Content & Design

drl_grasping is an open-source project focused on advancing robotic manipulation through deep reinforcement learning. It enables robots to acquire robust grasping policies for diverse objects using compact 3D observations in the form of octrees. The project emphasizes sim-to-real transfer, allowing policies trained in simulation to be evaluated on real robots with zero-shot transfer. It includes multiple RL environments for robotic manipulation, supporting continuous actions in Cartesian space and various observation variants like RGB images, depth maps, and octrees. The framework is compatible with Gym API and has been tested with end-to-end model-free actor-critic algorithms like TD3, SAC, and TQC, with a setup for model-based algorithms also provided.

fake-bpy-module is a collection of fake Blender Python API modules specifically designed to enable robust code completion in popular Integrated Development Environments (IDEs). This tool significantly assists developers working with Blender's Python API by offering accurate suggestions and documentation, streamlining the development of Blender add-ons and scripting tasks. It supports a wide range of Blender versions, from 2.78 up to the latest daily builds, and can be installed via pip, pre-generated modules, or by manual generation. The project emphasizes long-term support and provides resources for bug reporting, feature requests, and community discussions, making it an essential utility for Blender Python developers.

Relic is an award-winning spatial entertainment platform designed for creating and deploying interactive spatial experiences. It simplifies the content creation process by allowing users to generate 3D assets from 2D images/videos, animate, and rig them using simple prompts, eliminating the need for coding or complex game engines like Unity or Unreal Engine. The platform integrates various 3D AI tools into an agentic interface, making it accessible for creators of all skill levels. Content can be easily deployed on the Relic App, embedded via an SDK, or shared as a WebXR Link. Relic also enables users to watch 4D AR, 6DoF, and fully Spatial Movies, bringing characters to life and transforming viewing experiences into immersive worlds.

pointnet.pytorch offers a PyTorch implementation of the PointNet deep learning model, specifically designed for 3D classification and segmentation using point sets. This open-source tool facilitates research and development in 3D data processing, providing a robust and tested framework compatible with PyTorch 1.0. It includes functionalities for downloading and preparing datasets, training classification and segmentation models, and visualizing results. The repository details performance metrics on datasets like ModelNet40 and ShapeNet, allowing users to compare against original implementations. It's a valuable resource for developers and researchers working with 3D point cloud data.

ActionMesh is a generative AI model developed by Facebook that creates animated 3D meshes from video inputs. Users can upload a short video or a sequence of images, and the application will process them to produce an animated 3D mesh. This mesh can be viewed directly within the browser, offering an immediate preview of the generated asset. The tool provides the output as a GLB file, a standard format for 3D models, which can then be downloaded and utilized in various 3D applications and environments. It leverages a feed-forward approach to predict production-ready 3D meshes and adapts 3D diffusion techniques to incorporate a temporal axis, enabling the generation of synchronized latents for realistic animations.

LucidDreamer is an official implementation of a novel text-to-3D generation framework designed to create high-fidelity 3D models. It leverages Interval Score Matching (ISM) to address the common over-smoothing effect found in other Score Distillation Sampling (SDS) based methods, ensuring more consistent and higher-quality updating directions for 3D models. The framework incorporates 3D Gaussian Splatting into its pipeline, leading to superior texture and shape distillation from pretrained 2D diffusion models. Accepted as a highlight at CVPR 2024, LucidDreamer aims to significantly advance the quality and training efficiency of text-to-3D content creation, enabling users to generate detailed 3D assets from textual descriptions in approximately 35 minutes on an A100 GPU.

JustLearn is an online platform dedicated to empowering global communication skills through comprehensive knowledge articles tailored for immersive language learning. The platform offers a wide range of articles covering various languages, including Arabic, Chinese, English, French, German, Italian, Japanese, Portuguese, Russian, Spanish, and Swedish. Users can explore top articles for language learning, such as tips for full immersion, common mistakes to avoid, and essential phrases for travelers. JustLearn aims to support learners at all stages of their linguistic adventure, providing valuable insights and practical advice to strengthen their learning journey. The platform also offers daily learning tips delivered directly to the user's inbox, ensuring continuous engagement and improvement.

GameSmith AI is an AI-powered tool designed for generating and animating 2D sprites, making it a valuable asset for game developers and hobbyists. Users can create custom sprites by providing text descriptions and optional reference images, offering flexibility in design. The tool also enables animation of these sprites with various motions, streamlining the asset creation process. Once generated and animated, users can download sprite sheets for direct integration into their games. Hosted on Hugging Face, GameSmith AI leverages AI, specifically Gemini, to facilitate the creation of game-ready 2D assets.

SMPLer-X is an open-source AI tool designed for scaling up expressive human pose and shape estimation, as featured in NeurIPS 2023. It offers official code for comprehensive human pose and shape estimation, including functionalities for training, testing, and inference. The tool supports various pretrained models with different backbones and parameters, such as ViT-S, ViT-B, ViT-L, and ViT-H, optimized for performance across multiple benchmarks. SMPLer-X also provides Docker support for inference, simplifying deployment. An updated version, SMPLest-X, extends its capabilities with stronger foundation models, further enhancing its utility for researchers and developers in the field of 3D human reconstruction.

GeometryCrafter is an AI tool developed by TencentARC that processes uploaded videos to estimate consistent 3D geometry for each frame. It provides users with interactive 3D point-cloud models for multiple frames, a disparity video, and a downloadable file containing the generated 3D data. This tool is particularly useful for converting open-world videos into structured 3D representations, leveraging advanced diffusion priors. Hosted on Hugging Face Spaces, GeometryCrafter offers a streamlined approach to video-to-3D conversion, making complex 3D reconstruction more accessible for various applications.

ComfyUI-3D-Pack is an extensive node suite designed to enhance ComfyUI's capabilities for 3D processing. It allows users to work with 3D inputs such as Mesh and UV Textures, leveraging advanced algorithms like 3DGS and NeRF, and integrating models like InstantMesh, CRM, and TripoSR. The tool aims to streamline and simplify 3D asset generation within the ComfyUI environment. Key features include single image to 3D mesh generation, multi-view image processing, and 3D visualization. It supports various pipelines for generating 3D models with textures, normal maps, and even part segmentation, catering to a wide range of 3D content creation needs.

Multi-View 3D Visual Grounding is an AI application hosted on Hugging Face, designed to facilitate participation in visual grounding competitions. Users can explore detailed competition information, access relevant datasets, and monitor their progress through leaderboards. The platform also enables users to manage their submissions efficiently, providing a centralized hub for engaging with visual grounding challenges. This tool is part of the broader Hugging Face ecosystem, which offers various AI models, datasets, and spaces for machine learning development and deployment.

StableAnimator is an open-source, end-to-end ID-preserving video diffusion framework designed for high-quality human image animation. It synthesizes videos directly from a reference image and a sequence of poses, eliminating the need for post-processing tools like face-swapping or restoration. The framework incorporates a global content-aware Face Encoder and a novel distribution-aware ID Adapter to ensure identity consistency. During inference, it utilizes a Hamilton-Jacobi-Bellman (HJB) equation-based optimization to further enhance face quality. StableAnimator supports resolutions like 576x1024 or 512x512 and provides tools for human skeleton and face mask extraction, making it a comprehensive solution for pose-driven human image animation.

AI Figure Lab is an AI-powered platform designed for generating a wide variety of figures, characters, and packaging mockups. Users can create AI action figures, dolls (including Barbie-style), custom characters (fantasy, sci-fi, anime), avatars, and even viral starter packs. The tool simplifies the creation process, requiring no special technical or artistic skills; users simply describe their desired figure or choose from pre-made styles. It leverages advanced AI models like GPT-4o to produce high-quality results. Beyond figure generation, it also offers options to add packaging, customize details, and share creations, making it a comprehensive solution for creators looking to bring their imaginative figures to life.

3D-Adapter is an AI-powered tool designed to generate high-quality 3D models. Users can create detailed 3D visuals by providing descriptions or various inputs, making it accessible for a range of creative and technical applications. The tool leverages advanced AI to interpret user specifications and produce geometry-consistent multi-view diffusion, which is crucial for realistic 3D generation. Based on research papers, 3D-Adapter offers a demo to showcase its capabilities, making it a valuable resource for 3D artists, researchers, and anyone looking to efficiently produce 3D content without extensive manual modeling. It is available as a free-to-use application.

WD138 is an online platform dedicated to providing access to a wide selection of popular slot games. It ensures a secure and stable gaming environment for its users, featuring an easy-to-navigate interface suitable for both new and experienced players. The platform offers continuously updated links to ensure uninterrupted access to games across various devices, including desktop and mobile. WD138 emphasizes its reliability as an official online site, maintaining consistent performance for a comfortable gaming experience. Players can find a diverse range of slot games and benefit from active daily links to popular slot titles.

Diffusion Point Cloud is an AI tool designed for generating 3D point clouds, leveraging a probabilistic generative model. This innovative approach is inspired by non-equilibrium thermodynamics, allowing the tool to exploit the reverse diffusion process to effectively learn and reproduce complex point distributions. While the tool's specific applications are broad, its core functionality lies in creating detailed 3D representations from data. Currently hosted on Hugging Face, the space is paused, indicating a potential for future development or a need for user engagement to reactivate it. Its underlying technology suggests a focus on advanced 3D modeling and data generation tasks.

DetailGen3D is a Hugging Face Space application developed by VAST-AI that specializes in enhancing 3D models with realistic surface details. Users can upload a front-view photograph and a basic GLB mesh, and the tool will automatically generate intricate details that match the provided image. The application offers customizable settings such as seed and detail strength, allowing for fine-tuned control over the final output. This generative AI tool is designed to streamline the process of adding realism to 3D assets, making it valuable for creators looking to quickly refine their models without extensive manual texturing or sculpting.

Leia Inc.'s Immersity platform leverages proprietary Spatial AI and Switchable-Display Hardware to convert everyday 2D content into immersive 3D experiences. Designed for phones, tablets, laptops, and more, Immersity allows users to experience movies, images, and social media with a powerful sense of presence, as if they are part of the scene. The platform offers a professional 2D to 3D conversion service with multiple pricing tiers for creators and businesses, including options for images and videos up to 4K resolution. Immersity aims to unlock new immersive experiences without requiring new hardware, redefining how content is consumed on existing devices.

GaussianCity is an AI-powered tool hosted on Hugging Face Spaces that enables users to generate 3D city models with remarkable efficiency. This application provides an intuitive interface where users can manipulate four key sliders to control the camera's distance, height, and angle, as well as the map's center point. By adjusting these parameters, users can quickly create diverse perspectives of a sprawling city. The system processes these adjustments to render a detailed 3D city environment in a matter of seconds, making it ideal for rapid prototyping and visualization tasks. Its focus on speed and ease of use makes complex 3D city generation accessible.

SoTA-Point-Cloud is a GitHub repository offering an extensive survey of deep learning techniques applied to 3D point clouds. Published in IEEE TPAMI 2020, this resource covers major tasks such as 3D shape classification, 3D object detection, and 3D point cloud segmentation. It provides comparative results on numerous publicly available datasets, including ModelNet, KITTI, and Semantic3D. The repository also offers insightful observations and outlines future research directions, making it an invaluable resource for researchers and practitioners in the field of 3D computer vision. The maintainers regularly update the page with new results and suggestions.

Wonder3D is an advanced AI tool designed for 3D generation, capable of reconstructing highly-detailed textured meshes from a single input image in approximately 2 to 3 minutes. It leverages a cross-domain diffusion model to first generate consistent multi-view normal maps and corresponding color images. Following this, a novel normal fusion method is employed to achieve fast and high-quality 3D reconstruction. The tool provides training codes for users to train Wonder3D on their own data and offers both inference models and codes. It supports various setup environments including Linux and Windows (via a specific branch), and also provides Docker support. Wonder3D is continuously evolving, with updates like Wonder3D++ and Era3D offering enhanced capabilities such as automatic focal length and elevation degree estimation.

MiniWorld is a minimalistic 3D interior environment simulator specifically designed for reinforcement learning and robotics research. It allows users to simulate environments featuring rooms, doors, hallways, and various objects, making it suitable for tasks like training AI agents in office, home, or maze-like settings. Written 100% in Python, MiniWorld is easily modifiable and extensible, offering features such as few dependencies, good performance, lightweight design, and support for domain randomization for sim-to-real transfer. It also provides fully observable top-down views, depth map production, and the ability to display alphanumeric strings on walls. This project has been deprecated as of August 11, 2025, and is no longer receiving updates or support.

Dream Textures is a powerful Blender add-on that brings Stable Diffusion capabilities right into your 3D workflow. It enables artists to generate a wide range of assets, including textures, concept art, and background elements, simply by using text prompts. A key feature is its 'Seamless' option, which ensures textures tile perfectly without visible seams, making it invaluable for 3D modeling and game development. The tool also allows for re-styling animations using the Cycles render pass and offers 'Project Dream Texture' for texturing entire scenes with depth-to-image. Users can run models locally for faster iteration or utilize DreamStudio for cloud processing if hardware is limited. It also includes AI upscaling for low-resolution generations and a history feature to recall and manage past creations.