AI RESEARCH
REVIVE 3D: Refinement via Encoded Voluminous Inflated prior for Volume Enhancement
arXiv CS.CV
•
ArXi:2604.27504v1 Announce Type: new Recent generative models have shown strong performance in generating diverse 3D assets from 2D images, a fundamental research topic in computer vision and graphics. However, these models still struggle to generate voluminous 3D assets when the input is a flat image that provides limited 3D cues. We