As part of a collaborative project with the Weser Renaissance Museum at Brake Castle in Lemgo, AI-generated sculptures related to the (Weser) Renaissance were created and exhibited as 3D prints.
Project participants
Leonie Hans
Sascha Etezazi
Alexander Mann
Jakob Wagner
Stella-Lynn Krysztofiak
and many more
Project status
Completed.
Exhibition: From September 16th, 2025, to February 1st, 2026, at Schloss Brake in Lemgo
The Weser Renaissance Museum Schloss Brake and the Eulenburg in Rinteln aimed to create the first museum exhibition of AI-generated sculptures. The most important component, the AI-generated sculptures, was designed in collaboration with a master's course in media production led by Prof. Anke Stache.
Once the museums and students had finalized the concepts for the sculptures on the themes of the individual and perspective, the technical implementation of the concepts began at KIO. Among other things, we trained LoRA models and generated images and 3D models. KIO also documented the exhibition as a Gaussian Splat to preserve the work beyond the exhibition period.
Small AI model extensions were trained using images from the museum archive. These models can imitate various styles of the archive images. This allows an image AI model to generate images that resemble the originals.
In the next step, different models were created for the various concepts:
The exhibition is sponsored by the Ministry of Culture and Science of North Rhine-Westphalia and the Regional Culture Program of North Rhine-Westphalia. The exhibition is being held in cooperation with the following partners: Fraunhofer IOSB-INA, TH OWL, KI Akademie OWL, init TH OWL, Federal Ministry of Research, Technology and Space, KIO, Trinnovation OWL, Detmold University of Music, Department of Engineering and Mathematics at Bielefeld University HSBI, LWL Museum Ziegelei Lage, and is supported by Frauen für Lemgo (Women for Lemgo).
To maintain the integrity of the data used, both the LoRA training process and the generation of images were carried out on in-house graphics cards and the GPU cluster.