AI brings ancient Rome to life with highly plausible, historically accurate images

Two University of Zurich researchers have created Re-Experiencing History, an AI image generator that produces historically informed visualizations of ancient Rome and Greece based on scholarly sources. The platform represents a novel approach to historical education, using curated academic materials to train AI models that generate plausible visual representations of historical scenes rather than generic “ancient-looking” imagery.

How it works: Professor Felix K. Maier, an ancient history professor, and computational linguist Phillip Ströbel trained existing AI image generators using nearly 300 carefully curated images and captions from scholarly sources.

• The system draws from annotated materials including illustrations from academic books on Roman clothing, weapons, architecture, and triumphal processions.
• Each user prompt gets enriched with historically specific information retrieved from a database of around 70 research articles and books on Roman culture.
• Instead of broad or vague instructions, the refined prompts spell out concrete details about clothing, ritual actions, or settings, which makes the generated images far more specific and historically plausible.

What you can see: Sample outputs include vivid scenes like Roman soldiers playing games in military camps, the sack of Rome by Visigoths in 410 AD, and young women walking to religious festivals.

• A demonstration video created with OpenAI’s Dall-E 3 and fine-tuned Flux Dev shows sunlit outdoor markets, armor-clad soldiers, street scenes of poverty-stricken citizens, and rulers greeting crowds from chariots.
• For well-documented events like triumphal processions, the results can be impressively close to archaeological evidence, featuring recognizable elements like processional wagons, laurel crowns, and Forum architecture.

The limitations: The AI system struggles with less-documented historical scenarios and exhibits typical AI generation issues.

• When generating imagery for events with limited archaeological evidence, such as the pastoral festival Lupercalia, the models inevitably drift into conjecture.
• Comical glitches have produced anachronistic results, including Roman spectators holding smartphones and Cicero appearing to speak into a microphone.
• The technology also struggles with portraying people marked by age, labor, or illness—too often it defaults to idealized, flawless figures.

Why this matters: The researchers position their tool as a complement to human imagination rather than a replacement, targeting educators, researchers, documentarians, and museums.

• The creators explain that they cannot bring back a Roman triumph or a Greek festival, but by modeling them visually they provoke a dialogue between evidence and imagination.
• The platform produces visual hypotheses designed to encourage deeper historical engagement while making users aware of gaps, uncertainties, and biases in historical knowledge.

Availability: Re-Experiencing History is currently limited to University of Zurich email addresses, though anyone can register to be notified when it becomes publicly available.