archive
Every paper Pith has read. Search by title, abstract, or pith.
456 papers in cs.GR · page 7
-
RL training lets small LLMs close the CAD-CAE loop
Tool-Augmented Agent for Closed-loop Optimization,Simulation,and Modeling Orchestration
-
Hypergraph contrastive learning recovers 3D crowd meshes
Contrastive Multi-Modal Hypergraph Reasoning for 3D Crowd Mesh Recovery
-
MeshTailor traces seams vertex-by-vertex on 3D mesh graphs
MeshTailor: Cutting Seams via Generative Mesh Traversal
-
Gaussian face avatars split into editable parts without labels
FaceParts: Segmentation and Editing of Gaussian Splatting
-
Confidence values improve 3D Gaussian mesh extraction accuracy
Confidence-Based Mesh Extraction from 3D Gaussians
-
Domain co-variate keeps 3D diffusion outputs photorealistic
Realiz3D: 3D Generation Made Photorealistic via Domain-Aware Learning
-
Patchwork models 2D and 3D shapes with few parameters
Patchwork: A compact representation for 3D polygonal shapes
-
Neural skinning fields enable real-time physics animation across shapes
PhysSkin: Real-Time and Generalizable Physics-Based Animation via Self-Supervised Neural Skinning
-
Agent learns to sketch one part at a time
Teaching an Agent to Sketch One Part at a Time
-
STAC cuts memory 10x for streaming 3D reconstruction
STAC: Plug-and-Play Spatio-Temporal Aware Cache Compression for Streaming 3D Reconstruction
-
Simulator keeps gradients stable through frictional deformation
Fast and Reliable Gradients for Deformables Across Frictional Contact Regimes
-
Users prefer 2ms latency over 23ms when catching balls in VR
Perceptual Requirements for Low-Latency Head-Mounted Displays
-
Multi-agent RL tracks first assistive human interactions
Learning to Assist: Physics-Grounded Human-Human Control via Multi-Agent Reinforcement Learning
-
Auto-regressive diffusion fixes 3D reconstructions in sparse views
ArtiFixer: Enhancing and Extending 3D Reconstruction with Auto-Regressive Diffusion Models
-
Gaussian process surfaces render as exponential volumes
Macrofacet Theory for Gaussian Process Statistical Surfaces
-
New color space cuts UI color-difference error by 23 percent
Helmlab: A Two-Space Family of Analytical, Data-Driven Color Spaces for UI Design Systems
-
Gaussian groups by motion enable stable scene forecasts
Space-Time Forecasting of Dynamic Scenes with Motion-aware Gaussian Grouping
-
System differentiates geometry code without reformulation
Iskra: A System for Inverse Geometry Processing
-
Neural nets turn radiation pressure sims into fast optimizers
Photons x Force: Differentiable Radiation Pressure Modeling
-
Beltrami norm equals triangle angle distortion in discrete maps
Beltrami coefficient and angular distortion of discrete geometric mappings
-
Video models yield reusable neural materials from 17 frames
VideoNeuMat: Neural Material Extraction from Generative Video Models
-
ShapeUP edits 3D shapes from 2D image prompts with trained DiT
ShapeUP: Scalable Image-Conditioned 3D Editing
-
AGILE generates full meshes to track hand-object video interactions
AGILE: Hand-Object Interaction Reconstruction from Video via Agentic Generation
-
Text prompts yield valid origami folds via LLM and world-model planning
Learn2Fold: Structured Origami Generation with World Model Planning
-
Full-body 3D Gaussian avatars run in VR from head tracking alone
VRGaussianAvatar: Integrating 3D Gaussian Avatars into VR
-
Prioritized rules separate animation from Godot gameplay scripts
Data-Driven Animation Controller: A Prioritized Visual System for Decoupled Animation Logic in Godot Game Engine
-
2D Gaussians enable editable indoor scenes with consistent lighting
EAG-PT: Emission-Aware Gaussians and Path Tracing for Diffuse Indoor Scene Reconstruction and Editing
-
LLMs build manifold 3D meshes from extrusion sequences
Learning to Build Shapes by Extrusion
-
Synthetic abundance maps train unsupervised hyperspectral sharpening
Synthetic Abundance Maps for Unsupervised Super-Resolution of Hyperspectral Remote Sensing Images
-
LoRA adapts diffusion model for joint audio-visual video dubbing
JUST-DUB-IT: Video Dubbing via Joint Audio-Visual Diffusion
-
One diffusion step upgrades real-world video in space and time
Taming Real-World Space-Time Video Super-Resolution with One-Step Diffusion
-
Graphical model adds uncertainty handling to 4D Gaussian Splatting
Graphical X Splatting (GraphiXS): A Graphical Model for 4D Gaussian Splatting under Uncertainty
-
Foreground 4D proxy guides monocular video camera redirection
FreeOrbit4D: Training-Free Arbitrary Camera Redirection for Monocular Videos via Foreground-Complete 4D Reconstruction
-
NURBS curves admit explicit rational spline inverses
Explicit Inversion of Planar NURBS Curves
-
Physically based ISP module fixes photometric drift in radiance fields
PPISP: Physically-Plausible Compensation and Control of Photometric Variations in Radiance Field Reconstruction
-
Multi-view references create 3D-consistent videos
MV-S2V: Multi-View Subject-Consistent Video Generation
-
Distance fields in latent space generate sewing patterns
Learning Sewing Patterns via Latent Flow Matching of Implicit Fields
-
One lighting edit syncs all uncalibrated scene views
SyncLight: Single-Edit Multi-View Relighting
-
Synthetic abundances train super-resolution for real hyperspectral images
Unsupervised Super-Resolution of Hyperspectral Remote Sensing Images Using Fully Synthetic Training
-
Iterative agent improves image-to-program accuracy by over 100%
Vision-as-Inverse-Graphics Agent via Interleaved Multimodal Reasoning
-
Painting flows deform 3D Gaussians into scene-aligned brushstrokes
Thinking Like Van Gogh: Structure-Aware Style Transfer via Flow-Guided 3D Gaussian Splatting
-
Relaxing RoPE balances object preservation and scene harmony
LooseRoPE: Content-aware Attention Manipulation for Semantic Harmonization
-
New AI model creates long coherent dance from music and text
Listen to Rhythm, Choose Movements: Autoregressive Multimodal Dance Generation via Diffusion and Mamba with Decoupled Dance Dataset
-
Mesh deformation optimizes uniform spiral toolpaths on complex surfaces
Topology-Preserving Scalar Field Optimization for Boundary-Conforming Spiral Toolpaths on Multiply Connected Freeform Surfaces
-
Quaternions lead 3D rotation reps in efficiency
Representations of 3D Rotations: Mathematical Foundations and Comparative Analysis
-
WorldPlay streams 720p video at 24 FPS with lasting geometry
WorldPlay: Towards Long-Term Geometric Consistency for Real-Time Interactive World Modeling
-
Past frames guide future ones for consistent animations
Screen, Cache, and Match: A Training-Free Causality-Consistent Reference Frame Framework for Human Animation
-
Diffusion models now generate infinite realistic terrain
InfiniteDiffusion: Bridging Learned Fidelity and Procedural Utility for Open-World Terrain Generation
-
Topological analysis maps neural network class separation
HOLE: Homological Observation of Latent Embeddings for Neural Network Interpretability
-
Few satellite images reconstruct entire city blocks in 3D
From Orbit to Ground: Generative City Photogrammetry from Extreme Off-Nadir Satellite Images