diffusion-based generative models

This paper introduces CG3D, a novel method for generating scalable 3D assets from text prompts. Current text-to-3D methods struggle to produce detailed, multi-object scenes and maintain control over object configurations….