Interp3D: Correspondence-Aware Interpolation for Generative Textured 3D Morphing

Xiaolu Liu^1,2, Yicong Li^2*, Qiyuan He², Jiayin Zhu², Wei Ji³, Angela Yao², Jianke Zhu^1,4,5*

¹Zhejiang University    ²National University of Singapore    ³Nanjing University

⁴State Key Lab of CAD & CG, Zhejiang University    ⁵Shenzhen Loop Area Institute

^*Corresponding Authors

In this work, we propose Interp3D, a training-free approach that instantiates the progressive alignment principle based on generative priors for textured 3D morphing. 

Clone the repo:

git clone --recurse-submodules https://github.com/xialul2/Interp3D.git
cd Interp3D

Following the command in TRELLIS for the environment installation.

You can run the morphing process with the following command:

python example_interp.py --exp_id ./web_cases

To evaluate the effectiveness of our methods, we present Inter3DData, the benchmark dataset for assessing the generative texture 3D Morphing . You can download the dataset from Google Drive.

If you find this work helpful, please consider citing our paper:

@article{liu2026interp3d,
  title={Interp3D: Correspondence-Aware Interpolation for Generative Textured 3D Morphing},
  author={Liu, Xiaolu and Li, Yicong and He, Qiyuan and Zhu, Jiayin and Ji, Wei and Yao, Angela and Zhu, Jianke},
  journal={arXiv preprint arXiv:2601.14103},
  year={2026}
}