ProjFlow: Projection Sampling with Flow Matching for Zero-Shot Exact Spatial Motion Control

2026-02-26

AI Summary

This work addresses the challenge of enforcing precise spatial constraints on generated human motion without requiring any training, while preserving naturalness. The authors propose a zero-shot sampling method that formulates animation as a linear inverse problem, integrating flow matching with projection-based sampling to exactly satisfy constraints. A key innovation is the introduction of a skeleton topology-aware motion metric, which ensures constraint corrections are distributed coherently across the entire kinematic chain. Additionally, a time-decaying pseudo-observation mechanism is designed to handle sparse input constraints effectively. Evaluated on motion completion and 2D-to-3D pose lifting tasks, the method not only strictly adheres to imposed constraints but also achieves motion naturalness comparable to or better than existing zero-shot approaches, rivaling the performance of trained controller-based methods.

Technology Category

• Computer Vision: Motion & Tracking
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Intelligent Robots: Motion and Path Planning

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Generating human motion with precise spatial control is a challenging problem. Existing approaches often require task-specific training or slow optimization, and enforcing hard constraints frequently disrupts motion naturalness. Building on the observation that many animation tasks can be formulated as a linear inverse problem, we introduce ProjFlow, a training-free sampler that achieves zero-shot, exact satisfaction of linear spatial constraints while preserving motion realism. Our key advance is a novel kinematics-aware metric that encodes skeletal topology. This metric allows the sampler to enforce hard constraints by distributing corrections coherently across the entire skeleton, avoiding the unnatural artifacts of naive projection. Furthermore, for sparse inputs, such as filling in long gaps between a few keyframes, we introduce a time-varying formulation using pseudo-observations that fade during sampling. Extensive experiments on representative applications, motion inpainting, and 2D-to-3D lifting, demonstrate that ProjFlow achieves exact constraint satisfaction and matches or improves realism over zero-shot baselines, while remaining competitive with training-based controllers.

Problem

• spatial motion control
• zero-shot generation
• linear constraints
• motion realism
• human motion synthesis

Innovation

• Projection Sampling
• Flow Matching
• Kinematics-aware Metric
• Zero-Shot Motion Control
• Linear Inverse Problem