Geometry, acoustics and reconstruction hypotheses

Choose the representation before the tool#

Modelgenerator: preserve what the cameras support#

The current Processor Modelgenerator takes photographic material through a pinned DA3-based workflow to point clouds and camera paths. Its documented output should not be advertised as a finished textured mesh. Camera registration, visible surface coverage and successful transfer of output assets are distinct checks. A coherent cloud can still omit the interior of an object or include poorly supported surfaces.

Record the input photographs, processing configuration and coordinate relationship to measurements. Photograph-based geometry and a manually supplied dimension can be combined, but the resulting scale or correction should have a derivation record. A hidden pipe mouth, internal cavity or inaccessible mechanism cannot become an observed surface merely because software produces a complete-looking shape.

VAO-Pipe: make the geometric hypothesis inspectable#

VAO-Pipe belongs to a different route: procedural pipe geometry and passive pressure-acoustic evaluation. Its parameters make a hypothesis explicit and allow controlled variation. That is useful for investigating how selected dimensions and boundary assumptions influence a modeled result. It should remain distinct from a complete account of self-sustained sound production, historical voicing or the actual condition of a surviving pipe.

The thesis emphasizes the danger of tuning underdetermined parameters until a desired result becomes inevitable. Preserve the source of each dimension, distinguish measured and inferred values, and identify which observations were used to adjust the model. Compare against independent evidence where available. A close frequency agreement after optimization describes that model and fitting procedure; it does not automatically recover a unique historical geometry.

Simplification can remove the feature that matters#

A low polygon count is attractive for interactive playback, but visual economy and acoustic relevance do not always align. Section 8.5.1 reports that unweighted simplification below 50,000 polygons damaged features including labia, core gaps and cut-up edges in the investigated setting. A protected weighting of 100 belongs to that configuration; it is not a universal setting for every organ or reconstruction pipeline.

Retain a detailed research realization and derive a lighter presentation realization with a documented purpose. Check the features relevant to the question after simplification, not only the final file size. The two realizations can share an asset role while preserving their different bytes, derivations and intended uses. This makes a mobile visualization usable without silently replacing the geometric evidence.

PAMT: a proposed architecture for spatial acoustics#

PAMT is presented as a conceptual architecture, not a field-validated room-acoustic product. The thesis discusses coupling geometric and wave-based methods because geometric propagation alone cannot represent all low-frequency modal behavior. The transition needs to depend on the room and frequency regime, with reference to the room’s Schroeder frequency, rather than a universal cutoff applied everywhere.

No in-situ sound-pressure measurement campaign is reported for PAMT in the dissertation. Its proposed field validation and uncertainty budgets remain prospective. This matters when selecting a tool: the architecture can guide research questions and implementation, but it does not supply measured accuracy for a visitor’s room. A published conceptual method, a working prototype and a calibrated acoustic model are different deliverables.