论文HF Daily Papers · arXiv 2610.4075424 天前
DyCraft系统:分步生成4D场景解决具身动画物理违规
本文由 AI 根据公开内容整理,可能有误,请以原文为准:DyCraft: An Agentic System for Long-Horizon 4D Scene Generation(HF Daily Papers · arXiv 2610.40754)
核心内容
前沿编码智能体在生成连贯长时序4D动画时,常因缺乏物理约束导致物体穿模、支撑缺失及运动不自然,且难以通过单次脚本修复复杂时空错误。为此研究团队提出DyCraft系统,借鉴人类动画师工作流,将长时序动画生成拆解为编码智能体可可靠求解的分步任务,并在涵盖人形机器人、人物及物理机构的24项多镜头动画任务中进行验证。
DyCraft先将用户提示词扩展为详细计划并构建静态场景,再生成由关键姿态组成的3D故事板,使智能体能独立构建与调试每帧静态3D场景。通过将问题分解为经校验的起止状态对,智能体利用运动创作工具编写局部动作。评测显示,相比Codex中的原生GPT-6智能体及现有基线,DyCraft大幅减少物理违规,生成的4D场景在人工评分中获得更高偏好度。
关键看点
- 3D故事板机制:将长时序动画拆解为独立静态帧进行构建与调试,解耦复杂时空约束以支持局部动作精准编辑
- 物理合规性提升:针对物体穿模、缺乏支撑及非自然运动等违规问题,通过分步验证机制显著降低生成错误率
- 人形机器人适配:在24项长时序多镜头任务中验证,覆盖人形机器人、生物角色及物理机构等多类具身主体
- 优于GPT-6基线:对比Codex原生GPT-6智能体,DyCraft生成的4D场景物理一致性更强,人工评估偏好度更高
工程与产业观察
DyCraft验证了将具身仿真数据生成从端到端黑盒转向结构化分步验证的工程价值。其3D故事板机制有效解决了大模型在长时序物理一致性上的短板,为构建高质量、低违规率的具身智能训练数据集提供了可扩展的合成管线范式,降低了Sim-to-Real迁移中的数据清洗成本。
英文摘要 (Abstract)
While frontier coding agents can generate high-quality 3D scenes and articulated assets, they struggle to create coherent long-horizon animations. When equipped with general-purpose harnesses, they frequently produce physically implausible outputs, generating objects that interpenetrate, lack physical support, or move unnaturally. Such bare agents struggle to repair these errors because they default to scripting the full animation in one pass, making it difficult to verify their work and disentangle complex spatiotemporal constraints. Drawing on the practice of human animators, we introduce DyCraft, a system for animated 3D scene generation that breaks animation into steps that coding agents can more reliably solve. DyCraft first expands the user's prompt into a detailed plan and builds a static scene. It then creates a 3D storyboard of key animation poses, where each frame is a static 3D scene that the agent can build and debug in isolation. Storyboarding decomposes the problem into pairs of verified start and end states, allowing the agent to author local motion between them with a collection of motion authoring tools. We evaluate DyCraft on 24 long-horizon, multi-shot animation tasks covering humanoid robots, human and animal characters, and physical mechanisms. Compared with a native GPT-6 agent in Codex and prior agentic baselines, DyCraft substantially reduces physical violations and generates higher-quality animated 3D scenes that are more often preferred by human raters.
Hugging Face Daily Papers 热度:5 upvotes 作者:Jeff Tan, Hang Dai, Zeqian Long, Yanzhe Lyu, R. Kenny Jones, Shangzhe Wu, Chen Geng, Jiajun Wu
外部来源
本页事实对应的公开页面,以原文为准。(点击将离开本站)