Track manipulation, locomotion and embodied systems.
Search papers, labs, and topics across Lattice.
This paper introduces DiDE, a novel training-free framework for disentangled 3D stylization, addressing the limitation of existing methods that jointly transfer color and texture. DiDE leverages the overcomplete latent space of rectified flow-based 3D generative models by partitioning channels to independently encode color and texture, enabling interference-free composition at each self-attention layer. Experimental results on the new Disen3D-Bench benchmark demonstrate DiDE's superior performance in color fidelity, texture transfer, and content preservation compared to prior art.
Independent control over color and texture in 3D stylization is now achievable, decoupling these attributes for unprecedented creative flexibility.
Recent advances in rectified flow-based image-to-3D generative models have enabled high-fidelity 3D asset generation. Building on this, a growing line of work has exploited these strong 3D priors for training-free stylization, transferring visual attributes from a reference image onto a generated 3D asset. However, existing methods enforce an all-or-nothing paradigm: color and texture are transferred jointly, with no mechanism to control them independently -- a limitation we formalize as Disentangled 3D Stylization(Disen3D). To address this, we propose DiDE, the first training-free framework for Disen3D. Key to our approach is the observation that the structured latent space of image-to-3D models is overcomplete with respect to texture: texture information occupies only a small subset of the style-significant channels, leaving a free subspace available for independent color encoding. DiDE exploits this via a channel partition mechanism that processes a content image, a texture reference, and a color reference through dedicated branches and composes both style signals interference-free at every self-attention layer, preserving content geometry throughout. Experiments on Disen3D-Bench, our newly collected multi-reference benchmark, show that DiDE consistently outperforms 2D and 3D stylization baselines in color fidelity, texture transfer, and content preservation.