DreamFlow: High-quality text-to-3D generation by Approximating Probability Flow

Recent progress in text-to-3D generation has been achieved through the utilization of score distillation methods: they make use of the pre-trained text-to-image (T2I) diffusion models by distilling via the diffusion model training objective. However, such an approach inevitably results in the use of random timesteps at each update, which increases the variance of the gradient and ultimately prolongs the optimization process. In this paper, we propose to enhance the text-to-3D optimization by leveraging the T2I diffusion prior in the generative sampling process with a predetermined timestep schedule. To this end, we interpret text-to-3D optimization as a multi-view image-to-image translation problem, and propose a solution by approximating the probability flow. By leveraging the proposed novel optimization algorithm, we design DreamFlow, a practical three-stage coarseto-fine text-to-3D optimization framework that enables fast generation of highquality and high-resolution (i.e., 1024×1024) 3D contents. For example, we demonstrate that DreamFlow is 5 times faster than the existing state-of-the-art text-to-3D method, while producing more photorealistic 3D contents. 1 INTRODUCTION High-quality 3D content generation is crucial for a broad range of applications, including entertainment, gaming, augmented/virtual/mixed reality, and robotics simulation. However, the current 3D generation process entails tedious work with 3D modeling software, which demands a lot of time and expertise. Thereby, 3D generative models (Gao et al., 2022; Chan et al., 2022; Zeng et al., 2022) have brought large attention, yet they are limited by their generalization capability to creative and artistic 3D contents due to the scarcity of high-quality 3D dataset. Recent works have demonstrated the great promise of text-to-3D generation, which enables creative and diverse 3D content creation with textual descriptions (