One-Minute Video Generation with Test-Time Training
Karan Dalal, Daniel Koceja, Gashon Hussein, Jiarui Xu, Yue Zhao, Youjin Song, Shihao Han, Ka Chun Cheung, Jan Kautz, Carlos Guestrin, Tatsunori Hashimoto, Sanmi Koyejo, Yejin Choi, Yu Sun, Xiaolong Wang
2025-04-08
Summary
This paper talks about a new AI method that creates one-minute videos from text descriptions, helping AI tell longer stories like cartoon episodes instead of just short clips.
What's the problem?
Current AI video makers either get too slow with long videos or mess up story details when trying to handle multiple scenes, like forgetting character colors or mixing up actions.
What's the solution?
The method adds 'smart memory' layers (TTT) to existing AI video tools, letting them remember story details better and keep characters consistent across scenes by learning even while making videos.
Why it matters?
This could help create better AI-made cartoons, tutorials, or movie scenes that stay consistent over time, making long video creation easier for animators and content creators.
Abstract
Transformers today still struggle to generate one-minute videos because self-attention layers are inefficient for long context. Alternatives such as Mamba layers struggle with complex multi-scene stories because their hidden states are less expressive. We experiment with Test-Time Training (TTT) layers, whose hidden states themselves can be neural networks, therefore more expressive. Adding TTT layers into a pre-trained Transformer enables it to generate one-minute videos from text storyboards. For proof of concept, we curate a dataset based on Tom and Jerry cartoons. Compared to baselines such as Mamba~2, Gated DeltaNet, and sliding-window attention layers, TTT layers generate much more coherent videos that tell complex stories, leading by 34 Elo points in a human evaluation of 100 videos per method. Although promising, results still contain artifacts, likely due to the limited capability of the pre-trained 5B model. The efficiency of our implementation can also be improved. We have only experimented with one-minute videos due to resource constraints, but the approach can be extended to longer videos and more complex stories. Sample videos, code and annotations are available at: https://test-time-training.github.io/video-dit