First trailer (2025)
Raw demo (2025)
Majo is an AI/XR character that can converse, make (AI) art, print it and even see images in which she can recognize herself.
She was developed as my Bachelor Thesis “Storytelling with AI driven Characters in Mixed Reality” for the study program Augmented and Virtual Reality Design — at the Darmstadt University of Applied Sciences during the Winter semester 24/25.
Task:
To take a resolution to my conflict with the "lifelessness" of artificial intelligence through symbiotic AI-human experimentation, where I explore three approaches: the artist and his exploration of authorship, the technician and his delight with tinkering, and the human being with his heartfelt need of connection. Can artificial intelligence be imbued with life as if they were a blank canvas? Can it transcend replication and, through human partnership, expand the boundaries of possibility?
Idea:
Bringing to life an AI character named Majo, a virtual being who adopts the role of a young artist girl going through the Artist’s Dilemma. She thinks and expresses herself through AI, manifests through Mixed Reality on a spatial scale and attempts to find meaning through her interaction with humans. Majo invites users into her narrative journey, becoming a collaborative canvas where her emerging personality and artistic potential are co-created in real-time.
Solution:
A storytelling experience where Majo, an AI driven artist, comes to life in Mixed Reality through interaction, technology and creativity. By sensing her surroundings, understanding context, and generating art, she transforms human-computer interaction into an opportunity to co-create something novel. Her responses and artwork evolve with each interaction, blending in collaborative experience with each person, challenging the boundaries between human and artificial creativity.
Software:
Unity 2022.3
Meta Presence SDK
Ulip-sync (open source)
URP-PSX shader (open source)
AI:
LLM: Llama 3.2 (Groq)
VLM: Sonnet 3.5 (Claude)
Speech to Text: Whisper (Hugging Face)
Text to Speech: ElevenLabs
Text to Image: StabilityAI
Hardware:
Meta Quest 3
A very beefy PC
Cloud-based AI





