The paper introduces Relax Forcing, a training‑free memory mechanism for autoregressive video diffusion that structures temporal context into Sink, Tail, and History frames. By selecting History frames with a relaxation criterion, the method reduces error accumulation and attention overhead while preserving motion dynamics. Experiments on VBench‑Long demonstrate that this structured memory improves long‑video generation quality over existing baselines.
By Zengqun Zhao, Yanzuo Lu, Ziquan Liu, Jifei Song, Jiankang Deng, Ioannis Patras
arXiv:2609.23658v1 Announce Type: cross
Abstract: Despite impressive visual quality, state-of-the-art video diffusion models often generate content that violates real-world physical laws. While exist...
By Yueyan Li, Haibo Wang, Caixia Yuan, Xiaojie Wang
LongVU‑TTT is a causal test‑time training method for long‑video multimodal large language models that inserts a convolutional resampler with fast‑weight updates between the vision encoder and the LLM. The fast weights adapt per video and contextualize frame features before compression, while a hybrid selector keeps explicit visual evidence for downstream reasoning. Experiments show that TTT‑Conv outperforms TTT‑MLP and bidirectional Mamba2 on MLVU, and beats attention‑ and fixed‑state recurrent resamplers on three benchmarks, achieving competitive results on five video‑understanding tasks after reducing 512 frames to 128 LLM frames.
By Mahmoud Ahmed, Sameh Abdulah, Olatunji Ruwase, Sam Ade Jacobs, Mathis Bode, Mohamed Elhoseiny
World in World introduces a training‑free, inference‑time interface that transforms diverse control signals—such as source‑video observations, target‑view projections, geometry renderings, and retrieved states—into camera‑ and time‑labelled visual states. These states are processed by a frozen causal video model’s self‑attention, enabling tasks like camera‑controlled rerendering, long‑horizon revisiting, and human‑motion transfer without additional training. The method employs a correspondence router and evidence‑wise attention to align token identities and regulate auxiliary channel contributions during a single denoising pass.
By Chenxi Song, Yanming Yang, Chi Zhang
arXiv:2609.36832v1 Announce Type: new
Abstract: Text-to-video (T2V) diffusion models can generate realistic depictions of actions such as kicking, stabbing, and shooting, raising safety concerns that...
By Ping Liu, Chi Zhang
World in World introduces a training‑free inference interface that lets users control autoregressive video world models from new viewpoints. By converting diverse control signals—source‑video observations, target‑view projections, geometry renderings, and retrieved states—into camera‑ and time‑labelled visual tokens, the system uses a frozen causal video model’s self‑attention to maintain synchronization, complete unseen regions, and recover past appearances. The method supports camera‑controlled rerendering, long‑horizon revisiting, and human‑motion transfer while preserving perceptual quality, temporal consistency, and camera‑following accuracy.
arXiv:2608.22861v1 Announce Type: new
Abstract: State Space Models (SSMs) have surfaced as a promising architecture in Video Frame Interpolation (VFI), as they can capture long-range dependencies wit...
By Jaehyun Park, Nam Ik Cho
DynaTokens introduces lightweight, scene‑specific tokens that enable camera‑controlled video models to learn and generate dynamic scene motion while keeping the base model frozen. By leveraging cross‑attention, the tokens are trained from a few example trajectories, allowing the model to handle both camera‑induced motion and localized object dynamics. Experiments on VBench2 and WorldScore show that DynaTokens outperforms LoRA, block finetuning, and other trainable‑layer baselines in balancing dynamics and camera control.
arXiv:2607. 14194v1 Announce Type: cross Abstract: Text-to-video (T2V) generators can synthesize realistic and temporally coherent videos, but controllably removing a target concept from a generator remains difficult.
By Wenxuan Chen, Wenjie Feng
CameraEditor is a new framework that transforms camera-controlled image editing into a temporal sequence prediction problem. By using video diffusion models, it incorporates a geometric perception module and dynamic reference routing to create precise visual references through dynamic panorama cropping. The method also inserts intermediate transition frames to handle large perspective shifts, maintaining content identity and spatial coherence, and is evaluated on a dataset of 5,760 instances with a benchmark of 462 test cases, achieving state‑of‑the‑art performance.
By Xin Shen, Chengyou Jia, Keshuo Xing, Zifeng Zhu, Changliang Xia, Bowen Ping, Zhuohang Dang, Hangwei Qian, Minnan Luo
Long-video MLLMs must model temporal change before a limited visual-token budget removes most frame evidence. We introduce LongVU-TTT, which inserts a convolutional Test-Time Training (TTT) resampler...
arXiv:2609.24788v1 Announce Type: new
Abstract: In this paper, we propose SVEET, a framework that requires merely training on a pretrained bidirectional video diffusion model but supports high-qualit...
By Yujia Hu, Jiajun Li, Zihao He, Songhua Liu