Research map / Video / Video generation
Omni-Effects: Unified and Spatially-Controllable Visual Effects Generation
Fangyuan Mao, Aiming Hao, Jintao Chen, Dongxia Liu, Xiaokun Feng, Jiashu Zhu, Meiqi Wu, Chubin Chen, Jiahong Wu, Xiangxiang Chu
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 |
|---|---|---|---|
| replace single LoRA / mixed LoRA with LoRA-MoE | CogVideoX-5B · OpenVFX | FVD: 2026 (Single LoRA) / 2041 (Mix LoRA) -> 1628 | Improved |
| replace single LoRA / mixed LoRA with LoRA-MoE | CogVideoX-5B · OpenVFX | Dynamic Degree: 0.60 -> 0.66 | Improved |
| scale LoRA-MoE from 4 experts Top-1 to 8 experts Top-2 (ablation) | LoRA-MoE · OpenVFX | FVD: 1762 -> 1628 (params 18.9M -> 28.5M) | Improved |
| scale LoRA-MoE from 4 experts Top-1 to 8 experts Top-2 (ablation) | LoRA-MoE · OpenVFX | Dynamic Degree: 0.65 -> 0.66 | No clear effect |
| mixed training of VFX in a shared LoRA instead of per-effect LoRA | CogVideoX-5B · OpenVFX | FVD: 2026 -> 2041 | No clear effect |
| add Spatial-Aware Prompt (SAP) with IIF instead of ControlNet | CogVideoX · Single-VFX control (Melt, Levitate, Explode) | ECR: 0.54 -> 0.88 | Improved |
| add SAP with IIF instead of ControlNet for multi-VFX | CogVideoX · Multi-VFX Melt+Levitate | ECR: 0.05 -> 0.41 | Improved |
| SAP+IIF versus ControlNet on regional motion strength for Melt+Levitate | Omni-Effects · Multi-VFX Melt+Levitate | RDD: 3.18 -> 2.63 | Hurt |
| remove SAP attention masks from regions 2, 6, 7 (relax information flow) (ablation) | SAP+IIF · multi-VFX qualitative | effect leakage | Hurt |
| use full attention without any mask (ablation) | SAP+IIF · multi-VFX qualitative | spatial control | Hurt |
Closest papers by meaning
- EffectMaker: Unifying Reasoning and Generation for Customized Visual Effect Creation (CVPR 2026 · similarity 0.88)
- Tuning-free Visual Effect Transfer across Videos (ECCV 2026 · similarity 0.87)
- Prompt2Effect: Training-Free LoRA Synthesis for Controllable Video Effects (ECCV 2026 · similarity 0.85)
- ViTeX-Bench: Benchmarking High-Fidelity Video Scene Text Editing (NeurIPS 2026 · similarity 0.83)
- Generative Omnimatte: Learning to Decompose Video into Layers (CVPR 2025 · similarity 0.83)
- IllumiCraft: Unified Geometry and Illumination Diffusion for Controllable Video Generation (NeurIPS 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.
