Research map / Video / Video generation
Latent Knowledge-Guided Video Diffusion for Scientific Phenomena Generation from a Single Initial Frame
Qinglong Cao, Xirui Li, Ding Wang, Chao Ma, Yuntian Chen, Xiaokang Yang
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 latent scientific knowledge and quaternion prompts to SVD | SVD · DamBreak simulation | RMSE: 8.189 -> 4.921 | Improved |
| add latent scientific knowledge and quaternion prompts to CogVideoX | CogVideoX · DepthCharge simulation | RMSE: 4.775 -> 3.990 | Improved |
| use method vs LoRA fine-tuning on typhoon data | SVD · Typhoon 202001 | SSIM: 0.365 -> 0.446 | Improved |
| add the method on typhoon continuity score | SVD/CogVideoX · Typhoon 202001 | CS: 16.66 -> 19.19 | Hurt |
| add static knowledge (MAE embeddings) (ablation) | Ours(SVD) · Cylinder simulation | RMSE: 6.021 -> 5.852 | Improved |
| add dynamic knowledge (optical flow embeddings) (ablation) | Ours(SVD) · Cylinder simulation | RMSE: 5.983 -> 5.852 | Improved |
| add spatial-domain modeling (ablation) | Ours(SVD) · Cylinder simulation | RMSE: 6.423 -> 5.852 | Improved |
| add frequency-domain modeling (ablation) | Ours(SVD) · Cylinder simulation | RMSE: 5.923 -> 5.852 | Improved |
| use quaternion network instead of linear projection (ablation) | Ours(SVD) · Cylinder simulation | RMSE: 5.913 -> 5.852 | Improved |
| use quaternion projection instead of cross attention (ablation) | Ours(SVD) · Cylinder simulation | SFE: 70.012 -> 62.456 | Improved |
| vary LoRA rank from 4 to 2 or 8 (ablation) | Ours(SVD) · Cylinder simulation | RMSE: 5.852 -> 6.132 (rank 2), 6.018 (rank 8) | Hurt |
Closest papers by meaning
- Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition (ICLR 2024 · similarity 0.86)
- LaMo: Self-Supervised Latent Motion Priors for Physical Realism in Video Generation (NeurIPS 2026 · similarity 0.86)
- MeDM: Mediating Image Diffusion Models for Video-to-Video Translation with Temporal Correspondence Guidance (AAAI 2024 · similarity 0.85)
- Lost in Latent Space: An Empirical Study of Latent Diffusion Models for Physics Emulation (NeurIPS 2025 · similarity 0.85)
- LLM-grounded Video Diffusion Models (ICLR 2024 · similarity 0.85)
- SEINE: Short-to-Long Video Diffusion Model for Generative Transition and Prediction (ICLR 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.
