The P2.0 model represents a pivot from experimental AI output to technical utility. According to Chief Scientist Dr. Yanpei Cao, the core innovation lies in generating edge flows that mimic manual artist layouts, bridging the gap between generative speed and studio requirements. The system supports up to 25,000 faces for quad-based assets, ensuring compatibility with professional rigging and animation software.
Beyond mesh generation, the update introduces granular control tools. A new Mesh Edit feature allows creators to refine specific regions of an object without triggering a full regeneration. Coupled with a Smart UV function that automates 2D texture unwrapping, the model is designed to accelerate workflows for hard-surface objects, vehicles, and complex characters. Users can now generate four variations from a single prompt, offering flexibility in asset density and detail for interactive environments.
Comments (0)
No comments yet. Be the first!