Choreographing a World of Dynamic Objects

TL;DR

This paper introduces CHORD, a universal generative pipeline that synthesizes diverse 4D dynamic scenes by distilling rich motion information from 2D videos, overcoming limitations of traditional rule-based and data-intensive learning methods.

Contribution

We propose a category-agnostic, distillation-based approach to generate 4D scene dynamics from 2D videos, enabling scalable and versatile scene synthesis.

Findings

Successfully generates diverse multi-body 4D dynamics.

Outperforms existing methods in scene generation quality.

Demonstrates applicability in robotics manipulation policies.

Abstract

Dynamic objects in our physical 4D (3D + time) world are constantly evolving, deforming, and interacting with other objects, leading to diverse 4D scene dynamics. In this paper, we present a universal generative pipeline, CHORD, for CHOReographing Dynamic objects and scenes and synthesizing this type of phenomena. Traditional rule-based graphics pipelines to create these dynamics are based on category-specific heuristics, yet are labor-intensive and not scalable. Recent learning-based methods typically demand large-scale datasets, which may not cover all object categories in interest. Our approach instead inherits the universality from the video generative models by proposing a distillation-based pipeline to extract the rich Lagrangian motion information hidden in the Eulerian representations of 2D videos. Our method is universal, versatile, and category-agnostic. We demonstrate its effectiveness by conducting experiments to generate a diverse range of multi-body 4D dynamics, show its advantage compared to existing methods, and demonstrate its applicability in generating robotics manipulation policies. Project page: https://yanzhelyu.github.io/chord