### Install pixelSplat Dependencies

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Create a virtual environment and install project dependencies including PyTorch and other requirements.

```bash
python3.10 -m venv venv
source venv/bin/activate
# Install these first! Also, make sure you have python3.11-dev installed if using Ubuntu.
pip install wheel torch torchvision torchaudio
pip install -r requirements.txt
```

--------------------------------

### Install Custom Gaussian Rasterization

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Install a modified version of diff-gaussian-rasterization, pointing to CUDA 12.1 libraries.

```bash
LD_LIBRARY_PATH=/usr/local/cuda-12.1/lib64 pip install git+https://github.com/dcharatan/diff-gaussian-rasterization-modified
```

--------------------------------

### Train PixelSplat Model

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Use this command to start training the PixelSplat model with a specific experiment configuration. Ensure you have a GPU with sufficient VRAM.

```bash
python3 -m src.main +experiment=re10k
```

--------------------------------

### Install PyTorch with CUDA 11.8

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Install a version of PyTorch built with CUDA 11.8 if your system does not use CUDA 12.1 by default.

```bash
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
```

--------------------------------

### Run Ablation Study

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Execute ablation studies by using specific experiment configurations. This example shows how to ablate the epipolar transformer.

```bash
python3 -m src.main +experiment=re10k_ablation_no_epipolar_transformer
```

--------------------------------

### Evaluate PixelSplat Model (Real Estate 10k)

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Render frames from an existing checkpoint for the Real Estate 10k dataset. This command specifies the experiment, mode, view sampler, and checkpoint to load.

```bash
# Real Estate 10k
python3 -m src.main +experiment=re10k mode=test dataset/view_sampler=evaluation dataset.view_sampler.index_path=assets/evaluation_index_re10k.json checkpointing.load=checkpoints/re10k.ckpt
```

--------------------------------

### Evaluate PixelSplat Model (ACID)

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Render frames from an existing checkpoint for the ACID dataset. This command specifies the experiment, mode, view sampler, and checkpoint to load.

```bash
# ACID
python3 -m src.main +experiment=acid mode=test dataset/view_sampler=evaluation dataset.view_sampler.index_path=assets/evaluation_index_acid.json checkpointing.load=checkpoints/acid.ckpt
```

--------------------------------

### PixelSplat BibTeX Entry

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

The BibTeX entry for citing the PixelSplat paper in academic publications.

```bibtex
@inproceedings{charatan23pixelsplat,
      title={pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction},
      author={David Charatan and Sizhe Li and Andrea Tagliasacchi and Vincent Sitzmann},
      year={2024},
      booktitle={CVPR},
}
```

--------------------------------

### Reduce Memory Usage During Training

Source: https://github.com/dcharatan/pixelsplat/blob/main/README.md

Adjust the batch size to reduce memory consumption during training. This command sets the per-GPU batch size to 1.

```bash
python3 -m src.main +experiment=re10k data_loader.train.batch_size=1
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.