Modern 3D Data Processing System
Professional Point Cloud & Mesh Processing | Big Data Support | Cross-Platform Solution
ACloudViewer is an open-source 3D point cloud and triangular mesh processing software library. It supports rapid development of software for processing 3D data, highly based on CloudCompare, Open3D, ParaView and COLMAP, and integrates the PCL library.
Originally designed to compare two 3D point clouds (such as those obtained by laser scanning) or the difference between point clouds and triangular meshes. It relies on an octree structure highly optimized for this specific use case, capable of handling massive point cloud data (typically over 10 million points, up to 120 million points with 2GB memory).
Powerful 3D data structures and processing algorithms, supporting point clouds, meshes and various geometries
COLMAP-based scene reconstruction system, supporting complete workflow from images to 3D models
High-precision point cloud registration algorithms, including ICP, RANSAC and other methods
High-performance rendering engine based on VTK and OpenGL, supporting PBR physical rendering
Integrated with PyTorch and TensorFlow, supporting 3D deep learning applications
GPU acceleration for core 3D operations, supporting CUDA 12.x
Provides C++ and Python dual-language API, flexible and easy to use
Real-time CUDA-accelerated 3DGS rendering, remote training viewer, and novel view synthesis via the SIBR plugin
Native GGUF inference for depth (DA3), face recognition, feature matching (LightGlue), and 3D Gaussian splats (FreeSplatter) — no Python runtime
Rich plugin ecosystem, supporting custom feature extensions
Depth estimation, feature matching, and 3D Gaussian splats — powered by libAICore.so and GGUF models, fully integrated into the GUI
Turn a single photo into a depth map and 3D point cloud. Multi-view depth, camera pose, COLMAP export, and automatic reconstruction integration — all from quantized GGUF weights (~142 MB Base).
Run SCRFD/YuNet and ArcFace/SFace GGUF packs for face detection, identity registry matching, verification, and live multi-face recognition in one plugin.
Match corresponding features between two images in under a second. OpenCV RootSIFT + GGUF LightGlue matcher on GPU — hundreds of mutual matches with live green-line visualization.
Reconstruct a 3D Gaussian scene from as few as two uncalibrated photos, or collect guided Face Capture views. Optional camera pose estimation, SIBR-compatible PLY export, and one-click qSIBR preview.
Enable with
-DAICore_ENABLED=ON
plus individual PLUGIN_STANDARD_Q* flags. See the build guide for Vulkan/Metal/CUDA options.
All current and past release downloads are available on GitHub releases.
Download the .whl file for your system and Python version from GitHub Releases
💡 Due to file size exceeding PyPI limits, manual download is required
pip install cloudviewer-*.whl
Example: pip install cloudviewer-3.9.3-cp310-cp310-win_amd64.whl
Supports Python 3.10-3.12 | Ubuntu 20.04+, macOS 10.15+, Windows 10+ (64-bit)
Download the corresponding .whl file from GitHub Releases, then install:
pip install cloudviewer-*.whl
💡 Due to large file size, direct PyPI installation is not supported
python -c "import cloudViewer as cv3d; print(cv3d.__version__)"
import cloudViewer as cv3d
# Create sphere mesh
mesh = cv3d.geometry.ccMesh.create_sphere()
mesh.compute_vertex_normals()
# Visualize
cv3d.visualization.draw(mesh, raw_mode=True)
git clone --recursive https://github.com/Asher-1/ACloudViewer.git
cd ACloudViewer
mkdir build && cd build
cmake ..
make -j$(nproc)
./bin/ACloudViewer
For detailed compilation instructions, please refer to BUILD.md
Select the installer for your system from the Download section
Double-click the desktop icon or launch ACloudViewer from the Start Menu
File → Open to select your point cloud or mesh file
Supported formats: PLY, PCD, LAS, LAZ, E57, OBJ, STL, FBX, etc.
Explore ACloudViewer's powerful applications in different fields
