Monday, June 21, 2010

Canesta Manufactures Consumer 3D Sensors at TowerJazz

Business Wire: Canesta is using TowerJazz’s CIS process to manufacture its CanestaVision 3-D image sensors. CanestaVision chips are the single-chip 3-D image sensor SoCs for use in consumer desktop computing, TV and entertainment applications.

Canesta’s solution is the most robust in the industry, offering the best performance compared to any other competing solution. The very tight R&D collaboration and roadmap alignment between our two companies provides near and long term innovative solutions that will further improve performance and power consumption in future generations and will allow wins in many more applications, such as the mobile computing area,” said Dr. Avi Strum, VP and GM of Specialty Business Unit, TowerJazz.

Globes sources estimate that the deal will add $200-300M in annual revenue to TowerJazz.

While we are at ToF news, Softkinetic, Optrima and Metrological Media Innovations announced a partnership to offer set-top-box and television manufacturers an integrated 3D gesture-based solution. Mediaconnect TV, a hybrid consumer TV entertainment platform based on Intel CE Media Processing technology and powered by Metrological’s Metroconnect OS adds 3D gesture-based interfaces to its rich set of functions. The commercial deployment of Mediaconnect TV begins in Europe in the coming weeks and is expected to begin in North America by fall of this year.

8 comments:

  1. Anyone can tell us to who Canesta can sell 200-300M$ ToF 3D sensors ???

    Thanks !

    ReplyDelete
  2. At this time TOF sensors are relatively large chips and due to low volume, ASPs are high. It would be a mistake to imagine that will continue for long if TOF sensor adoption rate is high. While pixel sizes need to be large to capture enough reflective signal and keep NIR light sources at reasonable power levels, once bigger players enter this market, ASPs will drop like a rock. I remember in the old Photobit business plan we had VGA sensors at about $10 per chip. It was reasonable considering where CCDs were priced. People laughed when I predicted a one megapixel sensor would be $10 using CMOS.
    see: http://www.ericfossum.com/Publications/Papers/Ultra%20low%20power%20imaging%20systems%20using%20CMOS%20image%20sensor%20technology.pdf

    Now $10 for a 1Mpix chip is laughable in the other direction.

    Personally, I think the standard TOF sensor will have a short lifetime and these companies will be reduced to niche markets and collecting royalties where applicable. Monolithic Color 3D sensors is where the action is at (aka RGBZ).

    I was reminded the other day about a guy at National Semi (sorry, can't recall his name and it was not Dick Merill or Kevin Brehmer) but his mantra when we were doing CMOS APS tech transfer from JPL to National was "2-bucks, 2-bucks, 2-bucks, all chips wind up costing 2-bucks"

    ReplyDelete
  3. From technical point of view, the ToF method has the advantage to be able to keep the same precision over the distance if course the emitter is powerful enough. All the working ToF point sensors profite from highly collimated ultra powerful laser. The 2D ToF sensor is an attractive idea, but the problem is that it needs an even higher laser source becaue the optical energy is diluted in space.

    One of critical problems is that ambient light can not contribute to the measuring. From energtical point of view, ambient light/Sun light is a very powerful photon noise source. This makes actual ToF 3D sensor's effective frame rate low because temporal integration is needed. When the distance is large, the laser power will be prohibitive especially in consumer field where the eye safty is a key issue.

    For TV set, getsure based gaming, other solutions are possible such as motion analysis, passive stereo vision, structured light stereo vision, etc. All these approachs can work with normal or almost image sensors with reasonable cost and complexity.

    It seems for me that the computation power can be obtained straightforward with more and more advanced CMOS process.

    The absolute price is not a real issue, the real issue is: do we need this stuff to control my TV set ? can we use this device to play all the games ?

    All the major ToF sensor companies are NOT new companies, they are all OLD companies.

    INTEL CPU chip size is not necessarily larger than a CMOS sensor, but the price is much much higher !

    ReplyDelete
  4. But keep in mind that today´s time of flight cameras (I am speaking about canesta´s, pmdtec´s and mesa´s) are continuously increasing sensitivity. A few years ago, they required several watts of optical power. Now they demonstrate far better performance with few hundred milliwatts or less, with more and more acceptable resolution.
    We have worked with stereo approaches for many years. It is not only the computaional power stereo needs, for almost all applications we had to recognize serious problems with low contrast objects and callibration issues over time and temperature.
    Now we are working also with ToF systems, where many of the stereo-problems are solved.
    Dont´t you think there will be a sort of Moore´s law in optical power for ToF sensors?

    ReplyDelete
  5. Of course I didn't say that there is no progress in ToF solution. It's a very attractive solution indeed. But in ToF, we are facing more physical limitation which is not always related to scaledown process.

    As for digital computation, the electron number involved in each bit switching is still much much larger than the quantum limitation, so the scaledown process is still straightforward (1fF*1V = 6250 electrons).

    But for ToF solution, the number of photons on the photodiode is not that large, especially for short distance, the laser pulse width should be kept very small. If I remember well that MESA uses multiple measurements in order to determine the 4 parameters in a sin wave, while PMD uses an optical mixing approach to get the phase of the optical wave. In both case, the measureing reliability will depend on the final equivalent SNR.

    Theoretically speaking, you can use a high power laser with very short duration, but this kind of laser illuminator is not that easy anyway.

    ReplyDelete
  6. Dear bloggers, anyone know whose CIS used by Redone cinema camera ?

    ReplyDelete
  7. Hello Sir,
    I want 3D camera for video recording. I want it for 3D gesture recognition (with deph information). Please tell me its cost and how will I get it?
    Email ID: sonal.kumari1910@gmail.com


    --
    With thanks,

    Sonal Kumari

    ReplyDelete
  8. Can its 3D sensor be applied to shooting 3D photos or videos? I'm wondering whether companies with products like 3D cameras or 3D camcorder would be interesting in it...

    ReplyDelete

All comments are moderated to avoid spam and personal attacks.