Monday, January 29, 2018

Espros Presents 3D Face Scan with 130um Depth Resolution

Espros publishes a video of 3D face scan of its CEO Beat De Coi. The epc660 sensor scan features:
  • Simultaneous acquisition of 3D TOF and grayscale images
  • Distance resolution 0.13mm
  • 5 images per second with a 1 GHz ARM8 processor


  1. this 0.13mm z resolution means you need to measure the time between outgoing and incoming pulse within an time accuracy that it takes for light to travel 0.26mm? 300000km/s = 300m/us = 300mm/ns = 300um/ps. Its absolutely fascinating that such a magnitude is possible by using light speed as gauge

  2. What is depth resolution btw? We want to see absolute accuracy and noise. I guess per-pixel noise is at least 3mm without spatial and temporal smoothing.

  3. According to their datasheet the maximum modulation frequency of the epc660 is 20MHz resulting an unambiguity range of 7.5m.
    7.5m / 130µm = 57692, so they have a 16 bit ADC - except the datasheet says it's 12 bit and the resolution table there (page 67) says the resolution at 20MHz is 2.5mm.

    So I'm wondering where this 130µm comes from.

    1. oscillator/modulation frequency is irrelevant - there are oscillator multipliers in many state of the art semiconductors including but not limited to SoC processor systems.
      modulated signals are quite good for phase measurement. if you divide a 2PI range into e.g. 1024 phase steps - then you will get 7.5m / 1024 = 7,3mm. the precision number from above would have a factor of 5769. and the given precision is still not breaking the values that others have created using polarizer based surface reconstruction - some years ago. its a high value that needs more detail to understand its origin and the true nature of its meaning. but its basically in reach and seems to be in the area where bleeding edge results are suspected to lie in...

    2. In a system that works by sending modulated light the "modulation frequency" refers to that frequency. To quote the datasheet: "Selectable modulation frequencies 0.625 ... 20MHz resulting in unambiguity distance of 7.5m … 240m"

      130µm = 0.13mm
      7.5m = 7500mm
      7500mm / 0.13mm = 57692
      ... so I still think this resolution doesn't match their ADC.

    3. a high grade ADC (speed and quality) is often the key to make those technology fly. lets see what details will be further unveiled in this specific model.

      (sorry, it seems i was really off by a multiplier of 10 in the previous quick writings on the factor)

  4. I don't care what their engineers say, the video 3D imagery shown here says NOWHERE near that resolution. This claim of 130um is laughable !!!


All comments are moderated to avoid spam and personal attacks.