Sunday, September 16, 2018

GPixel 11um-Pixel Sensor in Moonless Night

Russian KB VITA kindly sent me a video of its camera that uses GSENSE400BSI sensor. The video was shot at moonless night, hence no shadows there. The camera has f0.95 lens and set at 22fps and 40ms exposure time:



24 comments:

  1. how many lux light?

    ReplyDelete
    Replies
    1. 2mLux, according to Wikipedia:

      https://en.wikipedia.org/wiki/Lux

      Delete
  2. Seems like exposure is not fixed. What's with the defects?, you'd think the whole row or column would be bad, why is the row partially bad?
    SNR = 49dB anyone else thinks this is low? Also, PRNU = 1.5% seems high...

    ReplyDelete
  3. The exposition is not constant, because the brightness of the ground and sky is very different, the exposure changes when the camera is moved to the sky. SNR does not affect sensitivity. The sensitivity is determined by the "noise read" that this sensor nread 2e. PRNU does not affect in low light scenes and it can be adjusted for bright scenes.

    ReplyDelete
  4. YouTube video quality 720p and compression is terrible. Is original available?

    ReplyDelete
    Replies
    1. See the video file below. This is what I've been sent:

      https://cloud.mail.ru/public/7nfY/if4pWk6Yd

      Delete
  5. If you remove the IRCut filter of a Sony A7s, you should be able to get much better results.

    ReplyDelete
    Replies
    1. Really? Why does the IR cut filter has a role to play here?

      Delete
    2. QE=95%@560nm, GSENSE400BSI

      Delete
  6. from link:
    https://habr.com/post/374163/
    video
    https://cloud.mail.ru/public/B1BK/fokxGJMPz
    Is it the same camera?

    ReplyDelete
    Replies
    1. Yes, it's the same camera, but only earlier winter tests with an F/1.2 lens. SWIR camera is also our development.

      Delete
    2. what cameras on video? are these three different cameras?

      Delete
  7. with sub electron RON and virtual zero DC, the 95% peak QE can not stand for comparison.

    ReplyDelete
    Replies
    1. Where do you see sub electron read out noise? It says it's 2e- and the noise is very visible.
      Does the analog gain change in this video too? Seems like in certain shots the noise changes dramatically...

      Delete
    2. "virtual zero DC" ??? This sensor has a tremendously high DC!
      Have a look at the DC of a state-of-the-art (2018) CMOS sensor, e. g. that one from Panasonic:
      https://astronomy-imaging-camera.com/wp-content/uploads/1600mc-cool-dark-current-vs-T1.jpg
      -dkf

      Delete
    3. What is the pixel size of this Panasonic sensor? Is it 11um?

      Delete
    4. Pixel size is 3.8┬Ám. But you should put the DC in relation to the sensor area. Then the DC for the GPixel is ca. 1650 e-/s/mm^2 and for the Panasonic it's ca. 26e-/s/mm^2 (extrapolated to -50°C)
      -dkf

      Delete
    5. i think about: CMOS-Chip (Panasonic MN34230) 3.8*3.8u
      https://www.astroshop.de/astro-kameras/zwo-kamera-asi-1600-mc-pro-color/p,56571#tab_bar_1_select

      Delete
  8. I think that this comment is related to Sony A7S sensor ...

    ReplyDelete
    Replies
    1. Well, if "virtual zero DC" is a nice circumscription for Sony's "star eater" problem, then the comment makes sense ;-)
      -dkf

      Delete
  9. https://www.youtube.com/watch?v=J9rgtz0vE_c

    ReplyDelete
    Replies
    1. Milky Way, fish-eye, gsense400 (kbvita)
      https://www.youtube.com/watch?v=MzdL9S_yh0o&list=UUg1-Iwu349iPhzBON9NVzhA&index=7

      Delete
    2. I do not see any satellite
      very long exposure

      Delete

All comments are moderated to avoid spam and personal attacks.