New Imaging Technologies announced release of its NSC1005 HD Ready sensor operating in 720p mode.
The NSC1005 has 1280x720 pixels of 5.6um size, with microlenses and/or color bayer filters. NSC1005 operates in rolling shutter mode up to a frame rate of 60fps. NIT's Native WDR technology is said to provide an instantaneous DR of more than 140dB, a logarithmic response of the signal. A custom de-bayerisation algorithm is available upon request from NIT in order to operate in the logarithmic response domain.
Are there any "hidden" drowbacks of the solar cell mode pixels compared to standard logarithmic pixels?
ReplyDeleteWhat kind of "hidden" drawbacks do you think about, please ?
ReplyDelete-yang ni
Simply I haven't read any scientific pubblications comparing the two methods. I just want to know if the solar cell mode is the future of logarithmic pixels since (according to NIT) the performance seem to be quite interesting...
ReplyDeleteWell, actually there are some shortcomings with this approach, for example: no share pixel possible, no way to compensate KTC noise, etc. But there are also a lot of advantages such as ultra wide dynamic range, contrast indexed imaging, good color rendition, etc...
ReplyDeleteWhy does this logarithmic sensor have the advantage of good color rendition?
ReplyDeleteWhy is white balancing not required for this sensor, according to the press release?
KTC noise ? From my understanding the log doesn't generate KTC noise in the pixel.
ReplyDeleteKTC noise comes from reset and you doesn't reset log pixel.
some precisions:
ReplyDelete1. color issue
At first our log sensor has a very precise and predictable response over all the dynamic range. This will give a perfect uniformity among the R/G/B channels. Logarithmic response transforms a multiplication into addition so the white balance operation is naturally merged with the display adjustment.
2. KTC noise
Indeed, in log operation mode, there is no KTC. But photodiode in solar cell mode generates Johnson noise and when you integrate this noise over the noise bandwidth which is determined by the dynamic resistance and the parasite capacitance, you will get a RMS noise which is exactly the same as KTC.
You can refer to my presentation at ISE2011 for more details.
-yang ni