Interferometric observations of the mm dust distribution in protoplanetary discs now show a ubiquity of annular gaps and rings, as well as a nontrivial occurrence rate of asymmetric substructures. The identification and accurate characterization of these features is critical to probing the physical processes responsible. This poster summarizes Frankenstein, an open source code that recovers the 1D brightness profile of a disc at super-resolution. The code uses a nonparametric, fast (<1 min) Gaussian process to directly fit the 1D visibility distribution, yielding an accurate match of the visibility amplitudes to longer baseline (typically by a factor of 2 - 5) than the Fourier equivalent of a CLEAN brightness profile. This yields a more accurate constraint on the widths and amplitudes of disc gaps and rings, and often an identification of substructures not seen in the source’s CLEAN image.