CVPR 2026 Submission · Supplementary Project Page

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

ProjFlow is a training-free projection sampler for flow-matching motion priors. It casts spatial motion control tasks as linear inverse problems and enforces linear spatial constraints exactly while preserving natural motion. This page collects qualitative examples and comparisons to state-of-the-art methods.

Spatial Control Exact Constraints

ProjFlow satisfies the constraints of any spatial motion-control task that can be formulated as a linear inverse problem.

Kinematics-Aware Geometry Kinematics-Aware Metric

A kinematics-aware metric spreads corrections coherently across the body during flow-based sampling.

Zero-Shot No fine-tuning

Control is applied to a pretrained text-to-motion prior without retraining or inner-loop optimization.

Representative Control Scenarios

All four applications shown below can be formulated as linear inverse problems: trajectory following, 2D-to-3D lifting, relative position constraints, and looped motion, all solved by ProjFlow within a single zero-shot framework.

(a)

Joint Trajectory

Precisely follow a specified joint trajectory.

(b)

2D to 3D Lift

Lift 2D keyposes and trajectories into 3D motion.

(c)

Relative Constraint

Maintain a fixed relative position between joints.

(d)

Looping Motion

Generate seamless loops by matching start and end poses.

Trajectory Control Benchmarks

We compare spatial accuracy for single-joint and multi-joint trajectory control. Each card fixes one joint or a set of joints to a target trajectory and shows OmniControl, MaskControl, and ProjFlow under the same control signal.

Methods: OmniControl, MaskControl, ProjFlow (ours), shown left to right in each triplet. Control: For each joint, all methods receive the same target trajectory and the videos visualize how well the motion follows the spatial constraints.
Anchor

Pelvis Tracking

Text prompt: "a person runs forward in an S path"

OmniControl
MaskControl
ProjFlow
Anchor

Head Tracking

Text prompt: "A person jumps and kicks a football in the air with their head"

OmniControl
MaskControl
ProjFlow
Anchor

Left Hand Tracking

Text prompt: "the person is boxing with their left hand and throws multiple punches."

OmniControl
MaskControl
ProjFlow
Anchor

Right Hand Tracking

Text prompt: "a person puts hands on the armrest"

OmniControl
MaskControl
ProjFlow
Anchor

Left Foot Tracking

Text prompt: "a person stands with both feet on the ground, kicks once with his left foot."

OmniControl
MaskControl
ProjFlow
Anchor

Right Foot Tracking

Text prompt: "a person side steps left and right."

OmniControl
MaskControl
ProjFlow
Anchors

Multi-Joint Tracking

Hands and Pelvis Circular-Arc Trajectories

Text prompt: "a person crosses their arms for chest fly"

OmniControl
MaskControl
ProjFlow
Right Hand and Pelvis S-Shaped Trajectories

Text prompt: "a person snakes forward while the right hand waves up and down"

OmniControl
MaskControl
ProjFlow

2D to 3D Reconstruction

For the 2D-to-3D reconstruction task, the input is initial 2D keypoints and their trajectories under a known orthographic camera. We compare Sketch2Anim with ProjFlow, which enforces the projection constraints exactly while preserving natural 3D motion.

Sketch2Anim
ProjFlow