[okooptics] has managed to obtain 3D images without using a lens. A thin layer of Scotch tape placed over the sensor instead of a lens on a Raspberry Pi camera disperses the image, while suitable deconvolution mathematics reconstructs an image from this blurred data.
Giving directional weighting to the point spread function used in the method provides views from different angles, as if the camera position had been shifted slightly. While the technique requires extensive mathematics, its limits of applicability and scientific foundations are also explained.
Why it matters
This method shows that in three-dimensional imaging, the computational approach can be as decisive as the optical components. The ability to produce an image as if it were being perceived from different angles makes it possible to evaluate the resulting data not merely as a flat image, but together with depth information. However, the result depends on correctly establishing the mathematical model used and the directional weighting applied to the point spread function; therefore, the process is not simply a matter of applying tape. The conditions under which the technique produces reliable results and the extent to which its scientific foundations can be generalized remain open questions that define its limits of applicability.