Research map / 3D vision / Face and shape
Controllable Shadow Generation Using Pixel Height Maps
Yichen Sheng, Yifan Liu, Jianming Zhang, Wei Yin, A. Cengiz Oztireli, He Zhang, Zhe Lin, Eli Shechtman, Bedrich Benes
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 Y-coordinate map input (ablation) | HENet · Real1500 val (sparse points), Pixel Height estimation | Abs error: 6.04 -> 5.92 | Improved |
| add Y-coordinate map input (ablation) | HENet · Real1500 val (sparse points), Pixel Height estimation | rel error: 1.98 -> 1.94 | Improved |
| add total variation loss (ablation) | HENet · Real1500 val (sparse points), Pixel Height estimation | Abs error: 6.12 -> 5.92 | Improved |
| add total variation loss (ablation) | HENet · Real1500 val (sparse points), Pixel Height estimation | rel error: 2.01 -> 1.94 | Improved |
| add real images to synthetic training data (ablation) | HENet · Real1500 val (sparse points), Pixel Height estimation | Abs error: 7.05 -> 5.92 | Improved |
| add real images to synthetic training data (ablation) | HENet · Real1500 val (sparse points), Pixel Height estimation | rel error: 2.34 -> 1.94 | Improved |
| predict pixel height with HENet instead of using Y-coordinate map directly | height estimation · Real1500 val (sparse points), Pixel Height estimation | Abs error: 10.84 -> 5.92 | Improved |
| render soft shadow from hard shadow plus softness (SSG) instead of mapping lightmap to shadow (SSN) | soft shadow generation · Soft shadow benchmark (20 unseen 3D assets) | Mean ZNCC: 0.370 -> 0.788 | Improved |
| use SSG instead of SSN | soft shadow generation · Soft shadow benchmark, hard shadows | Abs: 0.039 -> 0.025 | Improved |
| use SSG instead of SSN | soft shadow generation · Soft shadow benchmark, short shadows | Abs: 0.031 -> 0.033 | No clear effect |
| use SSG instead of SSN | shadow realism · User study 2AFC on natural images (mean preference rate) | preference rate: 0.26 -> 0.74 | Improved |
Closest papers by meaning
- Shadow Generation Using Diffusion Model with Geometry Prior (CVPR 2025 · similarity 0.87)
- Gaussian Shadow Casting for Neural Characters (CVPR 2024 · similarity 0.85)
- Shadow Generation with Decomposed Mask Prediction and Attentive Shadow Filling (AAAI 2024 · similarity 0.85)
- RNG: Relightable Neural Gaussians (CVPR 2025 · similarity 0.84)
- Shadow Generation for Composite Image Using Diffusion Model (CVPR 2024 · similarity 0.84)
- SoftShadow: Leveraging Soft Masks for Penumbra-Aware Shadow Removal (CVPR 2025 · similarity 0.84)
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.
