Research map / 3D vision / Face and shape
Refine Now, Query Fast: A Decoupled Refinement Paradigm for Implicit Neural Fields
Tianyu Xiong, Skylar Wurster, Han Wei Shen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add decoupled refiner network on cached embeddings (DRR) to embedding-only model (ablation) | DRR-Net · Nyx | PSNR (dB): 42.64 -> 44.69 | Improved |
| add decoupled refiner network to embedding-only model (ablation) | DRR-Net · MPAS-Ocean | PSNR (dB): 45.22 -> 49.26 | Improved |
| add decoupled refiner network to embedding-only model (ablation) | DRR-Net · Cloverleaf3D | PSNR (dB): 45.44 -> 48.69 | Improved |
| add condition refiner on top of spatial refiner (ablation) | DRR-Net · Nyx | PSNR (dB): 44.71 -> 44.69 | No clear effect |
| add condition refiner on top of spatial refiner (ablation) | DRR-Net · MPAS-Ocean | PSNR (dB): 48.43 -> 49.26 | Improved |
| add Variational Pairs (VP-S) augmentation (ablation) | DRR-Net · Cloverleaf3D | PSNR (dB): 47.47 -> 49.26 | Improved |
| add Variational Pairs (VP-S) augmentation (ablation) | DRR-Net · Nyx | PSNR (dB): 44.92 -> 44.61 | No clear effect |
| add Variational Coordinate (VC) augmentation (ablation) | DRR-Net · Nyx | PSNR (dB): 44.92 -> 43.11 | Hurt |
| add Variational Pairs (VP-S) augmentation (ablation) | FA-INR · Cloverleaf3D | PSNR (dB): 32.40 -> 49.13 | Improved |
| add Variational Coordinate (VC) augmentation (ablation) | Explorable-INR · Cloverleaf3D | PSNR (dB): 45.03 -> 42.41 | Hurt |
| replace MLP-based FA-INR with DRR-Net on MPAS-Ocean (unstructured mesh) | FA-INR · MPAS-Ocean | PSNR (dB): 52.13 -> 49.26 | Hurt |
| use DRR-Net instead of FA-INR for inference speed | FA-INR · Nyx | inference time (s): 287.2s -> 10.7s | Improved |
Closest papers by meaning
- Grids Often Outperform Implicit Neural Representation at Compressing Dense Signals (NeurIPS 2025 · similarity 0.87)
- PEINR: A Physics-enhanced Implicit Neural Representation for High-Fidelity Flow Field Reconstruction (ICML 2025 · similarity 0.85)
- Compute-optimal data scaling for neural surrogates via multi-fidelity training (NeurIPS 2026 · similarity 0.85)
- Derived Fields Preserve Fine-Scale Detail in Budgeted Neural Simulators (NeurIPS 2026 · similarity 0.85)
- RECOMBINER: Robust and Enhanced Compression with Bayesian Implicit Neural Representations (ICLR 2024 · similarity 0.85)
- Learning Transferable Features for Implicit Neural Representations (NeurIPS 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
