Research map / 3D vision / 3D generation
ShadowDraw: From Any Object to Shadow-Drawing Compositional Art
Rundong Luo, Noah Snavely, Wei-Chiu Ma
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 |
|---|---|---|---|
| condition on shadow contour instead of object-shadow composite (ablation) | line drawing generator (artist-source data) · Shadow-drawing (200 objects, top-4) | CLIP: 31.04 -> 31.38 | Improved |
| condition on shadow contour instead of object-shadow composite (ablation) | line drawing generator (artist-source data) · Shadow-drawing (200 objects, top-4) | Conceal: 0.225 -> 2.215 | Improved |
| condition on shadow contour instead of object-shadow composite (ablation) | line drawing generator (artist-source data) · Shadow-drawing (200 objects, top-4) | ImageReward: -0.0720 -> 0.1552 | Improved |
| condition on shadow contour instead of object-shadow composite (ablation) | line drawing generator (artist-source data) · Shadow-drawing (200 objects, top-4) | HPS: 0.2244 -> 0.2269 | Improved |
| train on large synthetic dataset instead of artist-sourced images (ablation) | shadow contour model · Shadow-drawing (200 objects, top-4) | CLIP: 31.38 -> 32.08 | Improved |
| train on large synthetic dataset instead of artist-sourced images (ablation) | shadow contour model · Shadow-drawing (200 objects, top-4) | Conceal: 2.215 -> 2.606 | Improved |
| train on large synthetic dataset instead of artist-sourced images (ablation) | shadow contour model · Shadow-drawing (200 objects, top-4) | ImageReward: 0.1552 -> 0.4177 | Improved |
| add scene parameter optimization (ablation) | shadow contour + synthetic data model · Shadow-drawing (200 objects, top-4) | CLIP: 32.08 -> 32.41 | Improved |
| add scene parameter optimization (ablation) | shadow contour + synthetic data model · Shadow-drawing (200 objects, top-4) | Conceal: 2.606 -> 3.006 | Improved |
| add scene parameter optimization (ablation) | shadow contour + synthetic data model · Shadow-drawing (200 objects, top-4) | ImageReward: 0.4177 -> 0.4441 | Improved |
| add scene parameter optimization (ablation) | shadow contour + synthetic data model · Shadow-drawing (200 objects, top-4) | HPS: 0.2294 -> 0.2373 | Improved |
| use trained shadow-contour pipeline instead of Gemini conditioned on shadow contour | ShadowDraw · Shadow-drawing (200 objects, top-4) | Conceal: 0.2421 -> 3.0059 | Improved |
Closest papers by meaning
- Learning to Shadow Hand-Drawn Sketches (CVPR 2020 · similarity 0.90)
- RASP: Revisiting 3D Anamorphic Art for Shadow-Guided Packing of Irregular Objects (CVPR 2025 · similarity 0.86)
- ArtiScene: Language-Driven Artistic 3D Scene Generation Through Image Intermediary (CVPR 2025 · similarity 0.84)
- MetaShadow: Object-Centered Shadow Detection, Removal, and Synthesis (CVPR 2025 · similarity 0.84)
- LineArt: A Knowledge-guided Training-free High-quality Appearance Transfer for Design Drawing with Diffusion Model (CVPR 2025 · similarity 0.84)
- Shadow Generation Using Diffusion Model with Geometry Prior (CVPR 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.
