Hacker News new | past | comments | ask | show | jobs | submit login

Or even digital imaging, which is a new technology forced onto an otherwise very conservative industry.

As an example, think of TSMC's upcoming 3nm process. It has a density of 300 million transistors per square millimetre. A Canon 5DS R has a 50 MPix resolution, which works out to about 60K photosites per square millimetre. That's a "budget" of 5,000 transistors per photosite!

That's more than enough to do "digital sensing". That is: accumulating photons in a digital counter instead of using an analog capacitor, allowing unlimited dynamic range. There would be no such thing as sensor saturation, allowing unlimited "electronic ND filter" effects without the colors being distorted even in a long time-lapse. It would allow "steering" of the data to nearby accumulators through an on-chip network at megahertz rates, providing near-perfect digital shake reduction without having to physically move the chip. Shake reduction could track moving objects and avoid blur even for very long hand-held exposures with cars or people in the scene, the same as what the Google Pixel does, but at a far higher quality. The motion vectors could be fed into a video compression stage, improving picture quality to above what the compressors can normally achieve by "guessing" the motion from 24fps still frames.

Etc, etc...




What does this mean...you still need some sort of analog capacitor to collect photons and convert them to electrons.


So instead of an ADC converting an analogue charge level to a 14-bit or 16-bit signal from a "big" capacitor, you instead have a tiny capacitor (or none at all) and collect individual 1-bit signals and accumulate them digitally. Photon counting, in essence.


so what kind of collecting device do you propose then?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: