Fus3D: Decoding Consolidated 3D Geometry from Feed-Forward Geometry Transformer Latents
project page / paper / arxiv / code
Fusing features of feed-forward geometry transformers into a consolidated 3D grid-structured latent space, decoding to SDFs.
Laura Fink
I'm a researcher working at the intersection of artificial intelligence, computer vision, and computer graphics. I explore how learning, geometry, and the physics of light can help us to capture and model the world in 3D. Currently, I'm a Visiting Postdoctoral Researcher at Inria, working with George Drettakis in the GraphDeco group. I received my doctorate (Dr.-Ing.) from Friedrich-Alexander-Universität Erlangen-Nürnberg, supervised by Marc Stamminger at the Chair of Visual Computing.
Mail / Scholar / Personal GitHub / LinkedIn / CV
Research
My main research is on the intersection of computer graphics, vision, and machine learning. First author papers are highlighted.
Refinement of Monocular Depth Maps via Multi-View Differentiable Rendering
project page / paper / arxiv / video / code
Dense and multi-view consistent depth maps from an monoscopically estimated initialization.
VR-Splatting: Foveated Radiance Field Rendering via 3D Gaussian Splatting and Neural Points
project page / paper / code
Foveated radiance field rendering with smooth Gaussian peripheral and neural point splats for crisp foveal rendering.
TRIPS: Trilinear Point Splatting for Real-Time Radiance Field Rendering
project page / paper / video / code
Rendering and optimizing neural point clouds via trilinear point splatting and tiny neural networks for novel-view synthesis.
LiveNVS: Neural View Synthesis on Live RGB-D Streams
project page / paper / video / code
Novel view synthesis on live RGB-D streams with feedback during capturing, robust to slam loop closures.
VET: Visual Error Tomography for Point Cloud Completion and High-Quality Neural Rendering
project page / paper / video / code
Improving point-based novel view synthesis quality by completing point cloud with 3D error volumes from 2D error maps.
Inovis: Instant Novel View Synthesis
project page / paper / video / code
Point-based novel view synthesis without per scene preprocessing or training using neural image based rendering. (* Joint first authors.)
Efficient Rendering for Light Field Displays using Tailored Projective Mappings
paper / presentation / code
Rendering only relevant fragments for parallax-based light field displays through device-tailored projection matrices.
UniTorch - Integrating Neural Rendering into Unity
Discussion and use-case study on the integration of neural rendering into the Unity game engine.
Time-Warped Foveated Rendering for Virtual Reality Headsets
paper / presentation / supplemental material
Foveated Rendering via temporal forward reprojection for fast, imperceptible VR rendering.
Lumipath - Towards real-time physically-based rendering on embedded devices
Spherical lightfield rendering based on Fibonacci point sets for immersive AR.
Hybrid Mono-Stereo Rendering in Virtual Reality
Approximating far regions of stereo-image pairs by a monoscopic centered view for VR.