This “camera” takes very special photos in which it does not capture the spatial information of the photons, but their temporal information

Digital cameras have sensors with millions of pixels that allow images to be formed by detecting the intensity and color of light at each point in space. Decades of research seem to make it clear that this is the optimal way to take photos, but a singular alternative has appeared.

A group of researchers from the United Kingdom, Italy and the Netherlands have created an imaging system that captures temporal information about photons instead of their spatial coordinates, allowing you to create animated 3D images. The idea is to apply this system to cars, mobile devices and health monitoring solutions.

Light is no longer (so) necessary

The process is started by using a detector that acts as a kind of photon timer. Instead of measuring the spatial distribution of photons’ color and intensity, as digital cameras do, the detector it only records the time it takes for photons produced by a laser to bounce in the objects of a certain scene and then reach that sensor again.

This “camera” takes very special photos in which it does not capture the spatial information of the photons, but their temporal information

Information about those times is reflected in the scene and recorded in a simple graph which is then transformed into a 3D image with an algorithm based on neural networks. That algorithm was trained with thousands of conventional photographs in which the team was shown in the laboratory moving and carrying objects, but in those photos that temporary information was also captured.

The neural network and the algorithm ended up learning how temporal data corresponded to spatial data, which allows to generate a kind of 3D animated photos -which at the moment do not have much to do with traditional photography, yes- from that temporary information.

Dr Alex Turpin, from the University of Glasgow, explained that with this system he is “capable of converting data in one dimension – a simple measure of time – into a moving image that represents the three dimensions of space in any given scene. The most important difference from conventional imaging is that our system is capable of decoupling the light completely from the process“.

For now, the neural system only recognizes scenes similar to those with which it has been trained, but with more training and even more advanced algorithms it could become, according to those responsible, a new alternative to visualize all kinds of scenarios and thus be applicable to various fields.