DreamControl: Control-Based Text-to-3D Generation with 3D Self-Prior

TL;DR

DreamControl introduces a two-stage framework for controllable text-to-3D generation that addresses geometry inconsistency issues by leveraging 3D self-prior and control-based score distillation, resulting in high-quality 3D content.

Contribution

It proposes a novel two-stage 2D-lifting framework with adaptive viewpoint sampling and boundary integrity metrics to improve 3D generation quality and consistency.

Findings

Achieves high-quality 3D content with consistent geometry and textures.

Effective in downstream tasks like user-guided generation and 3D animation.

Addresses Janus problem in text-to-3D generation.

Abstract

3D generation has raised great attention in recent years. With the success of text-to-image diffusion models, the 2D-lifting technique becomes a promising route to controllable 3D generation. However, these methods tend to present inconsistent geometry, which is also known as the Janus problem. We observe that the problem is caused mainly by two aspects, i.e., viewpoint bias in 2D diffusion models and overfitting of the optimization objective. To address it, we propose a two-stage 2D-lifting framework, namely DreamControl, which optimizes coarse NeRF scenes as 3D self-prior and then generates fine-grained objects with control-based score distillation. Specifically, adaptive viewpoint sampling and boundary integrity metric are proposed to ensure the consistency of generated priors. The priors are then regarded as input conditions to maintain reasonable geometries, in which conditional LoRA and weighted score are further proposed to optimize detailed textures. DreamControl can generate high-quality 3D content in terms of both geometry consistency and texture fidelity. Moreover, our control-based optimization guidance is applicable to more downstream tasks, including user-guided generation and 3D animation. The project page is available at https://github.com/tyhuang0428/DreamControl.