SpaceControl
Introducing Test-Time Spatial Control to 3D Generative Modeling

Elisabetta Fedele^1,2*, Francis Engelmann^2* Ian Huang², Or Litany^3,4, Marc Pollefeys¹, Leonidas Guibas^2
1ETH Zurich, ²Stanford University, ³Technion, ⁴NVIDIA

Code | Paper | Project Page

Generative methods for 3D assets have recently achieved remarkable progress, yet providing intuitive and precise control over the object geometry remains a key challenge. Existing approaches predominantly rely on text or image prompts, which often fall short in geometric specificity: language can be ambiguous, and images are cumbersome to edit. In this work, we introduce SPACECONTROL, a training-free test-time method for explicit spatial control of 3D generation. Our approach accepts a wide range of geometric inputs, from coarse primitives to detailed meshes, and integrates seamlessly with modern pre-trained generative models without requiring any additional training. A controllable parameter lets users trade off between geometric fidelity and output realism. Extensive quantitative evaluation and user studies demonstrate that SPACECONTROL outperforms both training-based and optimization-based baselines in geometric faithfulness while preserving high visual quality. Finally, we present an interactive user interface that enables online editing of superquadrics for direct conversion into textured 3D assets, facilitating practical deployment in creative workflows.

Check out our Project Page for more videos and interactive demos!

📦 Installation

1. Clone the repository:

git clone [email protected]:spacecontrol3d/spacecontrol.git
cd spacecontrol

2. Setup the environment: The code has been test with CUDA 12.8 (see nvcc --version) on an NVIDIA 3090 with torch 2.8.0+cu128

conda create -n spacecontrol python=3.10 -y
conda activate spacecontrol

# instructions for your setup: https://pytorch.org/get-started/locally/
pip install torch==2.8.0 torchvision==0.23.0 torchaudio==2.8.0 --index-url https://download.pytorch.org/whl/cu128

pip install pillow imageio imageio-ffmpeg tqdm easydict opencv-python-headless scipy ninja rembg onnxruntime trimesh open3d xatlas pyvista pymeshfix igraph transformers psutil viser tensorboard pandas lpips
pip install git+https://github.com/EasternJournalist/utils3d.git@9a4eb15e4021b67b12c460c7057d642626897ec8

pip install xformers==0.0.32.post1 --index-url https://download.pytorch.org/whl/cu128
pip install flash-attn --no-build-isolation
pip install kaolin -f https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-2.8.0_cu128.html
pip install spconv-cu120

mkdir -p /tmp/extensions
git clone https://github.com/NVlabs/nvdiffrast.git /tmp/extensions/nvdiffrast
pip install /tmp/extensions/nvdiffrast --no-build-isolation

git clone --recurse-submodules https://github.com/JeffreyXiang/diffoctreerast.git /tmp/extensions/diffoctreerast
pip install /tmp/extensions/diffoctreerast --no-build-isolation

git clone https://github.com/autonomousvision/mip-splatting.git /tmp/extensions/mip-splatting
pip install /tmp/extensions/mip-splatting/submodules/diff-gaussian-rasterization/ --no-build-isolation

cp -r extensions/vox2seq /tmp/extensions/vox2seq
pip install /tmp/extensions/vox2seq --no-build-isolation

💡 Usage

To start the web-based interactive demo:

python gui/gui_text_image.py

🙏 Acknowledgments

We thank the authors of TRELLIS for their excellent work and for making their code publicly available. We also gratefully acknowledge NVIDIA for their academic compute grant, which enabled the development of this method; these contributions were instrumental to the project.

📜 Citation

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

@article{fedele2025spacecontrol,
  title   = {{SpaceControl: Introducing Test-Time Spatial Control to 3D Generative Modeling}},
  author  = {Elisabetta Fedele, Francis Engelmann, Ian Huang, Or Litany, Marc Pollefeys, Leonidas Guibas},
  journal = {arXiv preprint arXiv:2512.05343},
  year    = {2025}
}