As photogrammetry processing evolves far faster than camera hardware, the real value of a project is no longer tied to a single output, but to the preservation of the original capture data.
Imagery collected today can be reprocessed tomorrow to extract new measurements, improved accuracy, and insights that were not possible at the time of capture. Retaining original datasets allows future tools to reassess assumptions and unlock additional value as methods continue to advance.
The attached video illustrates this using imagery captured earlier in 2025 and expressed through four different representations:
• Photogrammetry mesh: strong visual realism with analytical potential when properly constrained
• Photogrammetry point cloud: reduced visual polish but the most direct geometric record
• Gaussian splatting: exceptional realism for visualisation and context, though not inherently designed for metric analysis
• Transparent point cloud: a visualization approach that enhances spatial understanding without altering geometry
These outputs are not interchangeable. Some prioritise presentation, others prioritise measurement, validation, and analytical confidence.
I endorse Barry Bassnett’s emphasis on long-term data stewardship: "the idea that datasets captured today may be reprocessed years or decades from now using methods that do not yet exist". Preserving original photogrammetry data ensures future practitioners can remeasure, reinterpret, and extract new value as tools evolve.
