DenseDPO: Fine-Grained Temporal Preference Optimization for Video Diffusion Models
This work addresses a specific bottleneck in video generation for AI researchers and practitioners, offering an incremental improvement over existing DPO methods.
The paper tackled the problem of motion bias and coarse comparisons in Direct Preference Optimization (DPO) for video diffusion models by introducing DenseDPO, which uses aligned video pairs and segment-level preferences to improve motion generation with only one-third of the labeled data, achieving performance close to human labels when using automatic annotation.
Direct Preference Optimization (DPO) has recently been applied as a post-training technique for text-to-video diffusion models. To obtain training data, annotators are asked to provide preferences between two videos generated from independent noise. However, this approach prohibits fine-grained comparisons, and we point out that it biases the annotators towards low-motion clips as they often contain fewer visual artifacts. In this work, we introduce DenseDPO, a method that addresses these shortcomings by making three contributions. First, we create each video pair for DPO by denoising corrupted copies of a ground truth video. This results in aligned pairs with similar motion structures while differing in local details, effectively neutralizing the motion bias. Second, we leverage the resulting temporal alignment to label preferences on short segments rather than entire clips, yielding a denser and more precise learning signal. With only one-third of the labeled data, DenseDPO greatly improves motion generation over vanilla DPO, while matching it in text alignment, visual quality, and temporal consistency. Finally, we show that DenseDPO unlocks automatic preference annotation using off-the-shelf Vision Language Models (VLMs): GPT accurately predicts segment-level preferences similar to task-specifically fine-tuned video reward models, and DenseDPO trained on these labels achieves performance close to using human labels.