ply2splat 0.2.0

A high-performance library for converting Gaussian Splatting PLY files to SPLAT format

ply2splat

A Rust crate and CLI tool for converting Gaussian Splatting .ply files to the .splat format.

Features

• High Performance: Utilizes parallel processing (via rayon) for conversion and sorting.
• Fast I/O: Uses zero-copy serialization and large buffers for maximum throughput.
• Correctness: Implements the standard conversion logic including Spherical Harmonics (SH) to color conversion and geometric transformations.
• Python Bindings: Use the library directly from Python via PyO3.

Installation

CLI (From Source)

git clone https://github.com/bastikohn/ply2splat.git
cd ply2splat
cargo build --release

The binary will be available at target/release/ply2splat.

Python Package

Install from source using maturin:

pip install maturin
git clone https://github.com/bastikohn/ply2splat.git
cd ply2splat
maturin develop --release

Or build a wheel:

maturin build --release
pip install target/wheels/ply2splat-*.whl

Usage

CLI

ply2splat --input input.ply --output output.splat

Python

import ply2splat

# Convert a PLY file to SPLAT format
count = ply2splat.convert("input.ply", "output.splat")
print(f"Converted {count} splats")

# Convert without sorting (faster, but may affect rendering quality)
count = ply2splat.convert("input.ply", "output.splat", sort=False)

# Load PLY file and access individual splats
data = ply2splat.load_ply_file("input.ply")
print(f"Loaded {len(data)} splats")

for splat in data:
    print(f"Position: {splat.position}")
    print(f"Scale: {splat.scale}")
    print(f"Color (RGBA): {splat.color}")
    print(f"Rotation: {splat.rotation}")

# Access splats by index
first_splat = data[0]
last_splat = data[-1]

# Load existing SPLAT file
data = ply2splat.load_splat_file("output.splat")
print(f"Loaded {len(data)} splats from SPLAT file")

# Get raw bytes for custom processing
raw_bytes = data.to_bytes()

# Load and convert to bytes (for in-memory processing)
data, count = ply2splat.load_and_convert("input.ply")
print(f"Loaded {count} splats, {len(data)} bytes")

Development

Requirements

• Rust (latest stable)
• Nix (optional, for reproducible environment)

Running Tests

cargo test

Fuzzing

The crate includes fuzzing targets to ensure stability against malformed inputs.

# Install cargo-fuzz
cargo install cargo-fuzz

# Run fuzzing target
cargo fuzz run fuzz_conversion

Development Environment

This project supports both Nix and Devcontainers for a reproducible development environment.

• Nix: nix develop will enter a shell with Rust and dependencies configured.
• Devcontainer: Open the folder in VS Code and accept the prompt to reopen in container.

License

MIT