Boosting Zero-Shot 3D Style Transfer with 2D Pre-trained Priors

• URL: http://arxiv.org/abs/2605.30065v1
• Date: Thu, 28 May 2026 15:14:23 GMT
• Title: Boosting Zero-Shot 3D Style Transfer with 2D Pre-trained Priors
• Abstract summary: We focus on zero-shot 3D style transfer that can generate multi-view consistent stylized views of the 3D scene given an arbitrary style image.<n>Our core idea is to integrate a decoder pre-trained on large-scale 2D image datasets into the 3D style transfer pipeline.
• Score: 35.47950709037777
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In this work, we focus on zero-shot 3D style transfer that can generate multi-view consistent stylized views of the 3D scene given an arbitrary style image. We primarily tackle the issue of data scarcity in 3D style transfer, which arises when each model is trained on only a single scene, thereby limiting the number of available content images. This scarcity significantly hampers stylization performance, as model optimization relies on a sufficient number of content-style image pairs to provide supervisory signals. Our core idea is to integrate a decoder pre-trained on large-scale 2D image datasets into the 3D style transfer pipeline, thereby leveraging the prior knowledge encoded in the decoder from learning over numerous content-style image pairs. Our method combines feature Gaussian splatting and deferred stylization, enabling high-quality stylization with the data-sufficient decoder network while ensuring view consistency by unifying view-dependent operations into a view-invariant process. Experiments demonstrate that our Data-Sufficient StyleGaussian (DS-StyleGaussian) model outperforms existing zero-shot 3D style transfer methods in terms of visual quality across various datasets. This work also suggests that 2D pre-training can serve as a strong enhancement for 3D tasks, bridging the data gap between 2D and 3D.

Related papers

• DiffStyle3D: Consistent 3D Gaussian Stylization via Attention Optimization [22.652699040654046]
  3D style transfer enables the creation of visually expressive 3D content.<n>We propose DiffStyle3D, a novel diffusion-based paradigm for 3DGS style transfer.<n>We show that DiffStyle3D outperforms state-of-the-art methods, achieving higher stylization quality and visual realism.
  arXiv  Detail & Related papers  (2026-01-27T15:41:11Z)
• SSGaussian: Semantic-Aware and Structure-Preserving 3D Style Transfer [57.723850794113055]
  We propose a novel 3D style transfer pipeline that integrates prior knowledge from pretrained 2D diffusion models.<n>Our pipeline consists of two key stages: First, we leverage diffusion priors to generate stylized renderings of key viewpoints.<n>The second is instance-level style transfer, which effectively leverages instance-level consistency across stylized key views and transfers it onto the 3D representation.
  arXiv  Detail & Related papers  (2025-09-04T16:40:44Z)
• CLIPGaussian: Universal and Multimodal Style Transfer Based on Gaussian Splatting [4.597952319060107]
  We introduce CLIPGaussian, the first unified style transfer framework that supports text- and image-guided stylization across multiple modalities.<n>Our method operates directly on Gaussian primitives and integrates into existing GS pipelines as a plug-in module.<n>We demonstrate superior style fidelity and consistency across all tasks, validating CLIPGaussian as a universal and efficient solution for multimodal style transfer.
  arXiv  Detail & Related papers  (2025-05-28T20:41:24Z)
• Towards Multi-View Consistent Style Transfer with One-Step Diffusion via Vision Conditioning [12.43848969320173]
  Stylized images from different viewpoints generated by our method achieve superior visual quality, with better structural integrity and less distortion. Our method effectively preserves the structural information and multi-view consistency in stylized images without any 3D information.
  arXiv  Detail & Related papers  (2024-11-15T12:02:07Z)
• StyleGaussian: Instant 3D Style Transfer with Gaussian Splatting [141.05924680451804]
  StyleGaussian is a novel 3D style transfer technique. It allows instant transfer of any image's style to a 3D scene at 10 frames per second (fps)
  arXiv  Detail & Related papers  (2024-03-12T16:44:52Z)
• DatasetNeRF: Efficient 3D-aware Data Factory with Generative Radiance Fields [68.94868475824575]
  This paper introduces a novel approach capable of generating infinite, high-quality 3D-consistent 2D annotations alongside 3D point cloud segmentations. We leverage the strong semantic prior within a 3D generative model to train a semantic decoder. Once trained, the decoder efficiently generalizes across the latent space, enabling the generation of infinite data.
  arXiv  Detail & Related papers  (2023-11-18T21:58:28Z)
• 3DStyle-Diffusion: Pursuing Fine-grained Text-driven 3D Stylization with 2D Diffusion Models [102.75875255071246]
  3D content creation via text-driven stylization has played a fundamental challenge to multimedia and graphics community. We propose a new 3DStyle-Diffusion model that triggers fine-grained stylization of 3D meshes with additional controllable appearance and geometric guidance from 2D Diffusion models.
  arXiv  Detail & Related papers  (2023-11-09T15:51:27Z)
• Guide3D: Create 3D Avatars from Text and Image Guidance [55.71306021041785]
  Guide3D is a text-and-image-guided generative model for 3D avatar generation based on diffusion models. Our framework produces topologically and structurally correct geometry and high-resolution textures.
  arXiv  Detail & Related papers  (2023-08-18T17:55:47Z)
• AgileGAN3D: Few-Shot 3D Portrait Stylization by Augmented Transfer Learning [80.67196184480754]
  We propose a novel framework emphAgileGAN3D that can produce 3D artistically appealing portraits with detailed geometry. New stylization can be obtained with just a few (around 20) unpaired 2D exemplars. Our pipeline demonstrates strong capability in turning user photos into a diverse range of 3D artistic portraits.
  arXiv  Detail & Related papers  (2023-03-24T23:04:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.