AI Jesus: Historical Stock
An AI pipeline trained on first century Jewish culture to generate a historically grounded Jesus for sermons and documentaries. Built to replace the white default in stock footage with visuals closer to the lived reality of Roman period Judea, using ComfyUI and a LoRA trained on curated historical references.
This project involves training generative AI on first-century Jewish material culture to create historically accurate visual depictions of Jesus for use in sermons, documentaries, and spiritual media. The main issue is both practical and pastoral. Most stock libraries continue to default to a Europeanized image of Jesus. A brief search on popular marketplaces reveals a significant number of white portrayals, which then become the standard visual language for storytelling.
This visual standard is not impartial. It influences who audiences perceive as central to the narrative and can perpetuate the gap James Cone identified when he stated that depictions of Jesus often reflect power rather than historical accuracy. He argued that liberating theology must challenge this misrepresentation. By revisiting the cultural and geographic realities of Roman-era Judea and Galilee, the project seeks to assist creators in producing visuals that more closely align with historical research and with communities desiring a more authentic connection to the incarnation.
The workflow utilizes ComfyUI on a high-end GPU, employing a LoRA trained with carefully selected historical references related to clothing, textiles, grooming, skin tones, and built environments. This creates a reusable and controllable pipeline capable of generating consistent, production-ready full-body and scene images. These images can serve as ethical, historically informed alternatives to traditional stock footage conventions.
Project Gallery
