Applied research · 3D digital scale models
Applied research in digital scale models and audiovisual process automation: a method to reproduce real-world places in 3D, with a case study in the Ribeira neighborhood of Natal, Brazil.
This applied research aimed to investigate and develop an optimized process for creating three-dimensional digital scale models for audiovisual production, integrating 3D modeling techniques, process automation and methodological organization of production. The research was carried out through hands-on experimentation, iterative testing and a comparative analysis between traditional methods and the proposed method. The results point to a significant reduction in production time, greater standardization of digital assets and higher workflow efficiency, showing the potential of the solution for educational, cultural and audiovisual applications.
The advance of digital technologies has significantly expanded the possibilities of visual creation and representation in the audiovisual field. However, traditional processes for developing sets, environments and three-dimensional structures still have limitations in production time, high cost and low scalability.
In this context, digital scale models emerge as a viable alternative for simulating spaces, narratives and complex environments. Yet the lack of standardized methodologies and automated workflows can compromise the efficiency of the creative process.
This research arises from the need to investigate and structure a more efficient method for producing digital scale models, integrating 3D modeling practices, asset organization and automation, with a focus on practical application in audiovisual projects.
Location chosen for the research: the Ribeira neighborhood, in Natal/RN.
To develop and validate an optimized method for creating digital scale models applied to audiovisual production, aiming for greater efficiency, standardization and shorter development time.
To develop an automated procedure for creating 3D digital scale models used in animated shorts, mid-length and feature films, and video games.
Contemporary audiovisual production demands agile, scalable and economically viable solutions. Adopting digital scale models allows greater creative control, lower costs compared to physical set design, and flexibility in adapting narratives.
From a scientific and technological standpoint, this research is justified by contributing to the development of an applied methodology that joins theory and practice, and by generating knowledge that can be replicated in educational, cultural and institutional contexts.
The research was divided into stages, each carefully planned and carried out to ensure the precision and quality of the results.
A DJI Spark drone with a 4K camera captured aerial images of the neighborhood’s streets, sidewalks and architectural details.
A GoPro 9, also 4K, filmed and photographed signs, facades and other important elements.
The captured images were integrated into the inside of a 3D virtual sphere (Polycam), creating a rich, detailed environment on the computer.
Measurements of physical elements, such as street width and building height, were converted into pixels and adapted to the software’s working scale, via an FBX file.
Artificial intelligence algorithms processed the images and scale information, letting the tool draw the 3D model automatically with high fidelity to the real world.
In the Unreal Development Kit (UDK) engine, the model of the study area was drawn by hand, vertex by vertex, to assess the time and effort involved.
Every step was documented in a logbook, including difficulties, solutions and improvements.
Results and methods were published on a website dedicated to the research, with sign language (Libras) interpreter videos for accessibility.
The research resulted in an automated technique that cuts the production time of digital scale models to 50% of the estimated time (for production with vertex drawing), without compromising quality.
More than 500 aerial and 300 ground images ensured full coverage of the Ribeira neighborhood. The images were processed to remove distortion and improve visual quality.
The images were integrated into a virtual sphere using Polycam, UDK and 3DS Max, allowing an immersive view of the mapped environment.
Physical measurements were converted precisely, and the AI was trained to recognize and draw architectural elements from the images and scale data. The result was a 3D model that faithfully replicated the neighborhood, including facade textures and traffic signs.
Measurements were drawn by hand, without full precision, based on the available images and scale data. The resulting model represents the neighborhood with details such as facade textures and traffic signs, although it is not a fully faithful replica.
The logbook recorded every stage, with methods, challenges and solutions. Results were published on the research website, with explanatory videos and sign language interpreters.
Images of the Ribeira neighborhood model built in the engine, in wireframe views and with textures and lighting.
The automated technique developed in this research is a significant step forward in producing 3D digital scale models, offering an efficient and precise tool for design professionals. The methodology not only speeds up the process but also democratizes access to advanced 3D modeling technologies, benefiting small producers and independent professionals.
The tool used for photogrammetry and for building the automated model was the Polycam software. Files were exported as .FBX, and can also be exported in the formats below, which are compatible with many content-creation engines for games and animated films, allowing them to be used in virtual environments and multimedia applications.
The results and the full step-by-step of the research are published with images and attachments, aiming for full accessibility.
This project is open source. Any professional who feels able to update, improve or otherwise contribute to the development of the research is welcome to send their contribution.
The research remains open to future contributions, encouraging continued collaboration from the community.

This project was supported with funds from the Aldir Blanc Law 2, promoted by the José Augusto Foundation, Natal City Hall, the Ministry of Culture and the Federal Government of Brazil.
We thank the dedication and effort of the team involved and everyone who contributed to the success of this project, especially the José Augusto Foundation, Natal City Hall, the Ministry of Culture and the Federal Government.
Geographic data were collected from the Municipal Secretariat for the Environment and Urban Planning (SEMURB), through the Department of Information, Research and Statistics (DIPE), Research and Statistics Sector (SPE) of Natal City Hall.