ModelScope Text-to-Video Technical Report

12 Aug 2023  ·  Jiuniu Wang, Hangjie Yuan, Dayou Chen, Yingya Zhang, Xiang Wang, Shiwei Zhang ·

This paper introduces ModelScopeT2V, a text-to-video synthesis model that evolves from a text-to-image synthesis model (i.e., Stable Diffusion). ModelScopeT2V incorporates spatio-temporal blocks to ensure consistent frame generation and smooth movement transitions. The model could adapt to varying frame numbers during training and inference, rendering it suitable for both image-text and video-text datasets. ModelScopeT2V brings together three components (i.e., VQGAN, a text encoder, and a denoising UNet), totally comprising 1.7 billion parameters, in which 0.5 billion parameters are dedicated to temporal capabilities. The model demonstrates superior performance over state-of-the-art methods across three evaluation metrics. The code and an online demo are available at \url{}.

PDF Abstract


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Text-to-Video Generation MSR-VTT ModelScopeT2V FID 11.09 # 1
CLIPSIM 0.2930 # 4
FVD 550 # 3


No methods listed for this paper. Add relevant methods here