wildflow-splat 0.1.0

Fast PLY point cloud processing for 3D Gaussian splatting workflows

wildflow-splat

Fast PLY point cloud processing with Python bindings. Convert photogrammetry outputs into spatial patches for 3D Gaussian splatting workflows.

Installation

pip install wildflow-splat

Quick Start

from wildflow import splat

# Load PLY file and create configuration
config = splat.Config("model.ply")
config.sample_percentage = 50.0  # Process 50% of points

# Define spatial patch
patch = splat.Patch("section_1.bin")
patch.min_x = -100.0
patch.max_x = 100.0
config.add_patch(patch)

# Process point cloud
results = splat.split_point_cloud(config)
print(f"Processed {results['total_points_written']} points")

Features

• High-performance Rust backend with Python bindings
• Multi-threaded processing for large datasets
• Spatial partitioning with configurable bounds
• COLMAP-compatible output for 3D reconstruction pipelines
• Progress bars with interrupt handling
• JSON configuration support

API Reference

Config

config = splat.Config("input.ply")
config.sample_percentage = 75.0  # 0-100%
config.min_z = -50.0            # Z-axis filtering
config.max_z = 10.0

Patch

patch = splat.Patch("output.bin")
patch.min_x = -200.0  # Spatial bounds
patch.max_x = 200.0
patch.min_y = -200.0
patch.max_y = 200.0

Processing

results = splat.split_point_cloud(config)
# Returns: {'points_loaded': int, 'total_points_written': int, 'patches_written': int}

Configuration Files

Load settings from JSON:

config = splat.Config.from_file("config.json")

Example config.json:

{
  "input_file": "model.ply",
  "sample_percentage": 100.0,
  "minZ": -10.0,
  "maxZ": 50.0,
  "patches": [
    {
      "output_file": "patch_1.bin",
      "minX": -100.0,
      "maxX": 100.0,
      "minY": -100.0,
      "maxY": 100.0
    }
  ]
}

Requirements

• Python 3.8+
• Rust (for building from source)

License

MIT License