Based on the reformulation of the problem of galactic shape measurement, introduced by Tessore and Bridle, we develop and test new estimators which more efficiently solve what is now a statistical problem. Frequently only a single measurement is available from an observation made by telescopes, which may be distorted due to noise introduced in the measuring process. To soften these effects, we investigate a number of approaches, such as scaling, data censoring, or mixtures of estimators. We then run simulations for various set-ups which emulate real-life conditions, to determine practical suitability of all proposed estimation approaches.