computer vision

TanGO: Training-Free 3D Editing via Tangent-Space Guidance and Optimization

arXiv:2607.14927

summary

TanGO is a training-free framework that edits 3D content from flow‑matching generative models by applying adaptive per‑token control in the tangent space of the model's dynamics, reducing structural artifacts and improving transformation quality.

Abstract

While recent flow-matching 3D generative models (e.g., VecSet) adopt structured representations, their tokens share global context, causing conventional training-free editing to suffer from semantic artifacts such as collapsed preserved regions or incomplete transformations. To address this, we propose TanGO, a training-free framework that enables adaptive per-token steering in the tangent space of generative dynamics. To realize this selective control, we formulate a one-step optimal control rule and determine the strength of each token's control signal using a von Mises-Fisher inspired directional discrepancy derived from the source and target velocity fields. Experiments show that TanGO substantially reduces structural artifacts and achieves state-of-the-art performance, outperforming existing 3D editing baselines. The code is publicly available at https://github.com/siw00-lim/TanGO.

ECCV 2026

Topics & keywords

#3d editing#generative models#tangent space#training-free methods#optimal controltangent-space guidancevon Mises-Fisher discrepancyper-token steeringflow-matchingVecSet
TanGO: Training-Free 3D Editing via Tangent-Space Guidance and Optimization · wovepaper