Trajectory Forcing is proposed, an AR diffusion method that exploits the denoising trajectories of previous video chunks as context and achieves the highest overall quality among all compared methods.
Abstract
Autoregressive (AR) modeling is arguably the only way to extend a fixed-length video diffusion model to longer or even streaming videos. However, existing AR diffusion methods use only the previously predicted chunk as context for the next, discarding the intermediate predictions that constitute the chunk's generation history. In contrast, conventional AR models such as LLMs condition on the entire history of preceding tokens. To this end, we propose Trajectory Forcing (TraF), an AR diffusion method that exploits the denoising trajectories of previous video chunks as context. TraF operates under two training regimes: (1) with ground-truth videos, where pseudo-trajectories are constructed from Gaussian interpolations and re-predicted by the model, and (2) with a teacher model, where real trajectories are obtained from the student's own denoising rollout. The first regime naturally provides a strong initialization for the second. On the 5-second VBench benchmark, TraF achieves a Total score of 85.15, outperforming Self Forcing (+0.84) and Causal Forcing (+1.11), with Quality and Semantic improving simultaneously. On 30-second generation---6$\times$ the training horizon---TraF achieves the highest overall quality among all compared methods.
Can autoregressive video diffusion be accelerated further when only four denoising steps remain? Existing feature caching approaches struggle in this regime, as large feature changes between denoising timesteps undermine both direct reuse and lightweight predictors. We observe that features across video chunks remain s...
Zhuo Chen, Bin Zou, Fan-Yue Wei et al.· 0 citations
Streaming video world models generate the future chunk by chunk, conditioned on their own past and a stream of actions. Trained on short clips, they drift once a rollout exceeds the training horizon, as prediction errors re-enter the context and compound. A common remedy is to train the model on its own rollouts via di...
Yian Zhao, Yu Yan, Hong-Can Guo et al.· 0 citations
Glance Forcing is proposed, a lightweight distillation framework that decouples the denoising trajectory into two complementary slow and fast experts, each guided by a tailored, stage-specific objective, while keeping the pretrained teacher entirely frozen.
Zhuo-Bai Dong, Jun-Chao Yi, Hu-Jian Guo et al.· 0 citations
Conditional Residual Prediction (CRP), a simple recipe for reducing a model's reliance on a condition: the model first predicts the target without the condition, and the condition may only add a residual on top of this prediction.
Bo-Wen Zheng, Zhi-Guang Liu, Jia-Rong Ou et al.· 0 citations
Autoregressive video diffusion models generate videos chunk by chunk, yet producing plausible continuations does not guarantee that they capture how scenes evolve. Under teacher forcing, strong temporal redundancy may allow local visual continuity to serve as a shortcut for current-chunk denoising, reducing the pressur...
Ruo-Peng Gao, Yu-Kun Huang, Jia-Rui Hu et al.· 0 citations
Autoregressive (AR) video generation degrades over long horizons due to an overlooked train-inference discrepancy we term KV eviction mismatch: models train on short clips where all context frames reside in the KV cache, but at inference, memory constraints force distant frames to be evicted from the KV cache - removin...
Tri Cao, Hung Nguyen, P. Nguyen et al.· arXiv.org· 3 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.