RDPO builds preference pairs by reverse-sampling real video latents with a pre-trained generator, then fine-tunes with Flow-DPO, improving physics consistency metrics on two video models.
All are worth words: A vit backbone for diffusion models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
RDPO: Real Data Preference Optimization for Physics Consistency Video Generation
RDPO builds preference pairs by reverse-sampling real video latents with a pre-trained generator, then fine-tunes with Flow-DPO, improving physics consistency metrics on two video models.