CinemaTech has news on the new ARRI D-20 digital camera, just announced last week. And here are more details, directly from ARRI.

The ARRIFLEX D-20 camera features a a single 6-megapixel CMOS sensor that precisely matches the physical size of Super 35mm film aperture. Therefore, the D-20 utilizes standard 35mm film camera lenses.
This camera has been years in the making. They announced plans to build the D-20 back in 2003, and the first test shoot of the D-20 prototype was just after NAB 2005. A variety of well-known cinematographers have been able to use the camera, and responses seem to be positive.

One comment in particular caught my eye. Sam Nicholson says: “The depth of field and the lens choice it gives me are great, but the real advantage is the optical viewfinder. Judging focus is usually quite impossible on an electronic camera, unless you have a 30,000 USD monitor with you. The optical viewfinder of the D-20 will essentially save me 30,000 USD per shoot. Focus is the one thing that no one notices until it is out. In HD, often things get soft and you cannot tell. I would look to use the D-20 for some high quality keying on set for virtual backgrounds.”

ARRI has this to say about the choice to use CMOS in lieu of CCD’s:
“CMOS sensors inherently have superior power efficiency and a natural blooming immunity, plus it is possible to read out any portion of the sensor at any time. This has a wide range of advantages, including the ability to read out high frame rates despite the high pixel count and the ability to run speed ramps. It also means that the recording format can be freely chosen, so it is possible to trade spatial resolution for frame rate. Because CMOS is essentially a more flexible technology than CCDs, ARRI can experiment in the future with new and sophisticated features like higher frame rates or a double read-out of each frame to further increase dynamic range.”

They go on to say that the sensor and the camera data bus, are “prepared for frame rates up to 150 fps.”
Neato.

ARRI has really put a lot of thought and research into this solution. One such example is the choice of Video or Film output modes. In Video Mode, the camera internally processes the 2880 x 1620 image and outputs as 1920 x 1080 to a “variety of standard HD video signals for different recording formats, including HDCAM SR”.
In Film Mode, the D-20 outputs raw data from the CMOS sensor to a format that must then be processed offline using “complex 3D Look Up Tables (LUTs) before it is usable or even viewable.” I would liken this to still image RAW processing. It takes more space and requires processing, but you also retain more image information and quality. The good thing is that you can still monitor the live HD output as well when you are in Film Mode.
Another interesting feature is that ability store grading parameters as metadata piggybacked along with the raw, unprocessed image data. Pretty sweet stuff.