4DGS
PulseAugur coverage of 4DGS — every cluster mentioning 4DGS across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
SwitchSplat models discrete object states in 3D scenes
Researchers have developed SwitchSplat, a novel 3D Gaussian splatting representation capable of modeling discrete object states within casually captured scenes. Unlike previous methods that assume static environments or…
-
AEON framework unifies video and 3D world models with 4DGS
Researchers have introduced AEON, a novel framework designed to unify video and 3D world models. AEON bridges the gap between realistic appearance simulation found in video models like Sora and the geometric consistency…
-
Sparse Light Field Sampling Enhances 3D and 4D Reconstruction Quality
Researchers have developed a new method for 3D and 4D reconstruction that leverages sparse light field sampling from multiple viewpoints. This approach analyzes sensor-limited and exposure-limited multi-view scenarios, …
-
4DAnyone reconstructs 4D humans from single videos using novel diffusion techniques
Researchers have developed 4DAnyone, a novel framework capable of reconstructing three-dimensional human models from a single, uncalibrated video. This system generates consistent multi-view videos and then converts the…
-
New ERF-GS framework reconstructs fast motion using event-RGB fusion
Researchers have developed ERF-GS, a novel framework for reconstructing fast-moving objects in 3D scenes. This method integrates event data from high-frame-rate sensors into the Gaussian splatting pipeline, improving ac…
-
New research advances 3D Gaussian splatting for efficiency and applications
Recent research in 3D Gaussian splatting is exploring methods for efficient representation and manipulation of complex 3D scenes. Papers are introducing techniques for camera-agnostic pruning, hierarchical context aggre…
-
New AI model generates 3D dynamic humans from text in under a minute
Researchers have developed 4DHumanDiff, a novel diffusion framework capable of directly generating dynamic 3D human models from text prompts. This method bypasses the need for intermediate video synthesis or per-scene r…