Content & Design

LION (Latent Point Diffusion Models for 3D Shape Generation) is an open-source project presented at NeurIPS 2022, offering a robust framework for generating 3D shapes. This tool leverages advanced diffusion models to create 3D point clouds, enabling researchers and developers to explore and innovate in the field of 3D content creation. It includes functionalities for training VAE and diffusion prior models, with options for conditioning inputs like CLIP image embeddings for tasks such as single-view reconstruction or text-to-shape generation. The project provides detailed installation instructions, demo scripts, and evaluation tools, making it a valuable resource for those working with 3D shape synthesis and analysis.

Matterport3DSimulator is an AI research platform designed for deep reinforcement learning, computer vision, natural language processing, and robotics. It allows AI agents to interact with real 3D environments using visual information derived from panoramic RGB-D images. The simulator is based on the Matterport3D dataset, featuring 90 diverse indoor environments. Key capabilities include outputting real RGB and depth images, customizable image resolution and camera parameters, and support for off-screen rendering. It offers both C++ and Python APIs and is highly efficient, capable of around 1000 fps RGB-D off-screen rendering. The platform also includes the Room-to-Room (R2R) navigation dataset and task for training agents to follow natural language instructions.

OctNet is an open-source framework designed for deep learning with sparse 3D data, utilizing efficient space partitioning structures known as octrees. This approach significantly reduces the memory and compute requirements of 3D convolutional neural networks, allowing for the development of deep networks at high resolutions. By hierarchically partitioning space and storing pooled feature representations in leaf nodes, OctNet focuses memory allocation and computation on relevant dense regions. This enables deeper networks without sacrificing resolution, making it suitable for tasks such as 3D object classification, orientation estimation, and point cloud labeling. The framework includes core CPU and GPU code for network operations, data pre-processing tools, and a Torch wrapper for full network integration.

PointNet is a novel deep learning architecture specifically designed for processing point clouds, which are an important type of geometric data structure. Unlike traditional methods that convert point clouds into regular 3D voxel grids or image collections, PointNet directly consumes unordered point sets, respecting their permutation invariance. This approach makes it highly efficient and effective for a range of applications, including object classification, part segmentation, and scene semantic parsing in 3D. Developed by researchers at Stanford University, PointNet is available as an open-source project on GitHub, providing code and data for training classification and part segmentation networks. It has also served as a foundational work for subsequent advancements like PointNet++.

Trellis.2 AI 3D is an advanced online platform powered by Microsoft Research's 4-billion-parameter Trellis.2 AI model, designed to transform 2D images into high-fidelity 3D assets. Utilizing an innovative O-Voxel representation, it efficiently generates complex geometries and complete Physically-Based Rendering (PBR) material sets, including Base Color, Roughness, Metallic, and Alpha channels. The platform boasts remarkable speed, producing 3D models in seconds, and outputs standard GLB files compatible with major 3D software like Blender, Unity, and Unreal Engine. Trellis.2 AI 3D simplifies the 3D creation workflow by eliminating manual optimization, making it accessible for users to generate production-ready assets directly from an image.

Seedance 2.0 is a next-generation AI video generator designed to transform text and images into cinematic 15-second videos. It boasts advanced features like physics-based audio synchronization, ensuring realistic environmental sounds and dialogue that interact with the scene. The tool supports 2K resolution output at 24 FPS and offers multi-shot narrative capabilities with World ID technology for consistent character identity across frames. Users can generate videos from text prompts (up to 800 characters), images, or multimodal inputs combining up to 12 files. It supports various aspect ratios and styles, making it suitable for social media, marketing, and short films. Additionally, Seedance 2.0 provides video editing functionalities such as extension, in-painting, and character swap, along with API integration for automated workflows.

Archi AI is an innovative AI-powered platform designed for interior and exterior design, enabling users to generate photo-realistic images of their spaces quickly. Professionals and individuals can upload a picture of their room, select a desired style, and instantly create unlimited renders. The tool supports designing various rooms including living rooms, bedrooms, kitchens, bathrooms, and dining rooms, taking into account personal preferences to create customized designs. Archi AI aims to simplify the design process, saving time and money by allowing users to visualize different design ideas without physical alterations. It offers a cost-effective solution for transforming living spaces into dream environments.

AI Viggle is an AI platform designed to simplify video creation and animation processes by generating controllable videos. Users can leverage the platform to create dynamic video content from various sources, including still images, pre-existing video clips, and descriptive text prompts. The tool aims to make video production more accessible, offering features that assist in generating custom video content. It focuses on providing a streamlined workflow for transforming different media types into engaging video formats, catering to individuals and businesses looking to produce video content efficiently.

AI Action Figure Generator is an innovative tool that leverages artificial intelligence to create custom action figures from user descriptions. It allows users to detail character features, poses, accessories, and styles, which the AI then processes to generate unique figure designs. The platform is 100% free, offers unlimited image creation, and requires no sign-up, making it highly accessible. It boasts high-quality generation, customizable designs, fast processing, and support for multiple artistic styles. Users can download their creations in high resolution and share them, making it suitable for collectors, artists, game developers, and content creators looking for quick visualization and ideation.

CADflow.ai offers automated digital dentistry solutions specifically designed for dental labs of all sizes. This AI-powered platform streamlines the CAD workflow by automatically preparing 3D models for dental appliance production, significantly reducing the time required for these tasks. The cloud-based nature of CADflow.ai ensures 24/7 accessibility and scalability, allowing labs to meet varying demands efficiently. It aims to enhance productivity and accuracy in digital dentistry, making it an essential tool for modern dental practices looking to optimize their operations and improve turnaround times for dental prosthetics and appliances.

JustAHuman offers a unique gamified platform for 3D asset evaluation and labeling, allowing users to earn rewards while contributing to data annotation. Players accumulate points by completing challenges, which can then be converted into game credits, GenAI service provider credits, or crypto. This innovative approach aims to improve the efficiency and accuracy of AI model training by engaging users in a fun and rewarding way. The platform is designed to connect game creators with a community that can help process and label their 3D assets, making it a valuable resource for both players and developers.

Deco AI Room Design is an AI-powered application designed to help users generate personalized room design ideas. This tool enables individuals to visualize various interior design concepts, making it easier to explore different styles and layouts for their living spaces. It is particularly useful for homeowners looking for inspiration or interior designers seeking quick design options. The platform aims to simplify the design process by providing AI-generated suggestions, helping users to experiment with different aesthetics before making any physical changes. While specific features are not detailed, the core functionality revolves around providing visual design concepts.

FLUX.1 Open Ghibli Studio LoRA is an AI web application designed to convert uploaded images into the distinctive Ghibli art style. This tool allows users to transform their photos into animated artwork with customizable settings for image size, quality, and the intensity of the Ghibli-style effect. It also provides a detailed description of the original image, aiding in the creative process. Built with Gradio and utilizing the openfree/flux-chatgpt-ghibli-lora model, it offers a user-friendly interface for AI art creation and image style transfer. While currently paused, it aims to provide a unique artistic transformation for various visual content.

StableNormal is an open-source AI tool designed to enhance monocular normal estimation by reducing the inherent stochasticity of diffusion models. This approach leads to "Stable-and-Sharp" normal maps, outperforming various baselines in terms of accuracy and stability. The tool is presented as a research project from SIGGRAPH Asia 2024 and provides a Python-based pipeline for installation and usage. It includes a faster inference option, StableNormal-turbo, which is 10 times quicker. Users can compute metrics on datasets like DIODE, IBims-1, Scannet, and NYUv2 to evaluate performance, making it suitable for researchers and developers in computer vision and generative AI.

Daz 3D offers AI datasets specifically designed for character generation and machine learning. This platform provides fully licensed, structured 3D datasets that accelerate the training of AI models. With millions of production-grade assets, including rigged characters, morph systems, clothing, accessories, and environments, Daz 3D enables infinite dataset generation. The data is structured for AI with clean topology, rigged systems, and machine-readable scene structures, ensuring perfect ground truth with automatically generated precise labels, segmentation, and depth data. Daz 3D's synthetic data supports various industries, including gaming, computer vision, robotics, autonomous systems, virtual humans, retail, defense, AR/VR, and healthcare, by providing scalable and controllable training environments.

Poetry3D is an innovative artistic visualization project that transforms user-entered poems into unique 3D semantic trees. This tool serves as a practical demonstration of core AI concepts, including tokenization, vector embeddings, vector databases, and cosine similarity. By representing each word as a point in a multi-dimensional space, where words with similar meanings are positioned closely, Poetry3D visually explains how AI processes and understands language. Connections between words form branches, and parallel branches indicate similar phrases, revealing hidden patterns and semantic structures within the poem. The project flattens AI's 1,536-dimensional word space into a 3D visualization, offering a 'shadow of meaning' that is unique to each poem.

SVD_Xtend offers comprehensive training code and extensions for Stable Video Diffusion, allowing users to finetune SVD models for customized video generation. A key feature is tracklet-conditioned video generation, which provides precise control over object movement within videos using bounding box information. The tool supports various video data processing methods, including the use of datasets like BDD100K, and offers detailed training configurations. It also integrates methods from Boximator and TrackDiffusion for enhanced control and instance-level manipulation. SVD_Xtend is ideal for AI researchers and developers looking to experiment with and advance video diffusion models.

3D Generator is an AI-powered tool hosted on Hugging Face Spaces, designed to create detailed 3D images from simple text descriptions. Users can quickly generate 3D-looking art by typing in their desired visual concepts. This application simplifies the process of 3D rendering, making it accessible for various creative needs. It's particularly useful for generating visual content rapidly, whether for educational projects, fun art creation, or quick conceptualization. The tool aims to provide an intuitive experience, allowing anyone to produce 3D renders without requiring extensive technical knowledge in 3D modeling or design software.

3DTopia-XL is an AI-powered tool designed to streamline the creation of 3D models from 2D images. Users can upload an image, and the application automatically removes the background, generates a corresponding 3D model, and provides both video renders and a GLB file of the final model. The tool offers adjustable settings such as steps, seed, and resolution, allowing for fine-tuning of the generation process to achieve better results. This makes it a valuable asset for anyone looking to quickly convert images into usable 3D assets for various applications.

PROVEN Solution, founded in 2016, is a technology company specializing in AI, robotics, automation, data analytics, and immersive technologies like virtual and augmented reality. They offer expert consulting in business strategy, digital transformation, and operational efficiency, aiming to empower businesses with innovative tools. With a strong focus on emerging and future technologies, PROVEN Solution helps drive efficiency and transformation, particularly in Saudi Arabia and the GCC region. Their solutions are designed with a deep understanding of industry challenges, ensuring seamless integration and real-world impact. They also offer Arabic-first OCR platforms and energy-saving IoT solutions.

InstantMesh is an open-source framework designed for efficient 3D mesh generation from a single image. Built upon the LRM/Instant3D architecture, it provides a feed-forward approach to create detailed 3D models. The tool supports various sparse-view reconstruction model variants and includes fine-tuning code for Zero123++. Users can generate 3D meshes from images via a local Gradio demo or command line, with options to save as .obj files with vertex colors or texture maps. It also offers features like automatic foreground segmentation and support for multi-GPU setups to optimize memory usage. The project is available on GitHub, providing both inference and training code for researchers and developers.

MeshDiffusion is an open-source implementation of a diffusion model designed for generating 3D meshes. It leverages a direct parametrization of deep marching tetrahedra (DMTet) to create 3D models. The tool allows for both unconditional generation of 3D meshes and single-view conditional generation, where users can complete occluded regions of a mesh from a single view. It supports training diffusion models on custom datasets and provides pretrained models for various object categories like chairs, cars, airplanes, tables, and rifles. Additionally, MeshDiffusion offers functionalities for texture generation and visualization of generated meshes using Blender.

MotionGPT is an innovative open-source project that unifies human motion and language generation through large language models (LLMs). It treats human motion as a foreign language, converting 3D motion into discrete motion tokens similar to word tokens. This approach allows for language modeling on both motion and text in a unified manner, enabling the generation of high-quality motions and text descriptions across multiple tasks. MotionGPT supports text-driven motion generation, motion captioning, motion prediction, and motion in-between. It leverages prompt learning and instruction tuning to achieve state-of-the-art performance, demonstrating the potential of LLMs in motion tasks beyond traditional language generation.

PointLLM is a multi-modal large language model designed to understand colored point clouds of objects. It excels at perceiving object types, geometric structures, and appearance, effectively bypassing common issues like ambiguous depth, occlusion, or viewpoint dependency. The tool leverages a novel dataset comprising 660K simple and 70K complex point-text instruction pairs, enabling a robust two-stage training strategy. PointLLM also establishes two benchmarks, Generative 3D Object Classification and 3D Object Captioning, for rigorous evaluation. It offers capabilities for inferencing, chatting with 3D models, and evaluation using traditional metrics or GPT-4, making it a powerful resource for advanced 3D data analysis and robotics applications.