Jump to content

Aaton XTERA HD camera


Mitch Gross

Recommended Posts

JP of Aaton just announced it over on the CML, so I consider myself freed from any NDA.

 

The Aaton XTERA is an extension of the Aaton XTR camera family. It uses the standard XTR camera body, so one can reuse an old (1984 or later) camera body. The film gate is replaced with a guide channel and the film mag replaced with a data back/electronics module. Where the film gate used to lie on the mag now rests a single chip CCD or CMOS HD sensor the approximate size of the Super-16 film frame. On what was the take-up side of the film mag now lies a complete electronics module and on what was the feed side stores an isolation-mounted hard drive similar to the setup in the Aaton Cantar sound recorder. The hard drive is field swappable for "reloading" the camera and all the parts are easily upgradeable to prevent camera system obsolescence.

 

While it is currently unclear if the system is a permanent retrofit to the film camera or a swappable unit, the fact remains that the use of the old camera body with its excellent optical viewing system, Super-16 lens mount and available accessories means that this will be an accessible and relatively inexpensive way to enter into the world of high quality HD production. Single chip sensors using Bayer filtering and cine-style lenses is the way of the future as evidenced by a number of manufacturers including Kinetta, Arri, Dalsa and now Aaton. Prisms cause color fringing and fine detail loss. High quality optics are already plentiful for this format.

 

No pricing yet and the camera is expected to debut at NAB. The Aaton XTERA camera replaces the D-minima project that others may have heard about. While a availability is not yet known, this is a real camera from a real company and not "vaporware."

 

Look out Sony and Panasonic. This is the way the industry is headed.

Link to comment
Share on other sites

  • Premium Member

Hi,

 

Bayer filtering, mutter, grumble. Just 'cos they can't fit a three-chip block back there! Dalsa's claim of 4K resolution for their Origin camera is highly suspect on the basis that it uses a 4000-line sensor that is then Bayer filtered, providing a true RGB resolution of what, 3000 lines max with every mathematical gain in the world. What's the native resolution of this Aaton thing, Mitch?

 

Phil

Link to comment
Share on other sites

The first generation will likely be true 1080p (1980x1080), the same as the Kinetta and likely using the same sensors. There's not a lot of choices for what chips are available out there. But as noted, both the Kineeta and the Aaton concepts are modular and upgradeable, so a 720p chip could be used for TV production, a true 2k chip could be an upgrade when they become available in the next year or two and so on.

 

The way chip resolution is counted is something that PR people can try to fumble with but not techies. You and I were both in the room when we cornered John Coghill from Dalsa to explain the company's definition of 4k with Bayer. It is a far greater resolution than standard HDCam or even 2k, and it's not a misnomer like the errant Thomson PR that stated the Viper's chips at 9 million pixels, neglecting to mention that's before Bayer.

Link to comment
Share on other sites

There's a lot of good reasons to eliminate the 3 chip block in favor a a single Bayer-filtered sensor. Prism fringing is a major issue with the Sony cameras as evidenced in Geoff Boyle's comparison tests on the CML website. There are plenty of compromizes from having the video lens design of three separate color depths, and a truly sharp, clear color image is very difficult. Digital still cameras happily use single chip designs for maximum performance in the highest end cameras, and the only reason this has been avoided up until now is fresh rate issues. The quality of CCD or CMOS single chip designs at this point is surpassing three-chip prism systems and this is the way the industry is headed. That and the fact that the excellent cine lenses so readilly available can be used with them makes these chips the way to go.

Link to comment
Share on other sites

  • Premium Member

Hi,

 

> The first generation will likely be true 1080p.... 1920

 

I take this to mean that the image sensor has something like 2500 pixels on a line and uses the standard green-heavy weighting, depending how you're recording it. This I'd give them. But to create a 1920x1080 sensor BEFORE Bayer then call it 2K is not valid.

 

However, this may or may not affect it actually being any good in use; this is a rather theoretical concern.

 

Phil

Link to comment
Share on other sites

Yes, 1080p after Bayer.  That's only fair.

Is that a "for sure" thing, or are you just speculating that will be the case? And are these custom chips for Kinetta and Aaton, or something like the CMOS ProCam 3560 from Rockwell Scientific (if so, the ProCam is only 1920x1080, so that would be less resolution after bayer)?

 

Either way, at 1920x1080 with a bayer sensor, the resolution should still be pretty good. The resolution in 16x9 is very close to the old Nikon D1, which was very good for a 3 megapixel camera in RAW mode. The JPEG conversion in the camera left something to be desired, but the RAW output had fairly impressive resolution. And yes, it's 3 megapixels, but that is in 3:2, once you crop for 16x9, it's only got about 4% more resolution, which is pretty negligable IMHO. So if it can achieve the resolution of the D1 or the Canon D30, then we're in pretty good shape.

 

I think I'll bring an ISO 12233 res chart with me to NAB! :D

Link to comment
Share on other sites

Yes pixel count comparisons between Bayer filter and 3 chip are a somewhat apples and oranges comparison at the moment.

 

And quoted output resolutions could be derived from a sampled lower (or higher) res image so they are not to be trusted either.

 

Also like a prism system there are comprimises with Beyer colour fidelity that are subject dependent, which one is better? is another apples and oranges comparison.

 

 

Mitch which prism test do you refer?

 

 

Mike Brennan

Link to comment
Share on other sites

Mitch which prism test do you refer?

Mike,

 

Go to cinematography.net and check in the achived topics for something like F900/Viper/film comparison or F900 test or something to that effect. Geoff Boyle shot some tests that clearly illustrate how Sony's prism blocks can produce horrid color fringing, especially on the reds. It's a tough job making a clean alignment with a prism and even tougher in HD. Thomson seems to do a better job of it. The Varicam wasn't a part of the test but if it were I figure it would do about as well as the Sony.

 

Sorry I've been absentee to the discussions as of late. We've a brand new baby girl who is consuming all of my time and I'm happy to let her.

Link to comment
Share on other sites

  • Premium Member

Maybe someone should try using Kodak sensor technology? ;)

 

http://www.kodak.com/global/en/digital/ccd/sensorsMain.jhtml

 

http://www.kodak.com/global/en/digital/ccd...=0.1.4.12&lc=en

 

http://www.kodak.com/global/en/digital/ccd...0.1.10.10&lc=en

 

FWIW, Kodak inventor Bryce E. Bayer invented the Bayer filter array: B)

 

United States Patent 3,971,065

Bayer July 20, 1976

 

--------------------------------------------------------------------------------

Color imaging array

 

 

Abstract

A sensing array for color imaging includes individual luminance- and chrominance-sensitive elements that are so intermixed that each type of element (i.e., according to sensitivity characteristics) occurs in a repeated pattern with luminance elements dominating the array. Preferably, luminance elements occur at every other element position to provide a relatively high frequency sampling pattern which is uniform in two perpendicular directions (e.g., horizontal and vertical). The chrominance patterns are interlaid therewith and fill the remaining element positions to provide relatively lower frequencies of sampling. In a presently preferred implementation, a mosaic of selectively transmissive filters is superposed in registration with a solid state imaging array having a broad range of light sensitivity, the distribution of filter types in the mosaic being in accordance with the above-described patterns.

 

 

--------------------------------------------------------------------------------

Inventors: Bayer; Bryce E. (Rochester, NY)

Assignee: Eastman Kodak Company (Rochester, NY)

Appl. No.: 555477

Filed: March 5, 1975

Link to comment
Share on other sites

Are these sensors fast enough though to support up to 60p at 1920x1080 or 2K resolutions?

 

I think the Kodak DCS cameras look great, but their frame-rates are too slow to support a digital cinema camera-which is too bad-at 14 megapixels that would be one heck of a camera :)

Link to comment
Share on other sites

This reminds me of a question I've been meaning to ask.

 

Is it the image sensor (CCD/CMOS) that limits the successive capture rate, or the electronics that process the data from the sensor?

 

i.e., these 14MP sensors cannot be used for HD motion picture cameras because a) the sensors physically cannot capture 24 frames in one second, or b) the electronics cannot process such a large amount of data?

 

I hope this makes sense...

Link to comment
Share on other sites

  • Premium Member

Hi,

 

This is a slightly complicated subject, but basically, for a CCD noise is proportional to frame rate. Eventually it comes down to the fact that a CCD is an analogue device with a single output amp through which all the data on the sensor must eventually pass.

 

Each photosite accumulates photoelectrons (an electron motivated by the impact of a photon on the charged region) while there is a potential across the electrodes. You can control the time during which this potential is present, and thus the integration time. CCDs are fine for low light imaging with multiple-second exposures (notwithstanding the fact that they will collect cosmic rays VERY effectively.) They are also pretty good for short exposures, which is of course half the equation of high frame rate... so long as you only want to do it once.

 

When you come to read the image from a CCD, each vertical column of photosites (vertical usually because it's the shorter dimension) dumps its charge into the one below sequentially, and the row which falls off the bottom of the sensor ends up in a serialising bus. There is a limit to the rate at which this electron bucket brigade can operate, but it's very large, and isn't the limiting factor. The problem occurs when it comes time to try and measure the values in the serialiser, which is done by allowing the electrons to charge a capacitance of known value. The time it takes for the charge in the capacitor to stabilise is measured and by this we calculate the charge on the pixel.

 

In order to read the device rapidly, the value of the capacitance must be lowered, so that the electrons can charge it more quickly. This is bad news, because we're only talking about reading a few hundred to a hundred thousand electrons, and that's therefore an extremely small period of time. In practical designs this becomes a very small signal for the output amplifier to work with. In CCDs, frame rate is therefore directly proportional to noise.

 

CMOS is better for this kind of thing. CCDs are based on a rather old semiconductor technology, in which the sensitivity to light was actually a problem. It is difficult to make high-performance, modern devices using the MOS technology that CCDs rely on. However, it's possible to fabricate advanced semiconductor devices (amplifiers, DACs, logic) onto the same substrate as a CMOS image sensor. The critical advantage here is that you can choose to read out data more or less arbitrarily from anywhere on the substrate, rather than having to funnel it all down to a single output amp. It's possible to bin charge to almost anywhere on a CMOS sensor and read it out in any way that's convenient, so you can make a high frame rate sensor by creating multiple output zones spread across the physical real-estate of the sensor area. Interestingly this is kind of how the CCD in the F900 camera is done, but this is only possible because the F900 has a lot of off-sensor correction technology to match the quarters of the image to one another.

 

CMOS is good for other reasons. Because of the ease of fabrication, you can make more or less the entire camera on one chip - the sensor sites, summing and output amps, even the DACs and timing generators. This makes it cheap, and the imaging sensors in optical computer mice and some webcams are made this way. You can also integrate certain kinds of noise-cancelling measures directly behind the photosites, removing the need for a lot of off-board processing (CMOS television cameras could do lens shading adjustments directly on the sensor.) CMOS image sensors have greater dynamic range than CCDs, and a larger fill factor - up to 60% of the total surface of a CMOS sensor is actually photosensitive, as opposed to 40-50% of a CCD, increasing sensitivity. Decent, precision imaging CMOS sensors will be much better than the CCDs we have now.

 

I won't go on much more as this is now only peripherally on-topic, but there's plenty of reading out there.

 

Phil

Link to comment
Share on other sites

John P wrote

maybe someone should try using Kodak sensor technology?

 

John in respect of Bayer filtering is there a rule of thumb we can use to make a reasonable comparison between resolution of single chip Bayer filter imagers and 3 chip prism cameras?

 

For example does a 4k single chip bayer actually have 3k resolution?

 

 

Mike Brennan

Link to comment
Share on other sites

  • Premium Member

Hi,

 

Depends what exactly the pattern is, but usually they're green-heavy, since green is the largest contributor to the higher-definition luminance channel in YUV systems. Therefore, an RGBG Bayer pattern would have 50% of overall resolution in green, and 25% in red and blue. Now in the end you do have slightly better numbers than that because of the way the RGB data is reworked to produce a recordable YUV stream, but it's not much better than 50% overall. To call it a true 2K sensor with Bayer filtering, I'd want 4K on the chip with a 50% green pattern.

 

Phil

Link to comment
Share on other sites

  • Premium Member

What Phil said. :)

 

As I understand it, any camera using CMOS sensors is also likely to have a higher "fill factor" with more active area sensitive to light. A rule of thumb is that human vision gets 60% of sharpness information from the green, 30% from the red, and only 10 % from the blue. That's why Kodak's Bryce Bayer put most of the weighting on having green pixels.

 

http://www.kodak.com/global/en/professiona...8.3.18.30&lc=en

 

Here are some sample images from the 14 megapixel Kodak DCS Pro 14n digital camera:

 

http://www.kodak.com/global/en/professiona...8.3.18.16&lc=en

Link to comment
Share on other sites

  • Premium Member

Hi,

 

> A rule of thumb is that human vision gets 60% of sharpness information from

> the green, 30% from the red, and only 10 % from the blue. That's why Kodak's

> Bryce Bayer put most of the weighting on having green pixels.

 

I was working off the 601 specs for calculating YUV from RGB, which make a similar provision for similar reasons. It's worth mentioning a good practical reason to weight in favour of green with modern cameras is that data from the green image is proportionally strongest in the calculated luminance, and this data is stored at the highest resolution in many digital video systems. I don't know which came first here, but this is all based around the characteristics of human vision, certainly.

 

Phil

Link to comment
Share on other sites

Here are some sample images from the 14 megapixel Kodak DCS Pro 14n digital camera:

http://www.dpreview.com/reviews/kodakdcs14n/page16.asp

 

(scroll to the bottom of the page, the 400ISO test)

 

Mr. Pytlak, the images produced from this camera are certainly beautiful, but why such a heavy denoise filter? Maybe it's just me, but I think I would rather smooth out noise in Photoshop than have the "watercolor" look.

 

Can this feature be disabled?

Link to comment
Share on other sites

It seems this review is as of May 2003. No new sensor module. Oh, and I read this from the link you gave:

 

What about the issue of "noise" at higher ISOs?

Like many other professional digital cameras, noise in images taken with the DCS Pro 14n Digital Camera increases when shooting at higher ISO settings or in lower-light conditions. Images are greatly improved, however, once the user becomes proficient with the camera's extensive noise-reduction settings. We expect that some photographers will have to "experiment" with the camera to familiarize themselves with the settings, as with any high-end imaging tool.

 

I guess the review team didn't take the time to tweak any settings! :P

Link to comment
Share on other sites

(hello, first post)

 

I realize that NAB 2004 is only days away - but does anyone have an idea of what these solutions will end up costing even if it is a range? I've been following the development of the Kinetta for a little while (which led me to this forum) and I still haven't figured out if it is actually going to be selling after NAB or is it just showing? Same question about the Aaton.

 

Mark

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...