never even occurred to me. I would be a great idea to dumb some of the technology down to a prosumer/consumer still camera.

And lord knows Jim Jannard could design a goofy looking case for such an animal quite easily.
posted by melorama at 8:47 PM on September 24, 2006


This company is fairly well known in digital video circles. They sell things that real people own.

Really? You wouldn't know that from the web site, all it seems to show is mysteriousware. Although I will admit to owning some Oakleyware, it is more hype than engineering. Who would name their sensor Mysterium unless is was really imaginarium? Also, they do not seem to have any patents, at least in their name. I would kind of like to be wrong and see the technology march forward, but stories of little guys beating out the established world with "mysterious and secret" technology are almost always false.
posted by caddis at 9:04 PM on September 24, 2006


Mitheral: "What the heck bus are they using and what are they writing to? "

Enough answers here to make yer head spin...

http://www.hdforindies.com/2006/09/amsterdam-ibc-2006-red-news-redcode-4k.html
posted by melorama at 9:06 PM on September 24, 2006


And FWIW, it wont be long (end of 07) before we see 300GB 2.5" SATA drives with write rates exceeding 50MB/s minimum (prolly 75MB/s max). So I'm not worried at all about how they're going to write this stuff out to disk.
posted by SirOmega at 9:18 PM on September 24, 2006


Mandatory: Haven't you lost anything Doctor Bronx? Your purse? Your car keys?

But yeah, c'mon, poorly-conceived PR.

Anyhoo, I'm going to leave this just-completed cold fusion design on my desk while I go have lunch. Keep an eye on it, would ya?
posted by dreamsign at 9:18 PM on September 24, 2006


So they are using ~10:1 on board compression. Still as pointed out in that link at full resolution and frame rate they are taxing the limits of even non portable storage. Doable though.
posted by Mitheral at 9:25 PM on September 24, 2006


1. Some guy who runs a company that makes stylish bits of transparent plastic bonded to other bits of plastic thinks he can engineer a revolutionary video camera system? Yeah right!

2. 12 Megapixels * 3 bytes per pixel * 60 frames per second = 2,160 megabytes per second, or 17 gigabits per second. You could stream the data along an OC-768, but you would fill a 1TB hard drive in less than a minute (assuming you just stored the bits as they arrived).

3. The camera is a small part of the HD story. Video data transmission, storage, and editing are every bit as important as the camera. Given the cost of large storage arrays and rendering farms (for processing all of those frames on the NLE system), I suspect that the low price tag of the camera will matter little to the film industry. Home producers won't have the millions of dollars necessary to support the camera, so a $17k HD cam won't do them much good either.
posted by b1tr0t at 9:25 PM on September 24, 2006


Ok, so we are essentially talking about motion photography / videography, but still - how long do you think it will take Hasselblad to enable this insane beast - or at least its architecture, anyway - video capable?

Since it already has more than double the Red's resolution at not even twice the price...?

If there is a "disruptive tech" effect here, it'll be owed to one thing alone: not "Mysterium", but "Marketing".
posted by objet at 9:58 PM on September 24, 2006


b1tr0t:

1. The fact that he has started a successful company before is more important than what he actually invented before. It's unlikely he is doing the actual technical work.

2. Compression. Using your same logic my 5MP still camera would need 15 megs of space for every picture (it doesn't). Higher ratios of compression are achievable for movies than stills.

3. You can already today manipulate HD video on a PC.
posted by ill3 at 9:59 PM on September 24, 2006


RED doesn't need to "raise money" to build this product. Jim Jannard is self-made billionaire. He originally founded Oakley as an optical research company, not a sunglass/fashion company.

Okay okay. I suppose the wavelet compression that they're using could be responsible for the 'smoothness' and CG-look of everything in the images.

Still they really ought to hire a graphic designer. Every single image on their page looks CG-ish

His motivation to form RED was purely out of his own dissatisfaction with all of the digital cinema and HD cameras that are currently on the market

Yeah, the description sounds like what I would design if I were going to design a video camera :P Very nerd oriented. Ah well.
posted by delmoi at 10:15 PM on September 24, 2006


As modular as it is claimed to be, it would seem like a very high-end product could be created from this for the prosumer market, that would easily blow away entry- and mid-level digital SLRs.

It would be a huge waste of their bandwidth technology, though.
posted by delmoi at 10:19 PM on September 24, 2006



RED TEN'S VOICE: "Red Five, can you see them from where you are?"
posted by stbalbach at 10:24 PM on September 24, 2006


Ok, so we are essentially talking about motion photography / videography, but still - how long do you think it will take Hasselblad to enable this insane beast - or at least its architecture, anyway - video capable?

Kodak makes a 39 MP sensor (bottom of the list), but it can only capture at 0.9 FPS. Unless Hasselblad is using a dramatically faster sensor, it will never do video.



1. The fact that he has started a successful company before is more important than what he actually invented before. It's unlikely he is doing the actual technical work.

He started a successful company that made products out of bonded bits of plastic. Product design, development, and testing is radically different for video products. How successful do you think Bill Gates would be if he took over Quicksilver?

2. Compression. Using your same logic my 5MP still camera would need 15 megs of space for every picture (it doesn't). Higher ratios of compression are achievable for movies than stills.

If you were shooting in RAW mode, it would. When you shoot in JPEG mode, you lose a lot of dynamic range.

3. You can already today manipulate HD video on a PC.

at 12 MP / frame?
posted by b1tr0t at 10:45 PM on September 24, 2006


Reminds me of something I wondered about the other day. What's next with TV after HDTV? Are they even working on something yet?
posted by Brian James at 11:00 PM on September 24, 2006


OK just to throw my thing in here...
I know that it's real, and my friend has already preordered one.

B1tr0t, go read up in the DVXuser forums to find all the answers you're looking for. They're using a crazy JPEG2000-like RAW compression scheme that is done live by a dedicated onboard chip. It's for real, and it's driving Sony and Panasonic nuts - they can't figure out how they're able to pull down 4k images at 30fps, and my guess is one of those guys did a number on the office to see what they could find.

And you can manipulate HD, I'm working on a project in HD right now. It's slow to deal with, and 4k would be even more hard, but you render proxy clips in say DV and save the time markers etc, then switch in the real footage for the final product. So chill out and celebrate, this camera's going to be a revolution.

And you know that "switching industries" argument doesn't hold water... the guy is funding the R&D and directing the PR and such. He's not writing the compressor code or engineering the sensor - that's what all the people he hired are for.
posted by BlackLeotardFront at 11:12 PM on September 24, 2006


b1tr0t: "1. Some guy who runs a company that makes stylish bits of transparent plastic bonded to other bits of plastic thinks he can engineer a revolutionary video camera system? Yeah right!


You really need to RTFA(s).

Jannard isn't just some delusional rich guy who "thinks he can engineer" a camera (idealistic and risk-taking, maybe, but definitely not delusional). His company started as an OPTICAL RESEARCH COMPANY before they even got into the fashion-wear market. He is actually involved directly with the industrial design of the camera itself. In fact, Ted Schilowitz (formerly one of the main geeks at AJA, before leaving to join the RED team) mentioned in an interview that Jannard isn't just "coming up with ideas" for the camera, but he's actually responsible for 100% of the camera's external design, as in, the guy is actually working every day in front of a computer in a dinky little cubicle (there were a bunch of photos on the DVXuser forums IIRC) doing actual CAD work on the thing.

I doubt he's working directly on the hardcore geek-level sensor stuff and codec development, but since RED is a company, they have, like...you know...employees...who DO know this shit inside and out.


2. 12 Megapixels * 3 bytes per pixel * 60 frames per second = 2,160 megabytes per second, or 17 gigabits per second. You could stream the data along an OC-768, but you would fill a 1TB hard drive in less than a minute (assuming you just stored the bits as they arrived).


Again, you need to RTFAs.


3. The camera is a small part of the HD story. Video data transmission, storage, and editing are every bit as important as the camera. Given the cost of large storage arrays and rendering farms (for processing all of those frames on the NLE system), I suspect that the low price tag of the camera will matter little to the film industry. Home producers won't have the millions of dollars necessary to support the camera, so a $17k HD cam won't do them much good either.
"

RTFAs, please. This has been covered ad infinitum.

Also, its 2006. It no longer costs "millions of dollars" to work with HD material. In fact, it hasn't cost that much for at least 5-6 years now. The post production process is the cheapest part of the pipeline. Anybody with a copy of Final Cut Pro and fast enough disk storage can edit HD with no problems.

What makes this camera so revolutionary is that drastically lowers the price for acquisition, which is still by far the most expensive part of HD production. A Panasonic Varicam, for example (one of the defacto standard HD cameras in the industry) will cost you up to $350,000. Same with a Sony F950 or F900.
posted by melorama at 11:33 PM on September 24, 2006


Melorama, you make my point better and more thoroughly. Although I should add to your statement that my friend getting the RED also already has a Panasonic HVX, the consumer level Varicam (same sort of sensor, tech etc AFAIK) for shooting HD and it looks great, so HD is already acquirable, but not at the RED level of quality. The HVX cost about $6000, and that's no joke, but a lot of people who just dropped half a mil on Varicams and other options are going to be mad when RED comes out. Of course, with their money they will buy several...
posted by BlackLeotardFront at 11:46 PM on September 24, 2006


It looks totally bizarre and not at all like cigarette smoke.

That's because it's a gum bubble just after bursting.


No way. She's obviously a freaking ALIEN.


posted by duende at 12:00 AM on September 25, 2006


Reminds me of something I wondered about the other day. What's next with TV after HDTV? Are they even working on something yet?

Ultra HDTV
• HDR HDTV
posted by Blazecock Pileon at 12:04 AM on September 25, 2006


OK, really stupid question for the video guys in this thread...

What makes this camera so revolutionary is that drastically lowers the price for acquisition, which is still by far the most expensive part of HD production.

Are we using HD to refer to any really high-resolution video capture method? And this camera has accomplished some new height in this area? Or are we referring to HD as in the HDTV standard? Because if we're talking about the latter, then I don't see how this camera is such a breakthrough in a practical sense--yeah it kicks ass and all, but I don't see how it could make HDTV recordings look tremendously better, since the display resolution is going to be the same anyway.
posted by Brian James at 12:16 AM on September 25, 2006


If you were shooting in RAW mode, it would. When you shoot in JPEG mode, you lose a lot of dynamic range.

You would lose detail, but not dynamic range. You can probably boost the bit depth on jpg as well.
posted by delmoi at 12:22 AM on September 25, 2006


Kodak makes a 39 MP sensor (bottom of the list), but it can only capture at 0.9 FPS. Unless Hasselblad is using a dramatically faster sensor, it will never do video.

So are you saying that the youtube videos they put up are fake? Otherwise it looks like he's already done video, or at least a few seconds of it...
posted by delmoi at 12:25 AM on September 25, 2006


This is the sexiest thing I've ever seen. I'm saving up as we speak.
posted by stenseng at 12:37 AM on September 25, 2006


Brian James: " I don't see how this camera is such a breakthrough in a practical sense--yeah it kicks ass and all, but I don't see how it could make HDTV recordings look tremendously better, since the display resolution is going to be the same anyway."

Brian,

The fervor over this camera is not necessarily about the image quality, per se. After all, there are many cameras that exist now that make beautiful images at least as good as the RED ONE does.

The problem is, those cameras cost 10 to 20 times more (and in some cases, such as the Dalsa Origin camera, you can't even buy one...they only rent them out) than the RED will cost. And if the specs of the RED system are to be believed, it's arguably much more versatile and technically capable than any of its closest competitors.

For the average consumer of HD content, RED ONE isn't a breakthrough, at least directly. But it *is* an indirect breakthrough to them in that this camera will open up theatrical quality digital film production to a huge chunk of the independent film production market which is for the most part, currently shut out of 16/35/70mm film or digital cinema solutions, simply on acquisition costs alone.

This isn't just an "HD" camera. It's a true digital cinema camera. You can shoot at 2k or 4k, and project it in a theatre at full film quality, without worrying about the cost of film stock, telecine transfer sessions and distribution prints.

If they can actually deliver what they promise, this camera will do for acquisition what Final Cut Pro did for non-linear editing.
posted by melorama at 12:41 AM on September 25, 2006


melorama: "This isn't just an "HD" camera. It's a true digital cinema camera. You can shoot at 2k or 4k, and project it in a theatre at full film quality, without worrying about the cost of film stock, telecine transfer sessions and distribution prints."

This, by the way, is why companies like Sony, Panasonic, Arri, etc., are scared shitless by the RED camera.

It will either force them to drastically lower the cost of thier current professional cameras to remain competitive (much in the same way that Final Cut Pro has forced Avid to dramatically lower the cost of their Media Composer and XPress Pro editing products), or cause them to become a niche, "high end" camera company, catering to the type of customer who'd rather pay big bucks for the prestige of the brand (see also: SGI, Monster Cable, etc.) rather than the functionality of the product on its own merits.



In any case, it will be good for the consumer.
posted by melorama at 12:53 AM on September 25, 2006


Melorama:
Or, it will make those companies hire goons to case the RED offices.
posted by BlackLeotardFront at 12:57 AM on September 25, 2006


Oh I so much love these threads, usually

1. one thinks he is competent enough to address the newtech from a theoretical point of view
2. one thinks all the data he possess is enough to assess wheter the tech is possible or not
3. usually only x out of a lot larger number has all the knowledge to make this assertion with 99.9 confidence
4. all the others want it already, if anything because of it sexy gunmetal color and redness

Oh well , we will see ! If you excuse me, I will go find a 2560p monitor , the new hardware fast enough to edit, manipulate and play me that buttload of data in hope the difference from 1080i will be noticeable enough to give a shit. As you know, you can't possibily do it on the web , but don't worry procastinate your buy enough to let them show...mmhhh...Windows Vista 2560p edition.
posted by elpapacito at 2:21 AM on September 25, 2006


Snarks aside, are cameras improving as fast as I think they are?

I was prepared to blow ~USD1000 on a digital SLR and decided to spend roughly a quarter of that on something that'll do about 70% of what the SLR would do, plus operate on penlight batteries and be easy to use one-handed. I'm curious what roughly the same price points will get me five years from now.
posted by pax digita at 4:45 AM on September 25, 2006


paxdigita: the a630 is out and for $30 more then the 530 you jump from 6 to 8 Mpix. I'd do that. Also, the SLR , I still don't own a digital one so I am just relaying experience of a friend who claims he is the coolest ever AND that SLR behave very well at high end of ISO range, while you can't expect that from a compact. I really didn't bother looking more, but I will soon do as I am in the market for a new camera.

My guess is that in 5 years the same price point will give a 1x Mpix camera which also double as good quality videocamera that saves realtime compressed on memory card and puts out in wifi too.
posted by elpapacito at 5:04 AM on September 25, 2006


I love when peecee folks start talking about video as if they have a clue. Oh boy, I've played back quarter-res QuickTime in Windows, no wait I've edited home movies with Windows Movie Maker, no wait I've shot stuff on my 3-chip prosumer camera, and I am totally sure that this RED camera is BS!

OK, thanks for the snark opportunity ... now for some constructive commentary ...

cred: I care about video compression quality for a living, and work in the media industry, although only have about 5 years experience in video itself (i.e. cameras, studios).

First, the HDforindies link is very informative. b1tr0t made a few good points.

I would like to provide some context. 17 Gbit raw data out of the camera sensor sounds about right, and not obscene, because standard uncompressed HD uses 1.5 Gbit links to move the image around (HD-SDI aka SMPTE 292M). Getting hardware to pull in and push out that bitrate is a solved problem -- all of the HD TV networks (the big 4 plus the cable HDs like TNT) have had this running in their network plants for 3 years, and HD itself has been around for about 10 years so that technology is mature. The problem here is that this is 10x higher bitrate, and they need to make it fit in a camera. But considering where we were 10 years ago it's not out of the question.

Now, that was 1920x1080 HD, about a 2 Mpixel frame. Discerning film makers are not satisfied with HD, and so 4Kx2K has become the holy grail, or at least the minimum resolution at which nearly all filmmakers stop complaining about it. And over the past 5 years the whole filmmaking post-production chain has already converged on 4K as a production standard.

Theatrical projection is pretty much standardized on 2K/HD frames now -- Barco/Christie/TI are all furiously busy right now installing 2K projection systems around the world. Sony has a 4K projector, but the biggest problem with it is feeding the beast. Pushing 4Kx2K out of a server for 2 hours straight has been the hurdle that they've been working on for 2 years. Mark Cuban's Landmark Theaters signed up in Jan 2005 for these Sony 4K projectors, and after nearly a year of waiting they threw in the towel and went with a TI 2K rollout. Some Sony 4K's are sprinkled at Landmarks around the country (Atlanta has one in theater 8) but they just can't mass-produce them yet because it's too hard.

So all this is to illustrate that 4Kx2K storage is indeed hard but certainly doable, and I've just been talking about projection. RED is just capturing and moving images inside the camera! Not a big deal at all, except for the need to make it small and not consume hundreds of watts -- yep, that's hard to do, which is why you see this talk of SATA and hard drives.

Next thing: compression rates. The movie industry's digital cinema (DCI) standard uses JPEG2000 compression, which is intraframe-only. As far as I'm concerned, the reason to use intraframe and not interframe is so that each frame can stand all by itself and so you can edit at will. At the other end of the scale would be HDV, which is pure f*cking hell for editors. Nice low 15-25 Mbit bitrates that fit on those cute little tape cartridges, but a gigantic PITA to work with.

So, ill3 said "higher ratios of compression are achievable for movies than stills", and I say "no they are not if you're aiming at directors and DPs, because you can't use interframe compression."

Thanks melorama for the great links and commentary!
posted by intermod at 5:12 AM on September 25, 2006


Random conjecture:

The technology to do this seems pretty doable, it's not really that much data to push around. It's not like it will make for a light weight camera with all the hard drives, but this is pro kit so I guess it's not such a massive issue.

High end DSLRs can do 18MP easily for less money going at least 4fps. Sensor speed isn't the problem, storage, processing and moving the data around is the problem. Considering there are very few (none?) output devices used in conventional cinema capable of outputting images at that resolution, industry demand seems like it would be quite low, reducing major manufacturers desire to step up to such an extreme level of detail.

This camera isn't so much "revolutionary" as sort of pointless to most people.

But still, shiny! I wish my computer could decode video that large fast enough for realtime :(
posted by public at 5:37 AM on September 25, 2006


I was prepared to blow ~USD1000 on a digital SLR and decided to spend roughly a quarter of that on something that'll do about 70% of what the SLR would do...

Spoken like a true non-owner of a dSLR. Unfortunately, I own a dSLR and I'm thus never happy with my smaller, more portable digital camera anymore. The low light or high speed performance is just plain terrible. Digital SLRs have *huge* sensors compared to the smaller point and shoot models, which means they gather way more light. You can use much faster shutter speeds and still get a well lit photo, or go up to ISO 1600 in a dark setting with tolerable image noise. Honestly, the difference between an SLR and a point-and-shoot is night and day, even if on paper they look very similar.
posted by knave at 5:41 AM on September 25, 2006


Not a big deal at all, except for the need to make it small and not consume hundreds of watts -- yep, that's hard to do, which is why you see this talk of SATA and hard drives.

With the right controller, it's easy and fits in a backpack. You need 5 or 9 high capacity drives. Storage is more important than speed, so we'll go with the current bull-moose drives, 750GB per. We're going to lose on for parity, that's 4x750GB, or 3TB (or 6TB for 9 drives.)

You need a battery to run the drives -- not hard, even for these drives. We'll assume a pro, using pro drives, like the WD RE2 500GB. Total power dissapation per drive during read-write? 10W. Five of them is 50W. That's notebook levels of power draw. 9 is 90W, assuming another 10W for the controller, that's 100W, a hair high, but nothing a couple of 7Ah 12V batteries can't handle. Total backpack weight? Each drive is about 1.5 pounds, so 13.5 for 9 drives, or call it 20 pounds for drives, controllers and batteries. You can halve the battery and drives for a five drive version, that's 10 pounds. Let's add 50% for structure, so 30 for 6TB on your back, 15 for 3TB.

This is all very buildable, and wouldn't weigh more than anything cameramen aren't used to hauling now -- esp in a well designed pack.

For mounted cameras in studios, the right answer is fiber to a SAN. A dual-path FC connection is 4GB/s a second, a fully loaded CX-500 with 500GB SATA drives (mod the one DPE2 shelf that has to be FC) is 300GBx15+7*(500*15)=57TB raw, or about 50TB with parity. I'm sure they'll be selling 750GB drive for those shelves soon enough, pushing that CX-500 up to about 75TB. Not cheap, but compared to the film infrastructure, not expensive.

Better, SANs have things like "SAN copy", which is how you get the data off to post production. You dupe off the SAN overnight, and you're ready for the next day's shooting.

With 2.5" drives, we're looking at beltpack, not backpack, but we're looking at less shooting time, and less reliablity -- the tradeoff decision must be made by the guy getting the footage.

Oh, and you'll want RAID-3.

Never heard of RAID-3? Not surprising, it's not often used. RAID-3 is byte level striping. As a random access device, it sucks.

But there's nothing random about recording video streams. That's all about continous write speed, and in terms of write speed, the only thing faster than 5 disk RAID-3 in terms of continous write speed is a 9 disk RAID-3. Playback is even faster. You'd *never* put a transactional database on one of these, because random access sucks, but for long sequential streams of data that cannot afford latency (say, video), RAID-3 is the exact answer. The heads simply walk, slowly, across the disk as the data is recorded or played.

Even editing isn't that random. Yes, you jump to different cues, but all the heads jump at the same time to the same relative place on the disk. They then stream off the data until the next edit point. The recording volume, of course, is just writing data.
posted by eriko at 5:52 AM on September 25, 2006


Honestly, the difference between an SLR and a point-and-shoot is night and day, even if on paper they look very similar.

Amen. They're also faster, and hell, they point better. It had been a long time since I'd shot the Maxxum 7s, so running around A Famous Theme Park in CA with a 7D was confusing for about 5 minutes, then it all came back. Best was nailing a portrait of a friend without framing through the eyepeice -- the hands just know how the camera points.

The larger, heavier glass has a cost (wieght) but a bonus (light grasp, mass=steadier.)
posted by eriko at 5:54 AM on September 25, 2006


eriko/knave I think you two will enjoy this article
posted by elpapacito at 6:19 AM on September 25, 2006


I can see why people might think this is vapor, but 4k at 60 fps is not that hard to accomplish if you use more than one sensor at once. Specifically if the optics (the lenses) split the image, say four ways to four 12MP sensors, with each one having its own storage and accompanying bus, then you only need each sensor to run at 15fps, and your innovation is in some controller that times the roundrobin collection. As long as each frame (or even data stream, for that matter, because the timing and number of sensors is a known quantity to the controller) gets a unique name or number in a format common to all of them, you just reassemble the streams later when you edit.

In other words, 4 or more sensors, each sensor has 1 SATA bus and one storage unit, for a total of 4 buses and 4 storage units.

I'm just guessing, but that would accomplish all of the technical features and keep it under $20k.
posted by Pastabagel at 6:54 AM on September 25, 2006


Also, I'd like to point out that Sony et al probably aren't all that worried about this camera, because they have so many patents on sensors and digital photography in general that the red is invariably going to infringe somebody's patent on something, and which point they sue red and strike a cross license deal, get the rights to the tech, and in true Japanese spirit, manufacture it to insanely high precision, tolerance, and efficiency.
posted by Pastabagel at 7:03 AM on September 25, 2006


elpapacito, thanks for the link, that was great.
posted by knave at 7:35 AM on September 25, 2006


I'm still interested to see how this all pans out, but I'm still skeptical about the "break-in".

ps I used to have Oakley sunglasses and loved them, but now I'm poor and I have a cheapass pair of Julbos.
posted by exlotuseater at 8:32 AM on September 25, 2006


The images and videos shown are odd and highly contrived choices of subject matter to demo a camera, motion-picture or still-picture. Wouldn't budget filmmakers be interested in seeing how the camera worked in natural or partially-natural light, with something visible in the background? Or focused on something more than 8 feet away?
posted by Western Infidels at 9:56 AM on September 25, 2006


intermod:standard uncompressed HD uses 1.5 Gbit links to move the image around (HD-SDI aka SMPTE 292M).

Well, if we're going to start being pedantic, then for HD at 1080p you actually need SMPTE 372M dual links, totalling 3GBps.
posted by Luddite at 10:44 AM on September 25, 2006


this thread rocks
posted by bonaldi at 11:21 AM on September 25, 2006


This camera isn't so much "revolutionary" as sort of pointless to most people.

In terms of use, your probably right. But assuming that it's real and can do what they claim for the price that they have listed (a lot of assumptions, I know), it's usefulness for the average person will come from the huge jump in independent movies that they can watch.

And if it actually works as advertised, in a few years you can see this technology working it's way down to consumer level products. Which would be of use to most people.

I'd love for this to be real. Not that I could afford it, but it does help to reduce the barriers towards me making a bad movie one day.
posted by quin at 11:39 AM on September 25, 2006


Considering there are very few (none?) output devices used in conventional cinema capable of outputting images at that resolution, industry demand seems like it would be quite low

You can get 4K out of film quite easily. A number of movies have had 4K digital intermediates.
posted by Potsy at 1:17 PM on September 25, 2006


Western Infidels: My take on the demo shots was that they were interested in showing the dynamic range of the camera. All of them went from very dark black to very bright light, and, to my admittedly amateur eye, they looked pretty damn good. It's not nearly as hard to produce decent-looking images (whether video or still) in a well-lit outdoor environment as it is in a black studio with a single bright key light.

You could shoot some video outside, but you'd have grab stills from it and go to some trouble to point out the dark areas and the light areas and how well they both retain detail. Most of the frame would be midrange. The clips they shot were intended to impress pros, and I think they probably did.
posted by rusty at 1:56 PM on September 25, 2006


Their choice of a black background allows for excellent compression; is it possible that they are having trouble with the data-rate?
posted by Osmanthus at 6:55 PM on September 25, 2006


Here's my theory. I say they used black backgrounds because this magical sensor sucks ass and needs a shipload of light to operate. All of the scenes were extremely brightly lit, and still had a black background. Also you'll notice on the faces that the Depth of field was very shallow, indicating a wide aperture.

The videos were only a few seconds long. Show me a five minute movie shot outdoors, and I'll believe in it.
posted by delmoi at 11:25 PM on September 25, 2006


There are some cameras that might compare pretty well with a dSLR like the Sony R1 and Panasonic FZ30, but duh, that little cannon would really not compare to a dSLR. Yes, they will both "take pictures" but the pictures just won't look as good in most situations.

I have a little cannon myself (SD450) and I'm really disappointed with it: tiny lens, crappy dynamic range, etc. Sucks. My Sony V1 was much nicer, while still being a point'n'shoot.
posted by delmoi at 11:39 PM on September 25, 2006


if the optics (the lenses) split the image, say four ways to four 12MP sensors, with each one having its own storage and accompanying bus, then you only need each sensor to run at 15fps, and your innovation is in some controller that times the roundrobin collection

That's exactly what I was thinking as well, Pastabagel. Theoretically you could get 8k images right now using a similar technique: point 15 digital SLR's at the same target and trigger them at half-second intervals (more than enough to not flood the buffer). Of course, putting the images together would be complicated. As would aiming the beast. :)
posted by Civil_Disobedient at 4:22 PM on September 26, 2006


Don't forget you have the problem of parallax if the senors aren't all seeing the same image. 15 sensors arranged in a grid would induce a horrible flicker.

A split prism would work, but reduces the light available to each sensor. A rotating mirror would also work but you'd have to watch out for side effects like those that occur with a 35mm SLR shooting at higher than it's sync speed. 15 sensors mounted on a drum moving past the lense might do the trick.
posted by Mitheral at 4:56 PM on September 26, 2006


This sound very cool. Let's just hope Jannard can keep his head out of the microwave.
posted by idontlikewords at 10:47 PM on September 26, 2006


15 sensors mounted on a drum moving past the lens might do the trick.

It would be a beautiful Rube Goldberg device to behold, but I think we both know it couldn't possibly be this mechanically complex and still stand up to the rigors most DPs are going to require of it.

Don't forget you have the problem of parallax if the senors aren't all seeing the same image. 15 sensors arranged in a grid would induce a horrible flicker.

Parallax wouldn't be that bad. Of course, I think it will be far less than 15 sensors—you could probably hit the 4k numbers @30 fps with 6 sensors, so the distances won't have to be that far off. Even if you had a bunch of chips right next to each other, the angle of anything greater than, say, 2 feet is going to be imperceptibly small. Particularly if it's playing once every whatever-th of a second. There might be a very, very subtle blurring effect, however.

Alternately, you could always split the image a few times and up the sensitivity of the sensors to compensate for the lower light that would be hitting the array. That way you get (virtually) identical images on all the chips, no matter how spread apart they are. Either way, I find it far more impressive that they've come up with a way to syncronize this kind of bandwidth in an uber-buffer, then dump that gargantuan load of data downstream.

As a photographer, I've long thought that our days were numbered: the Canon 1D Mark 2 can do 3504 x 2336 pixels at 8 frames per second. You get four of those and you're basically there. Attach each to its own hard drive that can move faster than 10k rpm with a good cache and you just have the synchronization problem to address. They might use some kind of custom file system built for RAID, I dunno. Sure would be nice to see it, as I'm sure some other people were probably thinking.
posted by Civil_Disobedient at 4:58 PM on September 27, 2006


Parallax wouldn't be that bad. [...] Even if you had a bunch of chips right next to each other, the angle of anything greater than, say, 2 feet is going to be imperceptibly small.

I'll admit that I didn't work out the trig to be sure, but I am not sure thhis claim makes any sense. Optical rangefinders, such as those found on Leica and Fed/Zorki would not work. If you scale down to the photosite image stabilization wouldn't work either.

Either way, I find it far more impressive that they've come up with a way to syncronize this kind of bandwidth in an uber-buffer, then dump that gargantuan load of data downstream.

If they really made it work, and arne't just faking it for the demo (for which there is a long and honorable tradition), then the datacom is really the most interesting part of the camera. From what everyone else is guessing, it sounds like they did it with massive compression. That means that it will be difficult or impossible to recover the full dynamic range that the sensor apparently offers.

I'd put my money on this being a fake, with the real product still 2-5 years off.
posted by b1tr0t at 5:22 PM on September 27, 2006


From what everyone else is guessing, it sounds like they did it with massive compression.

That would likely have a dedicated encoding chip, or four, but certainly not outside the bounds of reason. JPEG2000 compression has an option to be lossless, so you're just reducing the overhead of the datastream, not changing the data.

Never underestimate the power of crazy people with enormous wealth.
posted by Civil_Disobedient at 5:52 PM on September 27, 2006


That would likely have a dedicated encoding chip, or four, but certainly not outside the bounds of reason. JPEG2000 compression has an option to be lossless, so you're just reducing the overhead of the datastream, not changing the data.

So is it wavelet or lossless? Wavelet compresison could yield good quality at a 10:1 compression ratio. Lossless tends to top out at around 2:1. If we drop down to 30 fps, then the raw bitrate drops from 17 Gb/sec to 8.5 Gb/sec. With lossless compression, we are down to 4.25 Gb/sec.

It is certainly possible to manipulate data streams like this, but it won't be cheap or easy. I still think the cost of the disk and rendering farm will far outweigh the cost of the camera. Throw in the salaries for the various operators, and this starts to look more like a nice savings and less like a disruptive technology.

Interesting post, whether or not the product is real.
posted by b1tr0t at 6:10 PM on September 27, 2006


I think if one was going to split the image up for capture by multiple sensors, one would probably use three sensors, one for red, one for green, and one for blue. You'd get triple the I/O bandwidth of a single sensor, you'd be able to get away with lower-resolution sensors, and you could reap significant image-quality advantages. Even some current consumer-level video cameras work this way.
posted by Western Infidels at 2:52 PM on September 29, 2006


« Older "The streets of 2030's New York remain the only ve...  |  A case of Horlicks for 1,000 -... Newer »


This thread has been archived and is closed to new comments