Research map / Image generation / Autoregressive and latent generation
Refracting Reality: Generating Images with Realistic Transparent Objects
Yue Yin, Enze Tao, Dylan Campbell
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 panorama synchronization (ablation) | Snellcaster · Six-scene sphere subset (before relighting) | PSNR: 17.98 -> 18.21 | Improved |
| add reflection modeling (ablation) | Snellcaster · Six-scene sphere subset (before relighting) | PSNR: 18.11 -> 18.21 | Improved |
| add panorama synchronization (ablation) | Snellcaster · Six-scene sphere subset (before relighting) | CLIP score: 33.88 -> 34.22 | Improved |
| add reflection modeling (ablation) | Snellcaster · Six-scene sphere subset (before relighting) | ImageReward: 0.67 -> 0.51 | Hurt |
| add relighting post-processing (ablation) | Snellcaster · Six-scene sphere subset (before relighting) | PSNR: 18.21 -> 17.10 | Hurt |
| add relighting post-processing (ablation) | Snellcaster · Six-scene sphere subset (before relighting) | CLIP score: 34.22 -> 34.61 | Improved |
| use Snell's-law refraction synchronization instead of plain FLUX-dev generation | FLUX-dev · Transparent-object scenes (300 combos) | masked PSNR: 12.68 -> 16.51 | Improved |
| use Snell's-law refraction synchronization instead of plain FLUX-dev generation | FLUX-dev · Transparent-object scenes (300 combos) | masked LPIPS: 0.48 -> 0.24 | Improved |
| use Snell's-law refraction synchronization instead of FLUX-dev | FLUX-dev · Transparent-object scenes (300 combos) | ImageReward: -0.20 -> -0.32 | Hurt |
| use Snell's-law refraction synchronization instead of FLUX-dev | FLUX-dev · Transparent-object scenes (300 combos) | CLIP score: 32.36 -> 32.85 | Mixed |
Closest papers by meaning
- MaterialRefGS: Reflective Gaussian Splatting with Multi-view Consistent Material Inference (NeurIPS 2025 · similarity 0.87)
- Differentiable Neural Surface Refinement for Modeling Transparent Objects (CVPR 2024 · similarity 0.87)
- Reflective Gaussian Splatting (ICLR 2025 · similarity 0.86)
- MirrorVerse: Pushing Diffusion Models to Realistically Reflect the World (CVPR 2025 · similarity 0.86)
- Factored-NeuS: Reconstructing Surfaces, Illumination, and Materials of Possibly Glossy Objects (CVPR 2025 · similarity 0.86)
- Multi-times Monte Carlo Rendering for Inter-reflection Reconstruction (NeurIPS 2024 · similarity 0.86)
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.
