Site-Archive Assembly (WebVR Prototype)
2024https://eytanmann.xyz/wadisalib/
The project, developed as part of my PhD dissertation, is a data-driven web-based virtual tour generated from the analysis of archival materials and digital scanning of architecture. It consists of three steps:
-
Automatically identifying and associating between data nodes - both archival and physical materials.
-
A Network Graph mixing the archive and scanned buildings - photographs, scientific writing, and memories, with point cloud data.
- Transforming archival materials into a continuous virtual tour experience using a web pipeline with GIS and public mapping resources.
The project invented a protocol to automatically identify and associate between data nodes - both archival and physical materials. This meant laser-scanning elements from within the site, in conjunction with archival research, and exploring data analysis tools to fragment data.
Machine-Learning models helped segmenting and classifying manuscripts (Using GPT-4) as well as Semantic Segmentation and Classification of point cloud data (PointNet model).
Machine learning models to spatialize archive materials were also deployed to extract spatial features from archival materials. These include photographs and film segments.
The scene above is extracted from Amos Gitai’s 1975 film ‘The Wadi Salib Events’ in which a man names Jehudah is seen standing on the plot where once his house stood in Wadi Salib, before its demolition during the 1960’s.
A Network Graph database allows to mix various materials, combining scanned buildings with photographs, scientific writing, and memories to provide multiple contexts.
Video capture of prototype:
The Network Graph databased was used as a navigation system for a virtual tour of the site-archive of Wadi Salib. This Video excerpt from web platform shows movement through this space. The system was developed using CesiumJS - javascript library for integrating 3D and Geographical system inside the web browser.