---
id: pypcd4
version: "1.4.3"
license: BSD 3-Clause License Copyright (c) 2018, Daniel Maturana Copyright (c) 2023-present, MAP IV, Inc. All rights reserved. Redistribution and use in source and binary forms, with or without modification,… (full text in the JSON record)
license_treatment: permissive
maintenance: active
---

# pypcd4 — Read and write PCL .pcd files in python

License: permissive · Maintenance: active · Downloads: 85.4K/mo

## What it is and what it does

pypcd4 is a modern Python library for reading, writing, and manipulating Point Cloud Data (PCD) files. It provides a PointCloud class that wraps point data and supports conversion to and from NumPy arrays, filtering by slicing or boolean masks, concatenation of multiple clouds, and integration with ROS PointCloud2 messages. The library depends on numpy for array operations, pydantic for data validation, and python-neo-lzf for compression support.

Typical workflows include loading PCD files from disk, extracting specific fields or subsets of points, transforming the data via NumPy, and saving results back to PCD format. It is designed for robotics and 3D perception applications where point cloud data is common, and offers convenience methods for common operations like field selection and point filtering without requiring manual NumPy indexing.

Use it for:

- Load a PCD file, filter points by spatial coordinates, and save the result for downstream processing.
- Convert between PCD and NumPy array formats to integrate point cloud data with machine learning pipelines.
- Merge multiple PCD files into a single point cloud using concatenation for multi-sensor fusion.
- Convert ROS PointCloud2 messages to PCD files for offline analysis or archival.
- Extract specific fields (e.g., x, y, z, intensity) from a point cloud for visualization or analysis.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

pypcd4 reads, writes, and manipulates Point Cloud Data (PCD) files, converting between PCD format, NumPy arrays, and ROS PointCloud2 messages.

Yes. pypcd4 is actively maintained, has no known vulnerabilities, installs with low friction, and is licensed permissively. It fills a clear need for point cloud I/O in Python with modern syntax and ROS integration. Suitable for robotics, 3D perception, and scientific computing workflows.

## Install

pip install pypcd4
uv add pypcd4
poetry add pypcd4

## Installing pypcd4

Before you install: Low install friction; pure Python wheel with three runtime dependencies (numpy, pydantic, python-neo-lzf). Active maintenance with recent commits and stable production status.

License in practice: BSD 3-Clause permissive license allows commercial and private use with attribution; no viral copyleft restrictions.

Quickstart:

pip install pypcd4

from pypcd4 import PointCloud
import numpy as np

pc = PointCloud.from_path("point_cloud.pcd")
array = pc.numpy()
pc_filtered = pc[pc.pc_data["x"] > 0.5]
pc_filtered.save("filtered.pcd")

Verify before relying:

- Whether ROS integration requires ROS to be installed and sourced, or if rosbags library alone suffices for reading/writing
- Performance characteristics when concatenating or filtering large point clouds (thousands or millions of points)

## Package facts

- License: BSD 3-Clause License Copyright (c) 2018, Daniel Maturana Copyright (c) 2023-present, MAP IV, Inc. All rights reserved. Redistribution and use in source and binary forms, with or without modification,… (full text in the JSON record) (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 85.4K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

point cloud data pcd files, read write pcd format, pointcloud numpy conversion, ros pointcloud2 message, 3d point cloud processing, pcd file manipulation, point cloud filtering, point-cloud-processing, ros-integration, 3d-data

[View on SkillFed](https://skillfed.io/packages/pypcd4) · [View on PyPI](https://pypi.org/project/pypcd4/)