CARVIEW

MOTORHOMES

Select Language

HTTP/2 200 server: GitHub.com content-type: text/html; charset=utf-8 last-modified: Thu, 23 Oct 2025 23:19:27 GMT access-control-allow-origin: * strict-transport-security: max-age=31556952 etag: W/"68fab7ff-cb56" expires: Sun, 28 Dec 2025 08:31:46 GMT cache-control: max-age=600 content-encoding: gzip x-proxy-cache: MISS x-github-request-id: C0D9:15317B:769C13:84E3D5:6950E898 accept-ranges: bytes age: 0 date: Sun, 28 Dec 2025 08:21:46 GMT via: 1.1 varnish x-served-by: cache-bom-vanm7210084-BOM x-cache: MISS x-cache-hits: 0 x-timer: S1766910106.985369,VS0,VE213 vary: Accept-Encoding x-fastly-request-id: 4f9d9602e87db14a3de681a33b392b7849587cbc content-length: 8874 Art3D: Training-Free 3D Generation from Flat-Colored Illustration

Art3D: Training-Free 3D Generation from Flat-Colored Illustration

Xiaoyan Cong ^{* †}

Jiayi Shen ^{* †}

Zekun Li ^*

Rao Fu ^*

Tao Lu ^*

Srinath Sridhar ^*

CVPR 2025 AI4CC Oral (Best Presentation)

^*Brown Univeristy, ^†Equal Contribution

Paper Presentation Code (Coming Soon) arXiv Data Gallery

Abstract

Large-scale pre-trained image-to-3D generative models have exhibited remarkable capabilities in diverse shape generations. However, most of them struggle to synthesize plausible 3D assets when the reference image is flat-colored like hand drawings due to the lack of 3D illusion, which are often the most user-friendly input modalities in art content creation. To this end, we propose Art3D, a training-free method that can lift flat-colored 2D designs into 3D. By leveraging structural and semantic features with pre- trained 2D image generation models and a VLM-based realism evaluation, Art3D successfully enhances the three-dimensional illusion in reference images, thus simplifying the process of generating 3D from 2D, and proves adaptable to a wide range of painting styles. To benchmark the generalization performance of existing image-to-3D models on flat-colored images without 3D feeling, we collect a new dataset, Flat-2D, with over 100 samples. Experimental results demonstrate the performance and robustness of Art3D, exhibiting superior generalizable capacity and promising practical applicability.

Pipeline

Diagram of the transformer deep learning architecture. — Art3D adds 3D illusion to the flat-colored image through ControlNet based on the structure features, e.g., canny edge or depth map. The mesh is then synthesized based on the proxy image and textured by baking information from the input.

Qualitative Results

We combine our augmentation module with InstantMesh [Xu 2024] and perform comparisons with state-of-the-art pre-trained image-to-3D methods Shap-E [Jun 2023], LN3Diff [Lan 2024], InstantMesh [Xu 2024], 3DTopia-XL [Chen 2024], LGM [Tang 2024] and Trellis [Xiang 2024] All the flat-shaped input images are from our curated dataset Flat-2D. Our augmentation module can significantly improve the geometry quality of generated 3D assets.

Visual Comparisons

Interactive 3D Mesh Gallery

Award

BibTeX citation

@misc{cong2025art3dtrainingfree3dgeneration,
    title={Art3D: Training-Free 3D Generation from Flat-Colored Illustration}, 
    author={Xiaoyan Cong and Jiayi Shen and Zekun Li and Rao Fu and Tao Lu and Srinath Sridhar},
    year={2025},
    eprint={2504.10466},
    archivePrefix={arXiv},
    primaryClass={cs.CV},
    url={https://arxiv.org/abs/2504.10466}, 
}

Original Source | Taken Source