Original
The significant new research award is given to a researcher who has made a recent, significant contribution to the field of computer graphics and is new to the field (i.e., received their Ph.D. or the equivalent up to seven years ago). The intent is to recognize people who have already made a notable contribution very early in their careers and are likely to make more. The award includes a $1,000 cash prize.
Current Recipient
Lingjie Liu
For outstanding contributions to neural representations and human performance animation.
ACM SIGGRAPH is pleased to present the 2026 Significant New Researcher Award to Lingjie Liu for her outstanding contributions to neural representations, human performance capture and animation, and 3D reconstruction.
Lingjie’s work on neural sparse voxel fields, NeuS, and NeuS2 has pioneered efficient sparse implicit neural representations and high-quality neural surface reconstruction. Her research on neural sparse voxel fields was among the first to dramatically accelerate neural radiance field rendering through efficient sparse implicit representations. She also contributed the innovative methods, NeuS and NeuS2, which integrate neural signed distance fields with volume rendering for accurate and efficient surface reconstruction from images. NeuS identified and addressed a fundamental limitation in signed distance field learning: bias accumulated during volume rendering, which had prevented prior methods from reconstructing complex surface geometry with high quality. NeuS2 derived an analytical second-order derivative formula for ReLU MLPs that better enforces the Eikonal loss, unlocking better surface geometry and more efficient training and rendering. Together, these contributions have substantially advanced neural rendering and established more practical and geometrically faithful approaches to image-based 3D reconstruction.
In 3D generation, Lingjie made pioneering advances in multiview consistency for 2D and 3D visual generative AI. In StyleNeRF, she enabled a 3D-aware generative model for photo-realistic high-resolution image synthesis with multi-view consistency. In SyncDreamer and Wonder3D, she pioneered multiview-consistent diffusion models that jointly generate consistent color images and normal maps, enabling efficient high-fidelity 3D generation from single images.
Lingjie has also made significant contributions to the area of human performance capture and animation. In particular, her work on Neural Actor enables flexible photorealistic free-view rendering of a person in novel poses from multi-view video input, modeling pose-dependent geometry deformations and pose- and view-dependent appearance. In PhysHMR, she developed a novel method that unifies video-based human motion estimation and physics-based human animation, enabling physically plausible human motion estimation through a learned physics-based control policy directly conditioned on video input.
She received a BEng from Huazhong University of Science and Technology in 2014 and a PhD from the University of Hong Kong in 2019 advised by Wenping Wang. She was a visiting researcher with Niloy Mitra at University College London from 2017 to 2018 and with Daniele Panozzo at New York University in 2019. She worked as a Postdoc at the Max-Planck-Institute for Informatics in Saarbrücken with Christian Theobalt and is currently the Aravind K. Joshi Assistant Professor in the Department of Computer and Information Science at the University of Pennsylvania.