Meta has unveiled a brand new artificial intelligence (AI) tool that generates high-quality 3D assets in only one minute using text descriptions. The corporate emphasized that this can be a groundbreaking advancement in the sector, because it enables the creation of high-quality 3D assets in a brief time frame.
VentureBeat reported on the 2nd (local time) that Meta introduced ‘3D Gen’, a 3D asset creation AI tool that may implement high-accuracy and high-quality 3D shapes and textures.
In accordance with this, 3DGen is an integration of â–²’3D AssetGen’ for text-3D shape generation and â–²’3D TextureGen’ technology for text-text generation.
It also supports Physically Based Rendering (PBR), allowing you to create realistic 3D assets under various lighting conditions.
In step one, 3DGen uses 3D AssetGen, a text-to-3D shape generator, to generate initial 3D assets based on text prompts. This process outputs a 3D mesh with textures and PBR maps in lower than 30 seconds.
Then, with 3D TextureGen, you may generate higher quality textures and PBR maps based on the 3D assets generated within the previous step in only 20 seconds. You may also use recent text prompts to specific textures for previous step creations or 3D meshes.
In this manner, 3DGen is explained as effectively combining the view space that expresses the image of an object, the amount space that expresses 3D shapes, and the UV space that expresses textures.
Since only sample footage has been released, actual performance has not been verified. Nonetheless, the flexibility to create high-quality 3D assets in a brief time frame is a transparent strength.
Particularly, by integrating AI-generated 3D assets into existing pipelines, it may possibly speed up the creation of virtual environments and digital twins similar to virtual reality (VR) or augmented reality (AR) in various industries. It is predicted to have huge potential for ripple effects, similar to accelerating the event of immersive experiences for education, training, entertainment, etc. by lowering the barrier to creating high-quality 3D assets.
Meanwhile, the technology to generate 3D assets based on 2D images or text prompts is a field that many firms have jumped into. It’s because there may be high demand for it in areas similar to game production and e-commerce, and it may possibly generate revenue straight away.
In March, NVIDIA introduced ‘Latte3D’, which may generate 3D objects and animal images in real time in only one second from a straightforward text prompt, and Stability AI also released an AI model capable of manufacturing 360-degree 3D rendering videos.
Reporter Park Chan cpark@aitimes.com
