StableDreamer enhances text-to-3D generation, resolving common issues in 2D diffusion models. It blends image and latent-space diffusion for better color and detail, uses 3D Gaussians for efficiency, and employs a unique training approach for clearer, high-quality 3D models.

StableDreamer enhances text-to-3D generation, resolving common issues in 2D diffusion models. It blends image and latent-space diffusion for better color and detail, uses 3D Gaussians for efficiency, and employs a unique training approach for clearer, high-quality 3D models. 

https://arxiv.org/abs/2312.02189