Edit4D: Editable 4D Scene Generation with Textual Control
We introduce Edit4D, an editable text-to-4D scene generation framework that constructs 4D scenes from separate background and dynamic character representations.
ProofPaper ↗
Key points
- Recent advances in generative models have enabled text-driven dynamic scene synthesis, but editable 4D scene generation remains challenging due to the need to preserve spatial geometry and temporal dynamics.
- Using these prompts, we synthesize multi-view scene images with text-to-image and camera-conditioned generative priors and reconstruct a background 3D Gaussian splatting representation.
- We then generate a character image, animate it with an image-conditioned video prior, reconstruct a 4D mesh, and bind Gaussian primitives to obtain a character representation.
- By integrating stage-specific components, Edit4D enables controllable camera motion, flexible object-level editing, and the practical construction of editable 4D scenes.
Sources (1)
- [1]Edit4D: Editable 4D Scene Generation with Textual ControlarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 15, 10:52 AM
We introduce Edit4D, an editable text-to-4D scene generation framework that constructs 4D scenes from separate background and dynamic character representations.
Recent advances in generative models have enabled text-driven dynamic scene synthesis, but editable 4D scene generation remains challenging due to the need to preserve spatial geometry and temporal dynamics.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 15, 2026RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent
- Oct 15, 2026VLA-$Δ$: Learning Physical State Transition for Vision-Language-Action Models
- Oct 15, 2026AVC-Reward: Event-Centric Audio-Video Consistency Evaluation
- Oct 15, 2026SUPRA-TEX: 3D Generation with Super-Aligned Texture
- Oct 14, 2026ReaDy: Recurrent States as Dynamic Context for Speculative Drafting in Hybrid Linear Attention Large Language Models
- Oct 14, 2026VASPO: Self-Distillation with Privileged Value Estimation