Researchers at Meta ( $Meta Platforms (META.US)$ ) and the U...
Researchers at Meta ( $Meta Platforms (META.US)$ ) and the University of Oxford have developed a new 3D generative AI model, VFusion3D, capable of generating high-quality 3D objects from a single image or textual description, potentially transforming fields such as virtual reality, gaming and digital design.
The new AI model VFusion3D is here!
The VFusion3D interface is simple and intuitive, allowing users to upload their images or select from a series of preloaded examples. The preloaded examples performed very well, generating 3D models and rendering videos that captured the essence and details of the original 2D images with remarkable accuracy.
The team’s use of pre-trained video AI models to generate synthetic 3D data allowed them to train a more robust 3D generation system, solving the long-standing problem of scarce 3D training data. With this breakthrough, it may ultimately accelerate innovation in 3D content across industries.
Generating AI+ 3D Models
Since ChatGPT caught fire globally, with the introduction of GPT-4, a multimodal large model that supports image input, as an opportunity. The multimodal computing potential of large models is starting to take technological advantage in the content generation space. Top tech companies and startups looking to capitalize on the AIGC windfall are starting to scramble to launch AI-powered 3D model generators.
The more widely used alternative technology path of generating 3D models from 2D diffusion models, which makes the leap from text to 3D models, hinges on two technical points.
That said, in recent times, the text-to-3D generation method has been able to generate impressive 3D effects, and the last time it came out of the ring was at SIGGRAPH. In 2021, NVIDIA introduced 3D modelling methods on this stage, which rocked the world with its faked effects.
And just last week at SIGGRAPH 2024, Jen-Hsun Huang, CEO of NVIDIA ( $NVIDIA (NVDA.US)$ ), spoke with Meta’s CEO Zuckerberg, with both sides endorsing the evolutionary direction of AI macromodeling.
In addition, Apple ( $Apple (AAPL.US)$ ) is researching generative AI technology “HUGS” which, can generate digital human doppelgangers in about 30 minutes, broadening the use of 3D AIGC technology scenarios.
The core of 3D AIGC and the virtual content field represented by XR is to improve the efficiency of creating 3D assets. This is especially true for the XR industry, which is based on virtual worlds, and the emergence of 3D AIGC technology is likely to revolutionize the XR content industry, which is still in its early stages of development.
Data show that WiMi Hologram Cloud (NASDAQ: WIMI), as a 3D vision and AI algorithm manufacturer, has accelerated the layout of 3D whole chain collaborative innovation in recent years to build a 3D ecosystem and industrialized applications based on independent technology, while strengthening the construction of strategic talent power and attracting and nurturing 3D technological innovations and industry-leading talents at multiple levels, which gives other platforms space for further development.
At present, WiMi combines artificial intelligence technology and 3D generation to achieve more efficient and higher quality 3D digital human content generation, and the generated digital model and texture are optimized to achieve personalized 3D bearded character generation. In addition, the generated characters can be viewed in a 360-degree view, and there is also the option to share or make the characters public, which greatly improves the efficiency of 3D digital human content production.
Today, AI+3D technology has been seen to grow tremendously in markets such as e-commerce, gaming, AR/VR, etc., and in the future, it will also empower the next phase of 3D content creation to help. In this regard, looking to the future, WiMi’s 3D generation capabilities + AI vision technology empower terminal perception, and will also continue to be combined with richer scenarios to continue to help more industrial development.
Conclusion
Overall, 3D AIGC technology has leaped nothing to something. In terms of application scenarios, all industries that are inseparable from 3D will become potential customers of 3D AIGC technology in the future, and in this regard, enterprises will become bigger winners in the future if they rush into the game to combine AI images, videos, 3D and real-time AI-generated models, and create fully immersive virtual environments through text prompts.
Disclaimer: Community is offered by Moomoo Technologies Inc. and is for educational purposes only.
Read more
Comment
Sign in to post a comment
