Digital image sensors are actually … analogue. They are -sort of- buckets that gets filled with (many) photons. The level of filling during an exposure is converted in a specific voltage that is used by the camera software to associate a certain level of brightness (it is a bit more complex but you get the idea).
It is only at this stage, when the image pixels are recorded that they are transformed into numerical -digital- values.
A team of researchers at Dartmouth has created a new technology, QIS: Quanta Image Sensor, based on nanoparticles that can change their structure when “hit” by a photon. Each nanoparticle has a 0 or 1 structure and this structure can be reset (forced to 0) over 1000 times per second. If a photon hits the nanoparticle it pushes it into the 1 structure and this change can be read by the sensor.
It becomes possible to read a single occurrence of a photon (for curiosity, our retina can also spot a single photon but the probability of a photon hitting a molecule of rhodopsin is such that on the average you need 50 photons to detect one). This clearly results in the ability to capture images in very very low light.
Even more interesting, this sensor acts like a movie camera: by reading the sensor 1,000 times every second one can track the movement of a light beam (of an object) so that an image becomes a video clip.
The developed prototype contains 1 million “sensing” nanoparticles, where each one can be roughly compared to a pixel. A million pixel camera would be too little today. The researchers feel they can scale up this technology up to 1 billion nanoparticles per sensor, potentially 1G pixel!
Notice that the prototype does not have any color filter so it is not able to take “colour pictures”, only black and white. It should be possible, however, to upgrade the sensor to detect colours (although it seems tricky to me using the normal filtering architecture, it wouldn’t work with nanoparticles), at least this is what the team of researchers is claiming.
Can we expect this technology to disrupt the consumer cameras world? Unlikely. According to the researchers present technology is so advanced in several directions (performance, economics, ease of manufacturing, software) that a disruptive one will have very little chance to take the market. However, it may hit a few applications niches in scientific application areas. If successful in some of them it might hope to extend to more and eventually invade the consumer market.