StyleFields improves 3D shape reconstruction with controllable detail mixing
StyleFields: Multi-Scale AdaIN-Modulated Implicit SDFs for Coarse-to-Fine 3D Shape Reconstruction and Editing
Computer Vision and Pattern RecognitionGraphics
Summary
Reconstructing 3D shapes often mixes up big overall form and small surface details. The authors introduce StyleFields, a way to separate and recombine these by changing the 3D shape model at different depths, so broad structure and fine details can be controlled independently. This method doesn’t need special labels or tricky training, and it helps create new shapes by mixing parts from different objects. They also show it can help optimize car shapes to improve airflow by editing coarse or fine features separately.
What this means in practice
- •For 3d modelers: Create new 3D shapes by blending large-scale forms and fine details from different models with controllable influence.
- •For automotive designers: Edit car shapes at global or surface-detail levels using a learned drag predictor to optimize aerodynamic performance.
Authors
Ehsan Garaaghaji, Nicolas Talabot, Pascal Fua, Doruk Oner
Abstract
We introduce StyleFields, a DeepSDF-based architecture for high-fidelity 3D reconstruction that enables controllable geometric style mixing: the coarse structure of one object can be combined with the fine-scale details of another. The core idea is depth-aware modulation: instead of a single global code, we inject latents via multi-level Adaptive Instance Normalization at several decoder depths, and supervise matching auxiliary heads with a coarse-to-fine schedule while gradually growing network depth. This aligns early layers with global shape and later layers with high-frequency detail, achieving content-style decoupling without part labels or adversarial training. StyleFields delivers faithful reconstructions, convincing cross-instance hybrids, and consistent gains in ablations over injection depth and supervision granularity. We further demonstrate a practical application in automotive aerodynamics: a learned surrogate drag predictor serves as a differentiable objective to optimize reconstructed cars, allowing targeted edits of global form or surface details by freezing the complementary latent stream. StyleFields offers a simple, effective recipe for controllable implicit reconstruction and downstream performance-driven design.