DreamReward: Text-to-3D Generation with Human Preference
This addresses the issue of misalignment with human preferences in text-to-3D generation, which is important for users in 3D content creation, though it appears incremental as it builds upon existing diffusion models.
The paper tackles the problem of text-to-3D generation models not aligning well with human preferences by introducing DreamReward, a framework that learns from human feedback to improve these models, resulting in high-fidelity and 3D consistent outputs with significant boosts in prompt alignment.
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.