Monday, September 07, 2020

Vivo Claims to Improve Sensitivity by 160% with RGBW Sensor with Sparse RB Colors

SparrowNewsGizmoChina, PCPop, Sina Technology, PRNewswire: Vivo announces that a new RGBW sensor will be used in its smartphones next year. The sparse blue and red colors coding is said to increase the sensor sensitivity by 160% over the regular RGB. The sensor is rumored to be manufactured by Samsung, named GN2 or GX1.

15 comments:

  1. Using W pixel certainly can help sensitivity overall. How about the chromatic aberration caused by W pixel? I don't think the quad pixel configuration is going to help for this.

    ReplyDelete
    Replies
    1. Chromatic abberration is not a significant issue for small lenses used in smartphones. CA is a problem if
      - the lens is large
      - there is a large gap between the lens and the sensor such as DSLRs have to acommodate their mirror/prism.

      While it is true that CA can be corrected in RGB CFAs such as Bayer, but cannot be corrected in the W color plane, W color plane is less affected by CA to begin with. This is because the quantum efficiency of R, B us lower than that of G, so the misaligned R, B have low impact on the reference G. In Bayer, on the other hand, R, B are amplified by the white balance step making their misalignment with G more objectionable.

      Furthermore, the W color plane is not output - it is used internally to clean up R, G, B. Also the misalignment of R, B with G, W in a RGBW sensor can be corrected in the ISP just as it is done in Bayer.

      In summary, RGBW's CA might be problematic for DSLRs and Medium format cameras with cheap lenses, but should be fine for smartphones, mirrorless cameras and cine cameras.

      Delete
  2. SONY/SAMSUNG/OMNIVISION need a RGBW sensor too.

    ReplyDelete
    Replies
    1. SONY 8 years ago - been there, did not work.
      http://image-sensors-world.blogspot.com/2012/09/sony-abandons-rgbw-coding-in-stacked.htmlSONY 8 years ago - been there, did not work.

      Delete
  3. More light at the expense of color accuracy

    ReplyDelete
    Replies
    1. The color accuracy of RWB and RYYB is low because they are missing green. The color accuracy of RGBW is high, being comparable to Bayer.

      Poor implementations of RGBW have suffered from low resolution and false color. These are not fundamental problems, however, and can be overcome with sophisticated demosaicking algorithms.

      Simply put, the Nyquist theorem is a sufficient but not a necessary condition. Natural images are compressible and that allows the recovery of high resolution luminance and chrominance signals from RGBW CFAs.

      Delete
  4. How does it compare to human vision? Is it more sensitive?

    ReplyDelete
  5. Reminds me of VHS color under.

    ReplyDelete
  6. Why vivo use Kodak 2005 patent as their own innovation? At least they should mention Kodak's name in this ppt.

    ReplyDelete
    Replies
    1. No, vivo specifically said this technology is not from them and available to all other smartphone companies. Many Chinese sources report vivo and another "TOP5" smartphone company are the first two to use.

      Delete
    2. Actually it was not Kodak who came first with panchromatic pixels in combination with a CFA. In the late '80s Sanyo already provided a CCD with panchromatic pixels combined with colour pixels (complementary colours at that time).

      Delete
    3. Sony used Canon's CFA, from this 1996 patent, in their IMX 278 sensor featured in the Huawei P8.

      Delete
    4. @AT Bayer US Patent No. 3,971,065 issued 1976 includes luminance and chrominance filter array elements. NOT SANYO! (unless you have some earlier reference than 1980 for Sanyo)

      Delete
    5. @EF : you are right, Bayer is already talking about luminance and chrominance. But he mentions in claim 2 that the luminance filter is a green filter. In the text he refers to luminance as the signal with a color vector which is the major contributor to the luminance signal. Does this mean that a panchromatic pixel is also considered ? Interesting discussion for patent lawyers.

      Delete
    6. @AT Well, I was not trying to argue patentability etc. but just to give credit where credit is due. But Bayer claim 1 is the independent claim and claim 2 is dependent where green can be considered as decent representation of luminance. But, you could have said luminance is NOT panchromatic since it is not "white" nor "clear." On the other hand, referencing Banning US Patent No. 2,755,334 1956, Fig 35 (not CFA but showing RGBW elements) it could be considered obvious to combine Bayer with Banning, esp. as Bayer references Banning and RGBW in his 1976 patent 20 years later. We need to give credit to Banning and Bayer today as we continue to use many of the ideas they came up with for electronic color imaging. And, let's not credit Kodak or Sanyo, let's also credit the individual men (and women).

      Delete

All comments are moderated to avoid spam and personal attacks.