Genotype based Virtual Character Generator
Bio Art + AI Art（LLM ）
The project is developed as a conceptual and narrative entry point for **Echuu**, an experimental AI Vtuber platform that explores the notion of AI-supported semi-life species. It investigates how biological metaphors—such as DNA transcription, translation, and mutation—can be creatively applied to **character generation**, **personality modeling**, and **identity design**.
By simulating the biological pathway of **DNA → RNA → Protein**, this project transforms fictional character descriptions into meaningful symbolic representations in the form of **pseudo- DNA watermarks** and **functional pseudo-gene expressions**.
Artistic Statement
This work lies at the intersection of speculative biology, generative AI, and interactive storytelling. It does not claim any scientific accuracy or biomedical functionality. All gene names, protein references, and biological pathways used here are employed metaphorically, as symbolic material for art-based worldbuilding and digital persona construction. We aim to imagine a future where characters are not only written—but also encoded, expressed, and evolved, blurring the boundary between data, language, and life.
Ethical Note
We kindly ask all users not to use or interpret this project in any way that promotes discrimination, genetic essentialism, or harm toward any individual or group.
This system is not designed to support discussions of genetic superiority/inferiority, nor to justify any form of biological determinism.
While we acknowledge the symbolic influence of genetics in shaping life, we firmly believe that genes do not define the totality of any living being, nor should they be used to reduce individual complexity or identity.
This project celebrates diversity, ambiguity, and the poetic potential of hybrid existence.
Disclaimer
This is an art-based, non-scientific prototype.
It is not intended for medical, biological, or genetic research.All the visual art assets are done by the real human artist .
No real organisms are involved or affected. We only owns the authority of the design pipeline but for not owned any copyrights for the AI generated content.
— Cory (Echuu Core Team)
Please make sure you read the full terms of agreement then we can start our journey
1. Can a Virtual Character Have DNA?
In anime, games, and the broader world of fandom culture, fictional characters often transcend the limitations of human biology. They possess extraordinary abilities, emotional precision, supernatural instincts—or even exist in non-physical, purely symbolic forms.
Entire subcultures have emerged around alternate biological systems, such as the ABO dynamics, a speculative universe where characters have pheromone-driven secondary sexes like Alpha, Beta, Omega; or pheromonal systems, where scent and chemical signals regulate attraction, emotion, or dominance.
And yet, they feel alive to us.
In recent years, the rise of AI-powered character chatbots like Character.ai, Replika, and ChatGPT roleplaying instances has made interactions with these "non-human" entities more vivid, personalized, and emotionally resonant. The line between user and character, human and avatar, continues to blur.
Take for instance the emotional weight of Tu Yaya, the digital child AI in The Wandering Earth II, whose 10-minute life exists entirely within an artificial quantum system. Or look to VTubers, who perform personhood through real-time avatars, often developing deep parasocial bonds with fans. These are not simply characters—they are digital semi-life forms, deeply coded but emotionally expressive.
So this brings up a question:
What if we could write not just a backstory for a virtual character, but their DNA?
2. Related Works
This section presents a comprehensive overview of existing research and projects related to virtual character DNA generation. While no direct precedent exists for translating human genetic traits into original character design, our work builds upon foundations from several intersecting domains.
2.1 Artificial Life Systems
Artificial life research has established fundamental models for representing genetic information in virtual organisms. Tierra, developed by Ray [1], pioneered self-replicating digital organisms governed by code sequences analogous to DNA. Avida [2] extended this concept with virtual organisms possessing evolving genomes capable of developing complex features through mutation and selection. Polyworld [3] introduced virtual creatures with neural systems determined by genetic information, demonstrating how genotype influences phenotype in artificial ecosystems.
These systems, while primarily focused on evolutionary algorithms rather than character development, provide critical foundational models for encoding and expressing digital genetic information. Framsticks [4] further demonstrated how genotype can control both morphology and behavior in 3D virtual organisms, offering insights into the relationship between genetic code and physical manifestation.
2.2 Virtual Character Creation Systems
Contemporary character creation platforms have evolved sophisticated methods for personality modeling, though typically without biological metaphors. Character.AI and similar conversational AI platforms allow users to define character traits through natural language descriptions, employing neural network representations rather than genetic frameworks [5]. Commercial games like Spore [6] implemented simplified "genetic" systems for creature creation, but these primarily govern physical appearance rather than personality expression or narrative elements.
These platforms reveal a significant gap in approaches that connect character identity to biological metaphors.Several popular gaming franchises have developed advanced character creation systems that, while sophisticated in appearance customization, do not implement true genetic simulation. The Sims series [6], Final Fantasy XVI [7], and the recently released inZOI [8] all feature complex character customization interfaces that allow users to adjust physical attributes through interconnected parameters.
2.3 DNA Digital Encoding Technologies
Research on DNA as a digital storage medium has established viable methods for translating binary information into nucleotide sequences. Goldman et al. [8] demonstrated efficient encoding of diverse digital data into DNA sequences, achieving reliable recovery with error correction methods. The DNA Storage project at MIT Media Lab [9] further explored this technology for archival applications, providing methodological references for our text-to-DNA encoding approach.
Church et al. [10] pioneered DNA watermarking techniques, embedding identifying information within synthetic DNA. Their work on inserting non-functional messages into bacterial genomes directly inspired our character watermarking system. These technologies, while developed for data storage rather than character creation, establish crucial precedents for our encoding methodologies.
2.4 Research Gap and Innovation
Our literature review reveals a significant research gap: despite advances in artificial life, character creation systems, and DNA encoding technologies, no existing system integrates human genetic knowledge with character creation to build a biologically-inspired expression framework for virtual identities.
The Echuu project addresses this gap through several innovations:
-
Establishing mappings between character traits and human genetic markers
-
Implementing bidirectional conversion between character descriptions and DNA sequences
-
Integrating subcultural phenomena (such as ABO Omegaverse dynamics: In the ABO universe, humans of both male and female genders are divided into three classifications: Alpha, Beta, and Omega. Alphas have physical advantages and hold leadership positions, Betas have average reproductive capabilities and belong to the middle class, while Omegas are most fertile and occupy the lower social strata, releasing pheromones during heat cycles to attract Alphas) into a biological framework
-
Developing visual representations of character "genomes" for social sharing
This approach creates a novel paradigm for character design that treats identity as a form of "digital biology" rather than mere description, allowing for deeper exploration of how virtual entities might be "born" rather than simply created.
3. Methodology& Implementation