Morphology of soot nanoparticles is characteristically complex and 3-dimensional, and plays a defining role in soot-related phenomena. Morphological characterisation of soot is essential to understand the extent of such effects, including harm to human health, and develop strategies to mitigate them. Use of 2D-TEM for characterisation is associated with numerous and significant sources of error and uncertainty related to a 2D-3D information gap. Volume reconstruction by 3D-TEM avoids many of these sources of error, and has been shown in simulation studies to be highly accurate. However, the technique has traditionally been too slow to permit study of enough individual structures to satisfactorily characterise a bulk soot-sample. Similarly, the prevalence of manual image processing in 2D-TEM studies of soot can limit characterisations to as few as 50 individual particles per sample. In this study a methodology has been developed to permit high-throughput tilt-series acquisition of soot nanoparticles for use in 3D-TEM volume reconstructions. This technique involves semi-automated 2D-TEM image processing via automated segmentation to screen large areas of TEM grids for suitable structures. Automated selections can be reviewed and corrected to allow rapid and high-volume soot sample characterisation in the traditional 2-dimensional approach. Application of this methodology to a GTDI soot-in-oil sample has enabled an intermediate characterisation of 30+ particles in 3D, and an advanced characterisation of 500+ particles in 2D. Results from both methods are compared to highlight the effect of different samples sizes and projection orientation in 2D-TEM, and the differences in results of 2D-TEM in comparison to more accurate 3D-derived results. On average, 2D-derived results deviated from 3D values by 30% in volume, 50% in surface area, and 22% in circularity, though 2D and 3D radius of gyration was fairly similar (<10%).