Search AI Products and News

Tencent HunYuan recently announced the launch of five new open-source 3D generative models. Built upon Hunyuan3D-2.0, these models boast faster generation speeds, richer details, and more realistic material expression. Simultaneously, Tencent's self-developed 3D AI creation engine has been upgraded, adding features such as multi-view input, intelligent model simplification, and full format compatibility. It is now fully open to C-end users. Among these new models, the Turbo series maintains high precision and quality while optimizing the geometric generation model...

Today, Tencent announced the launch of the Hongyuan 3D AI Creation Engine, designed to quickly generate high-quality 3D models through prompts or images. This creation engine integrates a 3D feature matrix, editing tools, generation workflows, and a rich library of creative materials, providing great convenience for designers and creators. Users only need to enter simple prompts, such as 'round brown long sofa' or 'owl, big eyes, dark brown,' to easily generate corresponding 3D models and choose different styles for download, supporting mainstream 3D formats.

Recently, a Ukrainian website focused on human 3D models, Trilegangers, faced an unprecedented traffic attack that caused its servers to crash. The site aims to provide a vast array of human 3D model data for 3D artists and game developers, but has been overwhelmed by frequent crawling from OpenAI's GPTBot. According to Trilegangers staff, despite the site's usage policy explicitly prohibiting unauthorized crawling and use, the situation has not been adequately addressed.

Startup company Meshy has launched its AI-driven 3D modeling tool Meshy-4, achieving significant improvements in model geometry and workflow after 16 months of development, greatly enhancing the creative efficiency of designers and developers. Meshy-4 excels in the field of AI-generated 3D models, producing smoother surfaces and clearer geometric details, approaching professional use standards. Its innovative text-to-3D model generation process is divided into two stages: modeling and texturing, increasing user control over the final output and paving the way for future specialized features.

Ali Tongyi XR Lab has open-sourced new Text-to-3D models, which can generate corresponding 3D models and textures from a text description. The open-sourced models include two major models: Text-to-ND and Text-to-ND-MV. The Tongyi XR Lab has released the basic version of Text-to-ND as well as the Multi-View version to meet different fine-grained algorithm development needs. It also synchronized the open-sourcing of distilled versions from the large model.

The AI software developed by the Martin Luther University team utilizes 3D models to interpret difficult-to-read cuneiform texts, yielding results that are more reliable than traditional methods. By comparing the content of multiple cuneiform tablets, it offers researchers a novel approach to study. The AI was trained on 3D models of nearly 2000 cuneiform tablets, enabling the possibility of interpreting various cuneiform texts and ancient inscriptions. This AI system not only performs well on high-quality images but can also successfully decode poorer quality images, opening new avenues for research.

Popular AI startup Kaedim falsely promotes its 3D model generation technology. Kaedim claims that its technology can automatically generate high-quality 3D models from 2D images; however, the quality of the AI-generated 3D images is poor. Kaedim sometimes requires artists to create high-quality 3D images completely from scratch. A warning to AI companies is that they should not overly exaggerate the current state of their technology.

Nvidia has launched FlexiCubes, utilizing AI to generate 3D meshes needed for various 3D applications. FlexiCubes can be integrated with physics engines to create soft objects within 3D models. This innovative mesh generation method is expected to revolutionize AI pipelines and provide high-quality meshes for various applications.