Research map / 3D vision / NeRF and radiance fields
CLIPascene: Scene Sketching with Different Types and Levels of Abstraction
Yael Vinker, Yuval Alaluf, Daniel Cohen-Or, Ariel Shamir
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 |
|---|---|---|---|
| learned implicit simplification with scene decomposition instead of CLIPasso with fixed stroke count | CLIPascene · User study vs CLIPasso | % preferring: 84.8% vs 7.0% | Improved |
| use deeper CLIP-ViT layers for the loss | CLIPascene · fidelity axis, MS-SSIM | MS-SSIM: 0.39 -> 0.17 | Improved |
| use ViT-based CLIP instead of ResNet CLIP for guidance (ablation) | CLIPascene · scene sketching (qualitative) | scene coverage | Improved |
| add foreground-background separation (ablation) | CLIPascene · scene sketching (qualitative) | visual quality | Improved |
| learn simplification with ratio loss instead of fixing stroke count (ablation) | CLIPascene · simplicity axis (qualitative) | smoothness | Improved |
| keep fine-tuning MLP_loc during simplification (ablation) | CLIPascene · simplicity axis (qualitative) | semantic preservation | Improved |
| use exponential rather than linear factor schedule f_k (ablation) | CLIPascene · simplicity axis (qualitative) | perceptual smoothness | Improved |
| replace ratio loss with target stroke count loss (ablation) | CLIPascene · simplicity axis (qualitative) | gradual abstraction | Hurt |
Closest papers by meaning
- SketchyScene: Richly-Annotated Scene Sketches (ECCV 2018 · similarity 0.89)
- SketchAA: Abstract Representation for Abstract Sketches (ICCV 2021 · similarity 0.88)
- SketchingReality: From Freehand Scene Sketches to Photorealistic Images (ICLR 2026 · similarity 0.88)
- Open Vocabulary Semantic Scene Sketch Understanding (CVPR 2024 · similarity 0.87)
- Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch (CVPR 2025 · similarity 0.86)
- How to Handle Sketch-Abstraction in Sketch-Based Image Retrieval? (CVPR 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.
