Content & Design

CityDreamer is an innovative AI tool hosted on Hugging Face Spaces, designed for generating immersive 3D city models. Users can easily create vast and detailed urban environments by manipulating simple camera controls such as distance, height, and angle, along with setting the map center. This intuitive interface allows for the rapid generation of unique cityscapes without complex modeling skills. While currently in a sleeping state due to inactivity, its core functionality promises a straightforward approach to 3D environment creation, making it suitable for various applications from conceptual design to virtual world building. The tool's ability to produce endless 3D cities positions it as a valuable asset for quick prototyping and visualization.

talking-head-anime-2-demo provides demo programs for the "Talking Head Anime from a Single Image 2: More Expressive" project. It features a manual poser for manipulating facial expressions and head rotation of anime characters via a graphical user interface or Jupyter notebook. Additionally, an iFacialMocap puppeteer allows users to transfer their own facial motion, captured by an iOS device, to an anime character image. The tool requires a powerful Nvidia GPU and specific software environments, including Python and PyTorch. It's designed for those interested in AI-driven animation and character manipulation, offering a hands-on approach to exploring expressive anime head movements.

Animated Drawings by Meta is an open-source project that provides an algorithm and tools for animating children's drawings of human figures. Users can upload their own drawn characters, and the system automatically detects, segments, and rigs them for animation using BVH motion data. The tool supports exporting animations as MP4 videos or transparent GIFs. It offers flexibility through configuration files for controlling characters, motions, and scenes, and even allows for animating multiple characters or adding background images. While primarily designed for human-like figures, it also supports custom skeletons. The project is available on GitHub and provides options for local setup or Docker container deployment.

4DGaussians is a research project presented at CVPR 2024, focusing on 4D Gaussian Splatting for real-time dynamic scene rendering. This method allows for very quick convergence and achieves real-time rendering speeds, as demonstrated on D-NeRF and HyperNeRF datasets. The project provides code for environmental setup, data preparation for synthetic and real dynamic scenes (D-NeRF, HyperNeRF, DyNeRF, and multiple views), training, rendering, and evaluation. It also includes helpful scripts for exporting 3D Gaussians, visualizing weights, and merging 4D Gaussians, making it a comprehensive resource for researchers in computer vision and graphics.

BlenderProc is a powerful open-source tool designed to create photorealistic synthetic training images using a procedural Blender pipeline. It's ideal for generating large datasets for computer vision models, offering extensive features for loading diverse object formats like .obj, .ply, .blend, and BOP datasets. Users can procedurally set object poses, apply physics for collision checking, and manipulate materials and lighting. The tool supports rendering various image types including RGB, stereo, depth, normal, and segmentation images, and can write results to .hdf5 containers with COCO & BOP annotations. It provides comprehensive documentation, tutorials, and examples to help users get started with synthetic data generation.

L4GM, or Large 4D Gaussian Reconstruction Model, is an AI tool designed to convert standard video inputs into advanced 3D Gaussian-reconstructed videos. This process significantly enhances the depth and realism of the original footage, offering a more immersive visual experience. While the Hugging Face Space for L4GM is currently paused, its core functionality aims to provide users with a powerful method for generating sophisticated 3D models from existing video content. This technology is particularly beneficial for those looking to elevate their video projects with cutting-edge 3D reconstruction techniques.

MMSkeleton is an open-source toolbox developed by OpenMMLAB, specifically designed for skeleton-based human understanding. It offers a highly extensible framework that systematically organizes code and projects, allowing for adaptation to various tasks and scaling to complex deep models. Key functionalities include 2D and 3D pose estimation, skeleton-based action recognition (like ST-GCN), and action synthesis. The toolbox also supports building custom skeleton-based datasets and creating personalized applications. It is part of the OpenMMLAB project, developed on the ST-GCN research project, and is released under the Apache 2.0 license.

InstantSplat is an official implementation of a sparse-view framework for large-scale 3D scene reconstruction using Gaussian Splatting. Developed by NVlabs, this tool supports 3D-GS, 2D-GS, and Mip-Splatting, allowing users to quickly generate detailed 3D environments from limited input. It is designed to be SfM-free, meaning it does not require Structure-from-Motion for camera pose estimation, significantly speeding up the reconstruction process. The repository provides comprehensive instructions for installation, including conda environment setup and Docker image usage, making it accessible for developers and researchers. InstantSplat is ideal for applications requiring rapid 3D model generation and free-view rendering, leveraging advanced techniques for efficient processing.

Street Gaussians is an open-source project presented at ECCV 2024, focusing on modeling dynamic urban scenes using Gaussian Splatting. This tool provides a framework for researchers and developers to reconstruct complex, moving urban environments from video data. It includes functionalities for data preparation, such as converting Waymo Open Dataset, generating LiDAR depth, and creating sky masks. Users can configure parameters based on 3D Gaussian Splatting, train models, render scenes, and visualize results. The project offers scripts for training and rendering on example and experimental Waymo scenes, making it a valuable resource for advancing research in dynamic 3D scene reconstruction.

SuperSplat is a free and open-source 3D Gaussian Splat Editor built on web technologies, allowing it to run directly in the browser without any downloads or installations. This tool enables users to inspect, edit, optimize, and publish 3D Gaussian Splats, making it accessible for various 3D modeling tasks. It supports local development with Node.js 18+ and offers localization capabilities, with currently supported languages available for translation. The project is actively maintained by an open-source community, providing a robust platform for working with 3D Gaussian Splats. A live version of the editor is available online, offering immediate access to its features.

Cascadeur is a standalone software designed for 3D keyframe animation, specifically catering to humanoids and other characters. It significantly simplifies the animation process through its AI-assisted tools, allowing users to quickly create key poses and instantly visualize physical results. The software also enables easy adjustment of secondary motion while maintaining full user control. Cascadeur supports managing scenes with multiple characters and offers various edit modes to suit different stages of the animation workflow. It's suitable for creating animations from scratch or editing existing ones, making it a versatile tool for animators and game developers.

SAM3D Body with Rerun is an AI tool designed for 3D body reconstruction, providing capabilities to visualize and analyze human bodies in three dimensions. This tool is particularly valuable for researchers and developers involved in AI model testing, offering a platform to interact with 3D body data. Hosted on Hugging Face, it aims to facilitate advancements in areas requiring detailed human body analysis. While the current live website indicates a runtime error, suggesting it's not fully operational, its intended purpose is to serve as a resource for those working with 3D human body models.

Gaussian-SLAM is an open-source project available on GitHub, designed for photo-realistic dense Simultaneous Localization and Mapping (SLAM). It leverages Gaussian splatting to achieve high-quality 3D reconstruction, offering a robust solution for researchers and engineers in computer vision and robotics. The tool supports various datasets including Replica, TUM_RGBD, ScanNet, and ScanNet++, and provides scripts for easy setup and data downloading. Users can configure and run SLAM experiments, reproduce results, and even generate fly-through videos based on reconstructed scenes. It's tested on powerful GPUs like RTX3090 and RTX A6000, ensuring performance for demanding tasks.

Zero-1-to-3 Live Demo is an AI tool designed for generating 3D models directly from single images. Hosted on Hugging Face Spaces by cvlab, it offers a live demonstration interface for users to explore the capabilities of this technology. While the intention is to provide an interactive experience for creating 3D representations, the current status indicates a build error, preventing the application from functioning as intended. The tool is associated with the Columbia University Computer Vision Lab and is licensed under the MIT License, suggesting an open and accessible approach to its development and use.

Find3D is an open-world 3D part segmentation model designed to identify and segment specific components within 3D objects. Users can upload their own .pcd files or select from provided samples to analyze point cloud data. The tool allows for precise part queries, enabling the segmentation of complex 3D objects into their constituent parts. This capability is particularly useful for applications requiring detailed structural analysis, object recognition, and component isolation within 3D environments. Developed as a Hugging Face Space, Find3D offers an accessible platform for researchers, developers, and enthusiasts working with 3D data and AI applications.

Open-DiffusionGS is an open-source project that implements a novel approach to single-stage image-to-3D generation and reconstruction by integrating Gaussian Splatting directly into a diffusion denoiser. This method allows for fast and scalable creation of 3D objects, including mesh exportation, and efficient scene reconstruction without the need for depth estimators. The tool is capable of generating 3D outputs in approximately 6 seconds, significantly faster than some state-of-the-art methods. It supports both object-centric image-to-3D generation and scene-level reconstruction, with evaluation capabilities for the latter using datasets like RealEstate10K. The project provides comprehensive scripts for environment setup, quick demonstrations, data preparation for both scene and object-level datasets (including G-Objaverse), evaluation, and multi-stage training of custom models.

CamPlan AI is a cutting-edge application designed for creating modern 2D and 3D floor plans, immersive interior design blueprints, and detailed room layouts. Leveraging AR scanning technology, LiDAR, and an AI Interior Design Assistant, it allows users to design and visualize spaces rapidly. Users can scan their home or house with an iPhone or iPad to generate detailed floor plans in 2D, 3D, and AR formats, enabling blueprint creation, space measurement, and CAD export. The AI Interior Design Assistant provides expert advice on layout, furniture, color schemes, lighting, textures, and materials, helping users create professional designs. CamPlan is optimized for iOS, iPadOS, and macOS, offering various viewing modes and export options.

SFA3D is an open-source PyTorch implementation designed for super fast and accurate 3D object detection using LiDAR point clouds. It features an anchor-free approach, eliminating the need for Non-Max-Suppression, which contributes to its speed. The tool supports distributed data parallel training, making it suitable for large-scale applications, and includes pre-trained models for immediate use. SFA3D is particularly relevant for autonomous driving and robotics, as highlighted by its use in the Udacity Self-Driving Car Engineer Nanodegree Program. It also offers ROS source code integration for robotics applications and provides detailed technical documentation and demonstration capabilities.

AI 3D Model Generator is a user-friendly tool hosted on Hugging Face Spaces that allows users to transform 2D images into 3D models. By simply uploading a photograph, the application generates a 3D mesh that can be viewed directly within the browser. The output is provided in widely compatible formats, specifically OBJ and GLB files, making them ready for integration into various applications such as games, augmented reality (AR), virtual reality (VR) experiences, or even for 3D printing. This tool emphasizes speed and accuracy in its image-to-3D conversion process, catering to individuals and professionals who require quick and efficient 3D model creation from existing imagery.

X&Immersion presents itself as a private website, with content indicating capabilities such as building websites, selling products, and writing blogs. However, all listed pages, including the homepage, pricing, plans, features, FAQ, and documentation, display a "Private Site" message. Users are prompted to log in to WordPress.com to request access, suggesting that the tool or service is not publicly available or is in a restricted development phase. Due to the private nature of the site, specific AI tools, services, or features related to video game studios, non-player characters (NPCs), or game design automation, as mentioned in the previous description, cannot be verified from the live content.

EmerNeRF offers a self-supervised approach for spatial-temporal scene decomposition using neural fields. It can effectively separate dynamic objects from a static background and estimate their motion without explicit supervision. The tool also enriches 2D features by lifting and 'denoising' them in 4D space-time, opening new possibilities for advanced scene understanding. EmerNeRF supports the NeRF On-The-Road (NOTR) dataset, derived from the Waymo Open Dataset, and NuScenes, with provisions for custom dataset integration. It is implemented in PyTorch and designed for researchers and developers working on neural radiance fields and 3D scene reconstruction.

Webots is an open-source robot simulator designed to provide a comprehensive development environment for modeling, programming, and simulating a wide range of robotic systems, including robots, vehicles, and other mechanical systems. Originally developed at EPFL for mobile robotics research, it was later commercialized by Cyberbotics and open-sourced in 2018. The platform is beginner-friendly, making it an excellent tool for introducing newcomers to the field of robotics. It offers pre-compiled binaries for easy installation and detailed tutorials to guide users through the simulation process. Webots supports continuous integration, nightly tests, and provides resources for building from source, updating, and reporting bugs, fostering an active development community.

3DGS.cpp is a high-performance, cross-platform implementation of Gaussian Splatting, leveraging the Vulkan API and compute pipelines for efficient rendering. This tool aims to democratize access to advanced point-based radiance fields, which are often limited to specific hardware or platforms. By utilizing Vulkan, 3DGS.cpp ensures broad compatibility across Windows, Linux, macOS, iOS, and visionOS, including support for Apple platforms where OpenGL is deprecated. Its compute capabilities are designed to be comparable to CUDA, with support for warp-level primitives (subgroups), making it a powerful alternative for developers and researchers. The project encourages contributions and offers a clear path for integrating new Gaussian Splatting variants, making it a valuable resource for expanding research reach.

awesome-human-pose-estimation is an open-source GitHub repository serving as a comprehensive collection of resources focused on human pose-related problems. It primarily concentrates on human pose estimation but also covers areas such as mesh representation, flow calculation, (inverse) kinematics, affordance, robotics, and sequence learning. The repository is continuously updated with the latest papers and resources, making it a valuable asset for researchers and students in the field. It provides an organized list of academic papers, categorized by topics like 2D and 3D pose estimation, human mesh, and real-time pose estimation, along with links to popular implementations in PyTorch, TensorFlow, and Torch.