Research map / Video / Video segmentation and tracking
Computational Mirrors: Blind Inverse Light Transport by Deep Matrix Factorization
Miika Aittala, Prafull Sharma, Lukas Murmann, Adam Yedidia, Gregory Wornell, Bill Freeman, Fredo Durand
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 |
|---|---|---|---|
| Parameterize both factors with one-off CNNs instead of optimizing factor entries directly (ablation) | Matrix factorization · Synthetic image-like matrices | qualitative factor legibility | Improved |
| Use a non-negative matrix factorization baseline instead of CNN factor generators | Matrix factorization · Synthetic image-like matrices | qualitative factor legibility | Hurt |
| Replace the CNN factor generators with direct matrix-entry optimization and an L1 smoothness prior (ablation) | Matrix factorization · Synthetic image-like matrices | qualitative factor legibility | Hurt |
| Use the SVD-basis CNN for transport generation rather than alternating 2D convolutions (ablation) | Blind inverse light transport · Blind light transport factorization | qualitative reconstruction | Improved |
| Use blind factorization rather than a measured transport matrix | Live-action Lego reconstruction · Live-action Lego | qualitative motion reconstruction | Hurt |
| Use the proposed blind factorization rather than the adapted Levin et al. approach | Blind inverse light transport · Blind light transport factorization | qualitative reconstruction | Improved |
Closest papers by meaning
- Using Unknown Occluders to Recover Hidden Scenes (CVPR 2019 · similarity 0.85)
- UniRelight: Learning Joint Decomposition and Synthesis for Video Relighting (NeurIPS 2025 · similarity 0.85)
- Neural Inverse Rendering from Propagating Light (CVPR 2025 · similarity 0.84)
- Diffusion Renderer: Neural Inverse and Forward Rendering with Video Diffusion Models (CVPR 2025 · similarity 0.83)
- LuxDiT: Lighting Estimation with Video Diffusion Transformer (NeurIPS 2025 · similarity 0.83)
- MetaGS: A Meta-Learned Gaussian-Phong Model for Out-of-Distribution 3D Scene Relighting (NeurIPS 2025 · similarity 0.83)
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.
