Research map / Video / Video generation
Dynamic-eDiTor: Training-Free Text-Driven 4D Scene Editing with Multimodal Diffusion Transformer
Dong In Lee, Hyungjun Doh, Seunggeun Chi, Runlin Duan, Sangpil Kim, Karthik Ramani
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 STGA (ablation) | Dynamic-eDiTor without STGA/CTP · DyNeRF (14 prompts, 1FPS) | Warping error (1e-3): 56.98 -> 38.64 | Improved |
| add STGA (ablation) | Dynamic-eDiTor without STGA/CTP · DyNeRF (14 prompts, 1FPS) | MEt3R (1e-1): 1.0721 -> 0.9277 | Improved |
| add CTP (ablation) | Dynamic-eDiTor without STGA/CTP · DyNeRF (14 prompts, 1FPS) | Warping error (1e-3): 56.98 -> 29.44 | Improved |
| add CTP on top of STGA (ablation) | Dynamic-eDiTor with STGA · DyNeRF (14 prompts, 1FPS) | PSNR: 28.08 -> 29.25 | Improved |
| add STGA (ablation) | Dynamic-eDiTor with CTP only · DyNeRF (14 prompts, 1FPS) | CLIP directional similarity: 0.1944 -> 0.1849 | Hurt |
| add CTP on top of STGA (ablation) | Dynamic-eDiTor with STGA · DyNeRF (14 prompts, 1FPS) | CLIP directional similarity: 0.1872 -> 0.1849 | Hurt |
| add Full Token Inheritance to CTP-Flow (ablation) | CTP · DyNeRF (14 prompts, 1FPS) | MEt3R (1e-1): 0.9205 -> 0.9074 | Improved |
| add Flow-guided Token Replacement to CTP-Full (ablation) | CTP · DyNeRF (14 prompts, 1FPS) | Warping error (1e-3): 33.22 -> 28.94 | Improved |
| restrict STGA to first 30 MM-DiT layers rather than all layers (ablation) | STGA · DyNeRF (14 prompts, 1FPS) | qualitative artifacts | Improved |
| Dynamic-eDiTor vs CTRL-D | 4D editing · DyNeRF (14 prompts, 1FPS) | PSNR: 31.06 -> 29.25 | Hurt |
| Dynamic-eDiTor vs Instruct4D-to-4D | 4D editing · DyNeRF (14 prompts, 1FPS) | CLIP directional similarity: 0.1077 -> 0.1849 | Improved |
Closest papers by meaning
- Instruct 4D-to-4D: Editing 4D Scenes as Pseudo-3D Scenes Using 2D Diffusion (CVPR 2024 · similarity 0.92)
- Efficient Dynamic Scene Editing via 4D Gaussian-based Static-Dynamic Separation (CVPR 2025 · similarity 0.91)
- Control4D: Efficient 4D Portrait Editing with Text (CVPR 2024 · similarity 0.89)
- CTRL-D: Controllable Dynamic 3D Scene Editing with Personalized 2D Diffusion (CVPR 2025 · similarity 0.89)
- Align Your Gaussians: Text-to-4D with Dynamic 3D Gaussians and Composed Diffusion Models (CVPR 2024 · similarity 0.89)
- Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models (NeurIPS 2024 · similarity 0.88)
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.
