DreamPolisher: Towards High-Quality Text-to-3D Generation via Geometric
Diffusion
DreamPolisher: Towards High-Quality Text-to-3D Generation via Geometric
Diffusion
We present DreamPolisher, a novel Gaussian Splatting based method with geometric guidance, tailored to learn cross-view consistency and intricate detail from textual descriptions. While recent progress on text-to-3D generation methods have been promising, prevailing methods often fail to ensure view-consistency and textural richness. This problem becomes particularly noticeable for methods …