1080P 60 from Texas Instruments
 

Looks like camera's are going to get a boost to 1080P 60 this year...

Texas Instruments delivers H.264 1080p at up to 60 frames per second decode or eight channels of D1 encoding with flexible new DaVinci(TM) video processor - Yahoo! Finance

Doesn't look like that chip is for cameras, to me.

It's a step in the right direction, it's all about the implementation of the codec that passes through the processor and which major player comes out with it first...

Jack, Most of your dream system can be had with a Sony EX1/EX3 coupled with a Nanoflash HD-SDI recorder to compress 4:2:2 to 160 MBits/sec. The Nanoflash will hold two 32 GB or 64 GB flash cards. Is it time to update your dream system?

I'm with you on wanting 1080p60.

The EX1 is a rolling shutter system. Plus I haven't mentioned that alongside that the "dream" camera have 1080p60, it should also have cranking to shoot at 1080p60 and record in 1080p24.

If SxS can write 200Mbps or above (which seems possible with the cranking modes I've seen), it would be better suited for a 200mbps codec than CF at the moment.

And with 10-bit AVC-I at 100Mbps possible, I think that doubling that data rate and doing 10-bit 4:2:2 1080p60 would be the "dream" codec.

Edit: Well, heavens to betsy... The Dream Codec is here: AVC-Ultra (even though it's shrouded in secrecy, which is why I never knew about it... The blog post on this on the Panasonic site has been deleted...)

OmniVision 1080p 60 sensor
 

"OmniVision’s new 14.6-megapixel HD sensor uses its most advanced 1.4-micron OmniBSI™ pixel architecture to achieve optimal performance and sensitivity in the industry’s smallest format. The 1/2.33-inch OV14810 has an active array of 4,416 x 3,312 backside illumination pixels operating at 15 fps in full resolution, or delivering full 1080p HD video at 60 fps, using a binning feature to achieve higher sensitivity. In full HD video mode, the sensor also provides additional pixels for electronic image stabilization (EIS)."


OmniVision

Nowhere does that mention CCD or CMOS. Plus it's a 1 chip solution. 3 chip solutions much better suit video applications.

Pay attention.... :-)

Not so sure about that one
 

The TI DaVinci series wouldn't normally be used for cameras. For one, most camera makers these days do their own video DSPs (eg, Canon's DIGIC, Sony's BIONZ, Panasonic's Venus, JVC Gigabrid, etc). The second is that this is doing playback, not encoding. They're claiming a full 1080/60p decode, which makes sense for the kind of portable devices generally built with the DaVinci. Or or multiple SD encodes... like a security camera interface application.

The DaVinci series have been used in quite a few PMPs, set-top boxes, wireless tablets, and even more special purpose embedded devices. TI used to make chips for digital still cameras, but they were specialized versions of their SOC (system on chip) products. And of course, their OMAP line, also similar, has been the applications processor on many cell phones.

The DaVinci has always been a system-on-chip. Like most TI SOCs, it includes an ARM processor, a TI DSP, and various acceleration engines to enable fast encoding or decoding of video. I don't think you're getting full 1080p encoding on this kind of chip.... yet. They claim to do 1080/30p encode and decode on this one (the latest DM6467 release), and show a roadmap with 1080/60p support in the future.

Which was the one reason I didn't go with that camera.

1080/50p/60p would be very welcome here too. I hate having to drop to 720 for 50p/60p shooting.

1 chip works for the Red and F35:)

One vs Three
 

Yeah... well... there are some very good one-chip cameras, and some very poor three chip cameras. But the basic idea of 3-chip is sound: for the same chip size, you'll get better video with three chips. This works because, within a format (eg, 4K, HD, etc) extra pixels cost you light sensitivity, and once you're at 1:1 with a 3-chipper, or 4:1 with a one chipper, there's no advantage to adding pixels.

A Bayer filter array is always going to cut out 2/3 of the light hitting your one sensors, so a 3-chip camera with 1/N" sensors get nearly 3x the light of a single chip of 1/N" sensor.

When you get to something like the Red camera, you have a very expensive large sensor. This would multiply the cost of the prisms and additional sensors, just as it multiplies the cost of lenses. As well, the distortions issues with a Bayer sensor hit a point of diminishing returns as you add pixels ... still cameras long ago abandoned the idea of RGB sensors (well, other than Sigma/Foveon), and they just keep getting better.

Of course, part of this is also the advance of technology. When video cameras began, it was hard to make a full frame CCD with any kind of sensitivity. So for pro work, they could make 1/2 rez sensors with a pixel offset, and get a much better result. That was repeated in the early days of HDV. But now we're at the state where $400 consumer cameras can have 1/2.5" chips with enough pixels to do pixel binning and eliminate the interpolation errors you get with 1:1 single chip imagers. There are both single and three chip cameras with great video results. I still have 3-chip cameras, but I acknowledge it's no longer strictly black and white anymore... you have to analyze the details (as anyone who's drooled over some well done Canon EOS 5D Mk II video can attest).