Official repo for the paper "iLRM: An Iterative Large 3D Reconstruction Model"

Check out the wide branch for wide-coverage scene reconstruction on the DL3DV dataset!

# create conda environment
conda create -n ilrm python=3.10 -y
conda activate ilrm

# install PyTorch (adjust cuda version according to your system)
pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt

We first release the 2-view RealEstate10K (256x256) checkpoint, which is the most common baseline in related works and provides a standard reference point for comparison. We will upload other checkpoints soon!

In training and evaluation, we used the dataset preprocessed by pixelSplat.

The model checkpoints are host on HuggingFace.

Update the dataset.roots field in config/experiment/re10k.yaml with your dataset path.

Update the checkpointing.load field in config/main.yaml with the pretrained model.

CUDA_VISIBLE_DEVICES=0 python -m src.main +experiment=re10k dataset/view_sampler=evaluation dataset.view_sampler.index_path=assets/evaluation_index_re10k.json

@article{kang2025ilrm,
  title={iLRM: An Iterative Large 3D Reconstruction Model},
  author={Kang, Gyeongjin and Nam, Seungtae and Sun, Xiangyu and Khamis, Sameh and Mohamed, Abdelrahman and Park, Eunbyung},
  journal={arXiv preprint arXiv:2507.23277},
  year={2025}
}