Research map / 3D vision / NeRF and radiance fields
REFRAME: Reflective Surface Real-Time Rendering for Mobile Devices
Chaojie Ji, Yufeng Li, Yiyi Liao
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 |
|---|---|---|---|
| replace direct gradient vertex updates with a hash-encoding vertex offset learner (ablation) | REFRAME · Toaster | PSNR: 17.01 -> 24.95 | Improved |
| replace direct gradient vertex updates with the vertex offset learner (ablation) | REFRAME · Materials | PSNR: 23.01 -> 29.58 (lr 0.001); 29.06 -> 29.58 (lr 0.0001) | Improved |
| add learned per-vertex normal offset (ablation) | REFRAME · Toaster (10k faces) | PSNR: 24.24 -> 24.62 | Improved |
| add learned per-vertex normal offset (ablation) | REFRAME · Materials (75k faces) | PSNR: 29.46 -> 29.58 | Improved |
| add a neural environment learner instead of feeding reflection direction to the tiny MLP (ablation) | REFRAME · Shiny Blender | rendering quality | Improved |
| learn environment feature map through an MLP instead of optimizing the map directly (ablation) | REFRAME · Toaster (900*1800) | PSNR: 23.59 -> 24.96 | Improved |
| learn environment map through MLP (ablation) | REFRAME · Toaster (900*1800) | cache size (KB): 2460 -> 504 | Improved |
| learn environment map through MLP at low resolution (ablation) | REFRAME · Materials (180*360) | PSNR: 29.58 -> 29.55 | No clear effect |
| use reflection-direction environment features and mesh geometry learner versus NeRF2Mesh | REFRAME · Shiny Blender | PSNR: 26.96 -> 36.02 | Improved |
| REFRAME mesh renderer versus 3DGS on glossy scenes | REFRAME · Shiny Blender | PSNR: 30.37 -> 36.02 | Improved |
| REFRAME on unbounded outdoor scenes with poor background meshes | REFRAME · Real Captured (3 outdoor scenes) | PSNR: 24.45 (Ref-NeRF) vs 22.50 | Hurt |
| remove anti-aliasing and bake textures for mobile rendering | REFRAME · Shiny Blender | PSNR: 36.02 -> 35.83 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- MobileNeRF — MobileNeRF: Exploiting the Polygon Rasterization Pipeline for Efficient Neural Field Rendering on Mobile Architectures (CVPR 2023) · better on 3 of 4 shared results
- Ref-NeRF — Ref-NeRF: Structured View-Dependent Appearance for Neural Radiance Fields (CVPR 2022) · better on 1 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ReFrame: Layer Caching for Accelerated Inference in Real-Time Rendering (ICML 2025 · similarity 0.87)
- Multi-times Monte Carlo Rendering for Inter-reflection Reconstruction (NeurIPS 2024 · similarity 0.86)
- Differentiable Neural Surface Refinement for Modeling Transparent Objects (CVPR 2024 · similarity 0.86)
- UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections (NeurIPS 2024 · similarity 0.85)
- Factored-NeuS: Reconstructing Surfaces, Illumination, and Materials of Possibly Glossy Objects (CVPR 2025 · similarity 0.85)
- SpecNeRF: Gaussian Directional Encoding for Specular Reflections (CVPR 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.
