Content & Design

SC-GS provides code for Sparse-Controlled Gaussian Splatting, designed for editable dynamic scenes. This open-source tool allows users to effortlessly edit and customize their digital assets through interactive features. It represents motion using sparse control points, which drive 3D Gaussians for high-fidelity rendering. The approach supports both dynamic view synthesis and motion editing, making it versatile for various applications. Recent updates include support for editing static Gaussians from .ply files, improved handling of real-world static objects, and video rendering with interpolation of editing results. It offers two ARAP deformation strategies for motion editing: iterative deformation and deformation from Laplacian initialization, giving users flexibility in achieving desired effects.

WildGS-SLAM is an open-source research tool designed for monocular Gaussian Splatting SLAM in dynamic environments. Developed for Computer Vision and Pattern Recognition (CVPR) 2025, it excels at accurately tracking camera trajectories and reconstructing 3D Gaussian maps for static elements from monocular video sequences, even when captured in the wild with dynamic distractors. The tool effectively removes all dynamic components to provide a clear static reconstruction. It supports various datasets including Wild-SLAM Mocap, Wild-SLAM iPhone, Bonn Dynamic, and TUM RGB-D, and also allows users to integrate their own custom datasets. WildGS-SLAM provides functionalities for camera pose evaluation and novel view synthesis, making it a valuable resource for researchers in the field.

nerfmm is an open-source implementation of Neural Radiance Fields (NeRF) designed to reconstruct 3D scenes and render novel views even when camera parameters are unknown. This tool jointly estimates camera poses, focal lengths, and the NeRF model, offering a robust solution for 3D reconstruction. It supports various datasets, including the LLFF dataset and a custom Blender Forward Facing (BLEFF) dataset, which is specifically designed for evaluating camera parameter estimation accuracy and image rendering quality under varying pose perturbations. nerfmm provides scripts for training from scratch, refining pre-trained models, and evaluating image rendering quality, novel view synthesis, and 3D pose visualization. It is particularly useful for researchers and developers in computer vision working on advanced 3D reconstruction and neural rendering techniques.

LimitPixel is a free AI-powered tool specifically designed for generating 3D animations. It focuses on creating lifelike animations, making it suitable for a range of uses. The platform allows users to produce 3D animations for different applications, from prototyping to final production. It serves as a valuable resource for 3D artists, animators, and game developers who need to efficiently create or prototype animated content.

Psmoveapi is a versatile, cross-platform library designed for 6DoF (six degrees of freedom) tracking of the PlayStation Move Motion Controller. It integrates advanced sensor fusion and computer vision techniques to provide precise positional and rotational tracking. The library also extends its functionality to include ambient display control through the PS Move's LED orb, enhancing user feedback and immersion. Developers can utilize psmoveapi to gain direct PC access to the PS Move controller, facilitating communication via both Bluetooth and USB connections. This makes it an ideal tool for creating custom applications, games, or research projects that leverage the unique input capabilities of the PS Move controller on various computing platforms.

sdfstudio is a unified, open-source framework designed for neural implicit surface reconstruction, leveraging the foundation of the Nerfstudio project. It provides a modular architecture that allows for the implementation and exploration of different surface reconstruction methods, such as UniSurf, VolSDF, and NeuS. The framework supports various scene representations and datasets, making it a versatile tool for advanced 3D modeling, research, and development in the field of neural implicit surfaces. Its open-source nature encourages community contributions and provides a flexible platform for experimenting with cutting-edge 3D reconstruction techniques.

taichi_3d_gaussian_splatting is an unofficial, open-source implementation of 3D Gaussian Splatting, designed for real-time radiance field rendering. This tool utilizes the Taichi programming language, known for its high-performance computing capabilities, to process and render complex 3D scenes efficiently. It takes multiple-view images, a sparse point cloud, and camera pose as input to train and optimize the point cloud representation. This allows for the creation of highly detailed and realistic 3D environments that can be rendered in real-time, making it suitable for applications requiring interactive 3D visualization or rapid scene generation.

Polycam is a versatile AI tool for 3D scanning and modeling, enabling users to capture reality and create digital twins across iPhone, iPad, Android, and web platforms. It leverages spatial AI for detailed documentation, measurement, and design, supporting LiDAR and photogrammetry for high-quality 3D models. Key functionalities include generating 2D floor plans, processing drone footage into 3D models, and capturing detailed metrics for various professional applications. Polycam is trusted by professionals in architecture, engineering, construction, forensics, product design, and media for its ability to accelerate workflows and provide accurate spatial data.

Board Ad is an AI-powered creative studio designed to generate marketing assets, including 3D product displays, model photoshoots, and infographics. The tool aims to streamline the creation of visual content for businesses, potentially eliminating the need for traditional design agencies. It is described as utilizing an intuitive interface powered by AI technology to produce these assets. However, the live website is currently redirecting, preventing access to detailed information regarding its features, pricing, or specific capabilities.

OpenPose is a powerful, real-time multi-person keypoint detection library developed by the CMU Perceptual Computing Lab. It is capable of estimating 2D and 3D keypoints for the human body, face, hands, and feet, offering a total of 135 keypoints. A key differentiator is its runtime, which remains constant regardless of the number of detected people for body/foot estimation, unlike other libraries whose runtime grows linearly. It supports various input sources like images, videos, webcams, and specialized cameras, and can output keypoints in multiple formats including JSON and XML. OpenPose offers both C++ and Python APIs for custom functionality and is compatible with Ubuntu, Windows, and Mac OSX, with hardware support for CUDA, OpenCL, and CPU-only versions.

I2L-MeshNet_RELEASE is the official PyTorch implementation of the I2L-MeshNet architecture, designed for accurate 3D human pose and mesh estimation from a single RGB image. This open-source project, recognized at ECCV 2020, won first and second place in the 3DPW challenge for part orientation and joint position metrics. The tool provides both lixel-based 1D heatmap predictions for mesh vertices and regressed SMPL parameters. It supports training and testing across various datasets like Human3.6M, MuCo, MSCOCO, 3DPW, and FreiHAND, making it a robust solution for researchers and developers in computer vision focused on human body reconstruction.

LangSplat is the official implementation of the paper "LangSplat: 3D Language Gaussian Splatting" (CVPR 2024 Highlight), a cutting-edge tool for generating 3D models with integrated language features. It offers a PyTorch-based optimizer to create LangSplat models from SfM datasets, a scene-wise language autoencoder to manage memory demands, and scripts to convert images into optimization-ready SfM data. The project also provides preprocessed datasets like 3D-OVS and expanded LERF datasets with COLMAP data, along with pre-trained models. LangSplat has seen significant performance improvements with LangSplat V2, achieving over 450+ FPS in rendering, and is expanding into 4D language fields with 4D LangSplat. It is ideal for researchers and developers working on advanced 3D reconstruction and language-driven scene generation.

Godot 3d Trucks offers an interactive truck-themed game experience directly within your web browser. This application, hosted on Hugging Face Spaces, leverages WebGL technology to deliver 3D graphics and gameplay. Users can launch the game and play without any installation, provided their browser supports WebGL. The game loads with a progress bar, indicating its readiness for play. It serves as a demonstration of the capabilities of the Godot Engine in creating engaging 3D environments and interactive experiences, making it accessible to anyone with a compatible web browser.

Godot 3d Voxel is an interactive web application hosted on Hugging Face Spaces, enabling users to play a voxel game without any downloads or installations. This tool provides a direct browser-based experience for exploring and interacting with 3D voxel environments, making it highly accessible. It's an excellent demonstration of the capabilities of the Godot Engine in a web context, suitable for both casual exploration and educational purposes for those interested in game development or interactive 3D experiences. The application is running and readily available for immediate use.

Interior Render AI is an AI-driven interior design tool that transforms indoor photos or design sketches into high-quality, cinematic interior design renderings. Users can upload an image, choose from a library of styles like Scandinavian or Japanese, or describe their desired aesthetic using natural language to generate photorealistic redesigns in seconds. The platform supports virtual staging for empty rooms, allowing real estate agents and homeowners to visualize spaces with realistic furniture and decor. It offers a free trial and a pay-as-you-go credit system, eliminating the need for subscriptions. This tool is ideal for concept exploration, client presentations, and creating professional-grade interior visualizations without requiring 3D software.

ID Animator is an AI tool designed for creating animations, enabling users to generate animated avatars and various other forms of animated content. While the tool aims to provide capabilities for animation generation, the current live website indicates a persistent runtime error, preventing its functionality. This issue suggests that the application is not operational at this time, impacting its ability to deliver on its intended purpose of animation creation. The platform is hosted on Hugging Face Spaces, a common environment for sharing and deploying machine learning applications.

NoPoSplat is an open-source research project that leverages 3D Gaussian Splats to generate high-quality novel views and perform accurate pose estimation from sparse, unposed images. Developed based on PyTorch, it allows users to predict 3D Gaussians in a canonical space, addressing challenges in 3D reconstruction where camera poses are unknown. The tool provides pre-trained checkpoints for various datasets like RealEstate10K and ACID, and includes detailed instructions for installation, training, and evaluation of both novel view synthesis and pose estimation. It's ideal for researchers and developers working on advanced computer vision tasks.

InVesalius is a free and open-source software designed for 3D medical imaging reconstruction, capable of generating detailed models from 2D DICOM files obtained through CT or MRI equipment. It boasts multi-platform compatibility, running on GNU Linux, Windows, and MacOS, and is internationalized with support for numerous languages including English, Portuguese, French, and German. The software provides a comprehensive suite of tools for medical professionals, encompassing DICOM support, image manipulation (zoom, pan, brightness/contrast), various segmentation methods (2D slices, watershed, pre-defined thresholds), and precise measurement tools. Users can create and export 3D surfaces in formats like STL, PLY, OBJ, and VRML, alongside high-quality volume rendering projections and picture exportation options.

SpacetimeGaussians is an open-source implementation of Spacetime Gaussian Feature Splatting, a technique for real-time dynamic view synthesis. Developed by OPPO US Research Center, this tool is designed for researchers and developers working on advanced 3D scene reconstruction and rendering. It supports various datasets including Neural 3D, Technicolor, and Google Immersive, offering both 'lite' and 'full' models for different computational needs. The project includes comprehensive installation guides for Linux and Windows (via WSL2), detailed instructions for data preprocessing, training, and testing. A real-time viewer, based on SIBR and Gaussian Splatting, is also provided, along with pre-trained models to facilitate immediate use and experimentation. The codebase is modular, allowing users to customize Gaussian representations, rendering pipelines, and even integrate new datasets.

torch-ngp offers a PyTorch CUDA extension implementation of instant-ngp, supporting both Signed Distance Functions (SDF) and Neural Radiance Fields (NeRF). It includes a graphical user interface (GUI) for training and visualization. The repository also features PyTorch implementations of TensoRF, adapted to instant-ngp's NeRF framework, and CCNeRF for compressible-composable NeRF via rank-residual decomposition. Additionally, it provides an implementation of D-NeRF for dynamic scenes and experimental features like text-guided NeRF editing. The tool supports various datasets and offers options for different backbones and optimization techniques, making it a versatile platform for neural graphics research.

Stable Zero123 is an AI tool designed for generating 3D models directly from 2D images. This capability streamlines the process of creating three-dimensional representations, making it particularly useful for rapid prototyping and research applications. By leveraging a single image, users can quickly transform 2D concepts into 3D assets, which can be beneficial in various fields requiring quick visualization or model generation. The tool is available as a Hugging Face Space, indicating its accessibility and potential for community-driven development and use.

HDM is an AI tool hosted on Hugging Face that specializes in the template-free reconstruction of human-object interaction. It leverages procedural interaction generation to achieve its results, making it a valuable resource for researchers and developers in the field of computer vision and human-computer interaction. The tool is designed to facilitate advanced studies and applications related to how humans interact with objects, offering a flexible and accessible platform for experimentation and development. Its availability as a free template on Hugging Face further enhances its utility for academic and research purposes.

posenet-python is an open-source project offering a pure Python implementation of Google's TensorFlow.js PoseNet model, designed for real-time human pose estimation. This port focuses on multi-pose detection and includes significant performance optimizations over a literal translation of the original JavaScript code, achieving throughputs of 90-110fps on a GTX 1080+ with its 'fast' post-processing. It supports various MobileNet models and provides demo applications for image processing, performance benchmarking, and webcam integration. Developers can easily install it within a Python 3.x environment with TensorFlow, making it accessible for integrating real-time pose estimation into their projects.

UnboundedNeRFPytorch is an open-source project providing a PyTorch implementation for generating large-scale Neural Radiance Fields (NeRFs). This tool is designed to facilitate the creation of advanced, efficient, and unbounded NeRFs, making it suitable for researchers and developers working on 3D reconstruction and rendering. Its primary purpose is to serve as a robust platform for benchmarking various large-scale radiance field algorithms, allowing for comparative analysis and performance evaluation. The implementation focuses on simplicity and speed, aiming to provide a straightforward yet powerful solution for complex 3D scene representation. As an open-source project, it encourages community contributions and offers a flexible environment for experimentation and development in the field of neural rendering.