Research map / 3D vision / 3D generation
Dynamic Inverse Rendering for Enhanced Material-Lighting Decomposition
Raza Yunus, Benjamin Ummenhofer, Jan Eric Lenssen, Eddy Ilg
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 |
|---|---|---|---|
| capture rigid hand-held motion instead of static dome (estimated poses) | Ours · Synthetic diffuse materials | albedo PSNR (dB): 32.68 -> 40.33 | Improved |
| hand-held capture instead of static dome | Ours · Synthetic original materials | relighting PSNR (dB): 38.14 -> 40.24 | Improved |
| hand-held rotations instead of single-axis turntable | Ours · Synthetic diffuse materials | albedo PSNR (dB): 35.65 -> 40.55 | Improved |
| turntable instead of static dome (GT poses) | Ours · Synthetic original materials | albedo PSNR (dB): 36.72 -> 36.15 | Hurt |
| use estimated instead of ground-truth poses in hand-held setting | Ours · Synthetic diffuse materials | albedo PSNR (dB): 40.55 -> 40.33 | No clear effect |
| our pipeline vs IRGS in static setting | Ours · TensoIR synthetic | albedo PSNR (dB): 33.41 -> 33.96 | Improved |
| our pipeline vs IRGS in static setting | Ours · TensoIR synthetic | normal MAE (deg): 3.99 -> 4.19 | Hurt |
| our SDF-based Gaussian pipeline vs extended IRGS on hand-held sequences | Ours · Hand-held sequences | relighting PSNR (dB): 32.99 -> 39.65 | Improved |
| share more frames per environment map in ReCap baseline (ablation) | ReCap · Hand-held sequences | albedo PSNR (dB): 21.91 -> 22.42 (10 vs 40 envs) | No clear effect |
| 3D Gaussian refinement with better matcher vs FMOV | Ours · HO3D | AUC ATE: 31.19 -> 35.26 | Improved |
| hand-held views instead of more static views (25 vs 400) (ablation) | Ours · synthetic | decomposition quality | Improved |
| uniform white lighting for hand-held capture (ablation) | Ours · synthetic | albedo quality | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- IRGS — IRGS: Inter-Reflective Gaussian Splatting with 2D Gaussian Ray Tracing (CVPR 2025) · better on 2 of 2 shared results
- SVG-IR — SVG-IR: Spatially-Varying Gaussian Splatting for Inverse Rendering (CVPR 2025) · better on 2 of 2 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
- Eclipse: Disambiguating Illumination and Materials using Unintended Shadows (CVPR 2024 · similarity 0.90)
- Inverse Rendering of Glossy Objects via the Neural Plenoptic Function and Radiance Fields (CVPR 2024 · similarity 0.90)
- Factored-NeuS: Reconstructing Surfaces, Illumination, and Materials of Possibly Glossy Objects (CVPR 2025 · similarity 0.90)
- ReCap: Better Gaussian Relighting with Cross-Environment Captures (CVPR 2025 · similarity 0.89)
- Neural Inverse Rendering from Propagating Light (CVPR 2025 · similarity 0.89)
- Inverse Rendering using Multi-Bounce Path Tracing and Reservoir Sampling (ICLR 2025 · similarity 0.88)
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.
