Nowadays, various AI image generation softwares are being developed, but they basically generate 2D images. A scientific research team from Singapore recently announced that it is developing an AI system that directly produces 3DVR images from text. If the project is completed, it will be used for future Drawing, film, animation and even games and other fields have caused a huge impact and impact.
•Singapore’s Nanyang Technological University research team recently published a paper titled “Text2Light: Zero-Shot Text-Driven HDR Panorama Generation”, talking about the planning of this AI project, which attracted the attention of the industry.
• With the current technical level of the AI industry, it is not realistic to directly generate 4K-level HDR3D images through AI. However, this paper describes how to implement the plan. First, the first stage of AI needs to generate a 360-degree frame of low-level resolution LDR. Then the second stage upgrades the results of the first stage to the HDR level, and the research is currently in active progress.
source
Further reading: