DreamReward: Text-to-3D Generation with Human Preference

Paper | Project Page

Junliang Ye^*1,2, Fangfu Liu^*1, Qixiu Li¹, Zhengyi Wang^1,2, Yikai Wang¹, Xinzhou Wang^1,2, Yueqi Duan^1,✉, Jun Zhu^1,2,✉

¹Tsinghua University   ²ShengShu  ^* Equal Contribution^✉ Corresponding Author

Our Code will be released soon... 🏗️ 🚧 🔨

Abstract: 3D content creation from text prompts has shown remarkable success recently. However, current text-to-3D methods often generate 3D results that do not align well with human preferences. In this paper, we present a comprehensive framework, coined DreamReward, to learn and improve text-to-3D models from human preference feedback. To begin with, we collect 25k expert comparisons based on a systematic annotation pipeline including rating and ranking. Then, we build Reward3D---the first general-purpose text-to-3D human preference reward model to effectively encode human preferences. Building upon the 3D reward model, we finally perform theoretical analysis and present the Reward3D Feedback Learning (DreamFL), a direct tuning algorithm to optimize the multi-view diffusion models with a redefined scorer. Grounded by theoretical proof and extensive experiment comparisons, our DreamReward successfully generates high-fidelity and 3D consistent results with significant boosts in prompt alignment with human intention. Our results demonstrate the great potential for learning from human feedback to improve text-to-3D models.

Comparison with MVDream

More Visual Results

Quantitative Comparison

We compared our DreamReward on 110 prompts generated by GPTEval3D. Left: User study of the rate from volunteers’ preference for each method in the inset pie chart, Right: Holistic evaluation using GPTEval3D.

We compared our DreamReward with DreamFusion, ProlificDreamer, Latent-NeRF, MVDream, and Fantasia3D. We calculate CLIP↑, ImageReward↑, GPTEval3D ↑ and Reward3D↑.

BibTeX

@misc{ye2024dreamreward,
      title={DreamReward: Text-to-3D Generation with Human Preference}, 
      author={Junliang Ye and Fangfu Liu and Qixiu Li and Zhengyi Wang and Yikai Wang and Xinzhou Wang and Yueqi Duan and Jun Zhu},
      year={2024},
      eprint={2403.14613},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}