Research map / Video / Video generation
Semantic-Aware, Physics-Informed, Geometry-Grounded Weather Synthesis
Chenghao Qian, Nedko Savov, Lingdong Kong, Yeying Jin, Rui Song, Wenjing Li, Zhun Zhong, Jiaqi Ma, Gustav Markkula, Luc Van Gool
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 particle conditioning, appearance anchoring and geometry to VACE | Wan2.1-VACE · Weather synthesis (overall) | VLM score: 0.75 -> 0.83 | Improved |
| add particle conditioning to VACE | Wan2.1-VACE · Weather synthesis (overall) | CLIP-D: 0.16 -> 0.17 | Improved |
| add particle conditioning to VACE | Wan2.1-VACE · Weather synthesis (overall) | CLIP-S: 0.82 -> 0.81 | Hurt |
| add particle conditioning to VACE | Wan2.1-VACE · Human evaluation | photo-realism: 3.58 -> 4.16 | Improved |
| add particle conditioning to VACE | Wan2.1-VACE · Human evaluation | physical-realism: 3.52 -> 4.12 | Improved |
| use full method instead of VACE for snow preference | Wan2.1-VACE · User preference (snow) | % chosen best: 26.1 -> 56.5 | Improved |
| add semantic parsing and effects reasoning to the image-edit prompt (ablation) | appearance anchoring · initial-frame editing | qualitative | Improved |
| add scene captioning to the video prompt (ablation) | appearance anchoring · video synthesis | qualitative | Improved |
| add gravity direction matching (ablation) | geometry grounding · tilted-camera videos | qualitative | Improved |
| add camera projection of particles (ablation) | geometry grounding · forward-driving video | qualitative | Improved |
| train segmentation with synthesized weather data | DAFormer / HRDA · ACDC and MUSES | mIoU: at least 5.4% and up to 14.5% | Improved |
Closest papers by meaning
- WeatherEdit: Controllable Weather Editing with 4D Gaussian Field (AAAI 2026 · similarity 0.89)
- End-to-End Video Semantic Segmentation in Adverse Weather using Fusion Blocks and Temporal-Spatial Teacher-Student Learning (NeurIPS 2024 · similarity 0.86)
- STORM: Spatio-TempOral Reconstruction Model For Large-Scale Outdoor Scenes (ICLR 2025 · similarity 0.85)
- Adaptive Flow Matching for Resolving Small-Scale Physics (ICML 2025 · similarity 0.85)
- ChronoMagic-Bench: A Benchmark for Metamorphic Evaluation of Text-to-Time-lapse Video Generation (NeurIPS 2024 · similarity 0.85)
- Semantic Segmentation in Multiple Adverse Weather Conditions with Domain Knowledge Retention (AAAI 2024 · similarity 0.85)
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.
