Paper entries are structured for thumbnails, abstracts, venues, years, tags, and external links.
Replace the placeholders with real paper metadata, images, and links.
ControlFace generates computer-graphics-friendly 3D faces from text and image guidance using a controllable diffusion model trained with few-shot 3D facial assets.
UniFaceGAN builds a unified latent space for high-quality 3D facial editing, supporting stylization, point-based drag manipulation, and local blending.
DreamPBR generates high-resolution physically based materials and SVBRDF parameter maps from text and multimodal controls.
DPWD embeds robust watermarks into diffusion models for point cloud generation, protecting model ownership while preserving generation quality.
Stable Video Signature integrates watermarking into latent diffusion video generation with temporal consistency and robustness against attacks.