Persona.ai
I worked extensively with researchers on image and geometry data processing, Unity prototyping, defined and designed early product vision for AI avatar generation by bridging research results and user research.
Reach out to learn about the product and read about the research paper here.
First Layer Face Generation
The user enters a text prompt for face generation. Persona.ai utilizes GANs to generate high-quality facial texture maps and 3DMM to create base geometries.
Reinforcement learning from human input
Three results will be generated from the text prompt for the user to choose from. The user selects their preferred result, which will then be used for reinforcement learning from human input.
Facial Feature Editing from Input
Users can further edit specific facial features by adding text prompts to the on-face edit areas they wish to change. The generated face will be added to the change history.
Export To Target Platforms
Once the user is satisfied with the result, they can share the editor via a web link or export it to the target platform, with options for different asset qualities and the addition of a rigged body.