Wenping Wang

A new method to speed up finding distances from points to mesh surfaces.

SyncDreamer enables creating multiple views from a single image easily.

New method improves surface extraction from neural unsigned distance fields.

A new approach to surface modeling using neural networks.

A method to train animated characters for realistic movements.

Model2Scene uses CAD models and language to improve 3D scene learning.

DirDist offers a better way to measure differences between 3D shapes without point correspondences.

Event-based cameras enhance the reading of blurred barcodes and signs.

New dataset aims to enhance robot object manipulation using human-like movements.

New method improves facial reconstruction accuracy using advanced image generation techniques.

New methods improve 3D model efficiency while retaining essential details.

A novel method for creating detailed 3D images from single images using multiview diffusion.

This paper presents new sampling methods for better generative model performance.

Deep-PE enhances pose selection accuracy in low-overlap point cloud scenarios.

DreamMat simplifies the generation of realistic PBR materials from text and 3D models.

A new method generates views from a single video, improving ease of use.

A new method improves reconstruction of hand-face interactions for AR and VR.

A new method improves how we work with complex 3D shapes.

Scientists develop methods to help robots predict human movements accurately.

Discover how SuperGaussians enhance image synthesis for realistic views.

Align3R ensures accurate depth estimation in dynamic videos with enhanced consistency.

A new method uses 3D scans for face predictions post-surgery.

Discover how AURORA streamlines virtual interior design for professionals and enthusiasts.