The first REAL digital cinema camera?
 

Even before the Viper has made it into regular production use, the Holy Grail of digital acquisition may be eclipsing it; or not. Guess we'll have to see when this baby premieres at NAB. Can't wait.

Check it out!

Wow, how about that... most interesting! I wish they had given their booth number for NAB; I couldn't find it on the website. This will definitely go on my "must see" list for NAB. Good find, Charles... I'm copying your post from the Film Look forum to the DV News forum. Yowza.

Must be special, I'm getting 500 and Server Too Busy errors.

Is this cross-posted?

<<<-- Originally posted by Robert Knecht Schmidt : Must be special, I'm getting 500 and Server Too Busy errors.

Is this cross-posted? -->>>

Ditto for me.

Ivan

Wow. That really looks awesome. A chip the size of 35mm film! Let's face it... it's only a matter of time before Digital replaces Film. Granted, we won't see anything of this magnitude in the XL2, but it's paving the way...

Just got in. The optical viewfinder will be a big hit with cinematographers, though I think its actual utility a bit dubious in the presence of a good video monitor. They also claim higher dynamic range, lower noise, and better sensitivity, but provide no white paper. 14 bits/channel means the camera will have ~14 stops of (luminance) exposure latitute, 254.6 dB quantization error SNR, and be able to output ~4.4 trillion colors. Will this be a three-chip camera or a yucky color filter array? Still only 60 fps max frame rate--you probably won't get much better slo mo out of this cam than a CineAlta. I wonder what they mean when they talk about "a traditional film camera feel." Does that mean the packaging has molding for the mags?

Here's to hoping the NAB demo includes an operational camera and not just a form factor mockup.

Hard to imagine that now, just a few short years after the first megapixel still camera came out, we can now see the advent of a camera that is basically capable of taking 60 8megapixel stills in one second. Just amazing.

Looks AMAZING... A few questions arise though...

1. where will it store these images on?

4046x2048
x5.25 -> (14 bits x 3 colors) / 8 bits per byte
x60
= 2,610,155,520

That is 2 GB per second... (uncompressed). If you
want to store this (compressed I assume) and want
to edit it you will need to have a massive system

(imagine 90 minutes uncompressed with this:
14,094,839,808,000 -> 14,1 TB of information.. WOW)

2. What type of compression and at what levels?

3. How many CCD's? They seem to be talking a lot in CCD
instead of CCD's....

Too bad they didn't put up a picture the machine recorded,
would have been awesome. Find it also a tad strange this
awesome camera (supposedly) just came falling down and
suprised everyone...

Guess we will have to wait and see.... Looking forward to it!

I'm predicting that when the "price" comes tumbling down, it'll crush all of us admirers looking up expectantly.

Robert:

<<The optical viewfinder will be a big hit with cinematographers, though I think its actual utility a bit dubious in the presence of a good video monitor.>>

An optical viewfinder allows the camera operator to do his job, which is to analyze everything in the frame. A while back I posted some incredibly long-winded list of duties in the operator position--maybe someone else can find that thread? Anyway, it's just not possible to do as good a job with a black and white CRT viewfinder (I won't even go into the color LCD as is found on stock on the XL1). Even a decent 7" onboard LCD provides nowhere near the information needed. A 14" broadcast monitor would rival an optical viewfinder, but where is that going to mount on the camera...? Really, it's not about vanity or tradition but all about catching the problems and improving the results before it commits to film/tape/digital media/oh boy, is it messy to talk about this stuff empirically...

Yes, I see what you're saying, Charles. Yet isn't the principal advantage of the optical viewfinder better gauging of focus? This doesn't help the operator much. If the operator's main task is composition, wouldn't he/she want to see the digital data stream (the image that's actually being recorded) rather than the approximation offered by the optical viewfinder?

Also, I'm not a cinematographer, but I've looked through a few Pana 35 finders on location and they tend to have optical defects--for example, some miniscule misalignment will cause one half or corner to be slightly out of focus, or there will be what looks like waterspotting or dirt, or so much light will be cut that the image looks darker than what the film is actually "seeing." Now for film cameras the video monitor tends to render an even further degraded image, but for CineAltas the monitor gives a nice clear picture, and it shows what you're actually putting down on tape. I'm guess I'm just a big fan of the monitor tent.

I think the post you are reffering to, Charles, is this one:

http://www.dvinfo.net/conf/showthrea...ator#post30132

Definitely "wow". I'd like to see how that thing looks. Transferring the footage to film should definitely look better than Star Wars Episode 2, which was shot on a standard HD camera and then cropped to fit the aspect ratio. DLP projection in theaters isn't even as good as HD yet (!). But regardless this is a camera to watch for.

I do find it humorous that the web site for the camera says it is targeted at cinematographers, not broadcasts. Yet they will be at the NAB (National Association of Broadcasters).

I guess soon we'll know if digital can _ever_ match 35mm.

Does anyone know what the resolution of 35mm would be in pixels? Does this camera ~(4k x 2k) match the resolution of 35mm?

Also I can't begin to imagine the processing and storage costs of such high resolution video, because to match film you'd have to not use a very high compression rate. Seems like it'd be at least a gig per minute. Wow.

When I was doing some research on a high res. digital still camera I found a statement in a European Photography magazine that you would need 24 mega-pixels to give the same resolution as 35mm film. So I am guessing they would need close to the same for this new “super” camera.

Alex

If you had 24 megapixels I think you'd probably be doing OK. I don't think you'd be sad at the lack of detail you were getting. :)

When motion pictures are scanned in for CGI work, lots of times the are scanned in at approximately 2k resolution (The Two Towers, for instance). If you look closely at TTT projected on a big screen, you can see some jaggies and sharpness artifacts. Sharpness artifacts look like white "halos" around contrasty edges. Some films are scanned in at 4k resolution and look much, much better. Stuart Little was scanned in (and rendered) at 4k resolution for example. Star Wars Episode 2 used a run-of-the-mill Sony HD camera modified to use professional lenses, and the backgrounds were all rendered at around 2k resolution (higher than the camera). So this camera *should* look damn good. But you will only be able to tell when it is transferred to film, as digital projection cannot even begin to approach this yet.

The question is are there any editing softwares and hardwares set up to handle this format yet?

Robert:

Focus--yes and no. Yes, the ground glass does deliver an image that is focus-critical, and the operator can and will make a determination if a given shot is soft, but it is ultimately the camera assistant who is performing the task entirely by instinct and feel.

<<If the operator's main task is composition, wouldn't he/she want to see the digital data stream (the image that's actually being recorded) rather than the approximation offered by the optical viewfinder?
>>

Compositionally, the optical viewfinder is EXACT, not approximate (unlike that rotten color viewfinder for the XL1!). I would expect an optical viewfinder for a digital camera would have framelines etched for TV transmission as well as TV safe, and of course you would be able to see outside those areas to watch for boom mikes and anticipate people entering the frame etc. That advantage cannot be understated.

As far as Panavision optics being defective as you indicated, I'm very surprised, Robert. Panavision, being a rental-only operation, keeps very high standards on their gear. Their viewing optics are usually flawless. I've never seen one with alignment/focus issues yet. Occasionally oil will get on the groundglass, causing that spotting issue you may have seen, but it is easily removed and cleaned (the assistants weren't on top of it, I guess!). There are various vintages of camera bodies available for rental, some of the models being nearly 30 years old (such as the Panaflex Gold, which was used for the first time on Sugerland Express, Spielberg's first picture). The viewing optics are not nearly as good on that camera as on the more recent Panavision offerings such as the Platinum or Millenium, which are plenty bright and sharp. Then again, imagine looking at the viewfinder of a 30 year old video camera!

The monitor tent is a great thing. It is the final word on the image being layed down, no question. But as far as operating the camera, given a choice between a good optical viewfinder and a black-and-white 1" CRT viewfinder, or even a standard resolution 7" LCD, you'll see much more useful information in that optical viewfinder. The latest HD onboard monitors are starting to approach a level of viewability that rival optical finders, but in bright sunlight, it's still better to have your eye in an eyepiece.

Curtis,

If they can already scan in film at 4K and handle it, yes, there
must be software out there thta can handle the output from
this camera (the resolution, not talking file formats here).

Any price on this baby yet? My guess is you could buy a small country for the same price.

No pictures of the kit, but interesting nevertheless.

It reminds me of this, the part about optical viewfinder and PL mount lenses: http://hugecgi.com/cgi-bin/ibc_dailynews1.cgi?db_id=21406&issue=5
Ikegami are doing the back end digital it seems. But you've got to ask yourselves what are Arri and Aaton planning. My guess will be some sort of digital back as well, though I wouldn't put it past Arri to bring out a whole digital cam based on the SR. They do after all have experience with digital already with their Arri laser post facility in Germany.

>>>>>>>>
Looks AMAZING... A few questions arise though...

1. where will it store these images on?

4046x2048
x5.25 -> (14 bits x 3 colors) / 8 bits per byte
x60
= 2,610,155,520

That is 2 GB per second... (uncompressed).
<<<<<<<<

If this is a 1 CCD camera, then recording the pixel data uncompressed in raw format from this camera takes only a little more storage space than it takes to record it from a 3CCD HD camera. Here are 8Mpixels versus 6Mpixels. If you record 60 frames/s, of course it is more data, but you shoot much less time
for slow motion.

So if framerate is 60/s, then data is less than 1GB per second. For 24frames/s the rate is less than half GB.

Balazs

There are quite a few companies preparing to launch digital movie cameras with more than normal HD resolution. I put here a few links about them:
This link is about a Lockheed Martin 12 megapixel cine camera which is still at an infantile stage:
http://www.uemedia.com/CPC/article_2980.shtml
A JVC movie camera utilising 3 pieces of 8 megapixel Photon-Vision chips:
http://www.towersemi.com/press/apr0402.html
About a Sony 10 megapixel camera for the shooting of Episode III:
http://www.forbes.com/asap/2002/1007/013.html
An Olympus 8 megapixel movie camera, coupling four 2-megapixel CCDs. It has already been demonstrated:
http://www.olympus.co.jp/Special/Info/n020522aE.html

Balazs

Balazs,

The 1 versus 3 CCD's has nothing to do with the actual data
stream, why? Because it will still have three colors: Red, green
and blue. The difference between 1 and 3 is that with 1 chip it
outputs all of the colors at the same time (but there are still 3
there!!). With the 3 chippers each chip outputs its own color
(actually, black and white).

So this will result in the same amount of bits (assuming they
both allocate the same amount of bits for each channel
ofcourse) for each pixel.

Cheers!

Rob, I think Balazs__ is correct. 1-chippers user a CFA post-processing scheme to derive RGB from luminance data from a CCD overlayed with a known-pattern color filter array, typically [GRGR/BGBG].

Thus raw data from a 1-chip camera is actually 1/3 the size of raw data from 3-chip cameras--the colorization is essentially a post process, and thus the color data doesn't need to be stored.

(Foveon chips notwithstanding, of course, and "pixel shift" schemes such as the one employed by the XL1 make things more complicated still...)

Robert,

Okay. My bad then. But, I was talking about the format they are
laying back to tape/harddisk/whatever. If we take the DV
cameras as an example the DV spec only allows for one format.
So this format has the same bandwidth with a one chip or three
chip camera! I suspect the "professional" systems to do the same.

But if the raw datastream is working like that (which would mean
that actually the resolution is lower than a 3 chipper as well
since they are using less bits then a 3 chip would -> lowering
(color) resolution) I have said nothing :)

This may sound stupid, but now that the formats are close to the same size ( canon's 11 megapixel CCD is the size of 35mm film) I was wondering what the size of one pixel on a ccd is compared to a grain of film?

A ONE CHIP camera? Why not 3 chips? Isn't 3 chips immensly superior?

Yes the resolution at that 4K would make it equivalent to film, but still less resolution than the human brain and eye combo....

Todd's right, of course, but camera technology can't be appropriately compared to the HVS. Our clarity of visual perception drops off significantly outside of a small foveal region. Designing cameras to behave like a human eye would mean chips with high density of small pixels in the center and fewer, blockier pixels on the periphery.

Also there's a significant amount of post-processing done that makes our images much appear to us to be much better than they really are.

Chris.... They do have a booth number up now.

Its SU6245

Go take a look and report to all of us stuck in PAL Land :-)

Hans Henrik

One of the misconceptons is that 3 CCD chipset has better low light sensitivity than a single CCD. This is not the case. There is certain amount of light coming from the lens and in a 3CCD system you split it into 3. Overall the same amount of light is reaching the chips.

3 chip CCD system has more accurate colors. You pick up luminance and the complimentary colors at the same time in the same point during the scan. In 1 CCD system there is a shift among the 3 signals.

For RGB signal 3 CCD's are a must, for 4:2:2 comression there is a significant improvement with 3 CCDs. For 4:1:1 one CCD is fine.

"Overall the same amount of light is reaching the chips. "

Since there's always some transmission loss in a medium (the mediums in this case being a prism and a color filter), one should find the effective ASA of a three-chip camera is reduced by the splitting--i.e., less light reaches the CCDs than with a CFA 1-chipper. (Of course it could be a nominal amount in comparison to other factors: the optics, the iris, etc.)

The loss in the prism is minimal. The 3-chip camera has 3 filters; the 1-chip camera has individual filters one ach pixel.

I think the future will be single CCD designs. Canons CMOS chip in the EOS 10D is a 6 Mega pixel chip that gives 35mm film quality up to about 16 X 20. That camera cost $1,500 and Canon makes their own CMOS chip. If they applied a little of that technology to the next generation XL series, just imagine what we might be shooting.

In still photography 1 CCD is fine. You read the RGB pixels, store that information and print that information. With moving images you use more compression. If you sense in RGB, record in RGB, and play in RGB, one CCD would be fine. The way they process the color signal, by recording one lumimnace signal and 2 compressed color components, the 3 CCD have an advantage, because at any instance you read 3 colors and from that you create the 3 signals. With one CCD you can't read the colors simultaneously. So unless you get to CineAlta SR with 4:4:4 RGB stream, 3 CCDs are an advantage.