Proportions vs dimensions: shedding a different light on the analysis of 3D datasets

Résumé En

In the last decades, many methods (e.g., digital photogrammetry, laser scanning, computer vision, etc.) have been introduced that result in a renewed capacity of academics to produce large 3D datasets. Naturally research objectives, technological suites, levels of geometric accuracy required, or scales of objects under scrutiny do strongly vary - hence a wide range of “outputs” corresponding to various data interpretation strategies[1][2][3].But with that renewed capacity methodological questions emerge: does the “massive amount” of 3D data a survey results in really corresponds to the analytical need? Ultimately, is the added-value of “going massive” undeniable? We argue that this capacity to “go massive” can also open opportunities to investigate new analytical filters. We base on two observations:-more 3D data does not imply abandoning our capacity to reduce, to sum up, to spot significant features [4],-flexible, low cost survey suites can give us a chance to revisit fundamental metrics in the history of architecture : ratios, proportions, [5][6]rather than exhaustive dimensioning. We investigate how a low-res 3D point cloud can be re-read with the aim of identifying simple ratios and geometric relations, in other words of extracting from it meaningful architectural features. The paper underlines the cognitive potential of reading proportions in the history of architecture (both at design and analysis levels) and focuses on an experimentation conducted on thirty “comparable” edifices[7]. The approach exemplifies a shift from a one-shot, exhaustive documentation of one edifice to a workflow dedicated at decoding and visualising relations inside a collection.