Oral Histories Breathe Life Into 3D Digital Reconstructions of Ohio Communities Displaced by Interstate Highway Construction

The Ghost Neighborhoods project uses AI and elder stories to digitally rebuild Columbus communities demolished by 1950s highway construction.

By: AXL Media

Published: Mar 23, 2026, 11:16 AM EDT

Source: Information for this report was sourced from Ohio State University

The Digital Resurrection of Lost Urban Landscapes

Scientists from The Ohio State University have reached a five-year milestone in their effort to digitally reconstruct three historically Black communities fragmented by the 1956 National Interstate and Defense Highways Act. By overlaying "ghosts" of demolished homes onto current 3D models of the roadways that replaced them, the project provides a stark visual representation of the human cost of 20th-century urban renewal. Senior author Harvey Miller, a professor of geography, noted that while the initial focus was on digitized infrastructure, the project is now evolving into a community-engaged effort to validate these models against the lived experiences of former residents.

Bridging the Gap Between Data and Memory

The research team is moving beyond purely scientific 3D modeling by integrating oral histories to fill gaps that technology alone cannot address. According to project leaders, talking to the people who once lived in these neighborhoods allows researchers to add crucial environmental details such as specific tree placements, vintage cars, and even the "life stories" tied to individual addresses. This collaborative approach ensures that the digital recreations serve as more than just architectural skeletons, instead reflecting the vibrant, thriving atmosphere of the communities as they existed before the construction of I-70 and I-71.

Streamlining the 3D Modeling Workflow

To accelerate the reconstruction process, graduate student Tshui Mum Ha is developing advanced clustering methods that categorize buildings based on architectural styles and semantic attributes. By analyzing construction materials, building heights, and interior lines found in historic Sanborn fire insurance maps, the team can generate plausible 3D models at a much larger scale than previously possible. These automated templates are designed to create a "first approximation" of a neighborhood, which can then be refined in real-time during engagement sessions where elders provide feedback directly on laptop-based models.