license: other
license_name: stabilityai-ai-community
license_link: LICENSE.md
pipeline_tag: image-to-3d
tags:
- image-to-3d
- checkpoint
inference: false
ReLi3D
ReLi3D is a multi-view image-to-3D reconstruction model that takes object images and camera poses and generates a textured, UV-unwrapped, relightable 3D mesh asset.
It is the official model for the paper ReLi3D: Relightable Multi-view 3D Reconstruction with Disentangled Illumination.
Please note: For individuals or organizations generating annual revenue of US $1,000,000 (or local currency equivalent) or more, regardless of the source of that revenue, you must obtain an enterprise commercial license directly from Stability AI before commercially using ReLi3D, or any derivative work of ReLi3D or its outputs, such as fine-tuned models. You may submit a request for an Enterprise License at https://stability.ai/enterprise. Please refer to Stability AI's Community License, available at https://stability.ai/license, for more information.
Model Description
- Developed by: Stability AI
- Authors: Jan-Niklas Dihlmann, Mark Boss, Simon Donne, Andreas Engelhardt, Hendrik P. A. Lensch, Varun Jampani.
- Model type: Transformer multi-view image-to-3D model
- Model details: ReLi3D is trained to reconstruct a relightable 3D mesh from multiple 512x512 object images with known camera poses. The model outputs UV-unwrapped geometry and texture, and predicts material properties such as roughness and metallic values, together with an estimated illumination representation for downstream rendering workflows.
License
- Community License: Free for research, non-commercial, and commercial use by organizations and individuals generating annual revenue of US $1,000,000 (or local currency equivalent) or less, regardless of the source of that revenue. If your annual revenue exceeds US $1M, any commercial use of this model or derivative works thereof requires obtaining an Enterprise License directly from Stability AI. You may submit a request for an Enterprise License at https://stability.ai/enterprise. Please refer to Stability AI's Community License, available at https://stability.ai/license, for more information.
Model Sources
- Repository: https://github.com/Stability-AI/ReLi3D
- Project page: https://reli3d.jdihlmann.com/
- arXiv page: https://arxiv.org/abs/2603.19753
Files
- config.yaml: ReLi3D inference config
- reli3d_final.ckpt: ReLi3D model checkpoint
Training Dataset
The training process uses renders from Objaverse and curated subsets of additional sources with license review and filtering for training suitability.
Usage
For installation and full usage instructions, please refer to the ReLi3D GitHub repository.
Quickstart
To run inference on a set of images and camera poses, you can use the following command from the repository:
python demos/reli3d/infer_from_transforms.py \
--input-root demo_files/objects \
--objects Camera_01 \
--output-root outputs \
--num-views 4 \
--texture-size 256 \
--overwrite
Intended Uses
Intended uses include the following:
- Generation of artworks and use in design and other artistic processes.
- Applications in educational or creative tools.
- Research on reconstruction models, including understanding model limitations.
All uses of the model should be in accordance with our Acceptable Use Policy.
Out-of-Scope Uses
The model was not trained to be factual or true representations of people or events. As such, using the model to generate such content is out-of-scope of the abilities of this model.
Safety
As part of our safety-by-design and responsible AI deployment approach, we implement safety measures throughout the development of our models, from the time we begin pre-training a model to the ongoing development, fine-tuning, and deployment of each model. We have implemented a number of safety mitigations that are intended to reduce the risk of severe harms. However, we recommend that developers conduct their own testing and apply additional mitigations based on their specific use cases. For more about our approach to Safety, please visit our Safety page.
Contact
Please report any issues with the model or contact us:
- Safety issues: safety@stability.ai
- Security issues: security@stability.ai
- Privacy issues: privacy@stability.ai
- License and general: https://stability.ai/license
- Enterprise license: https://stability.ai/enterprise
Citation
@inproceeding{ dihlmann2026reli3d,
author = {Dihlmann, Jan-Niklas and Boss, Mark and Donne, Simon and Engelhardt, Andreas and Lensch, Hendrik P. A. and Jampani, Varun},
title = {ReLi3D: Relightable Multi-view 3D Reconstruction with Disentangled Illumination},
booktitle = {International Conference on Learning Representations (ICLR)},
year ={2026}
}