[Technical Report] EchoWM: Open and Enterable Omnimodal World Models
Songchun Zhang^*, Yaowei Li^*, Junhao Zhuang^*, Weiyang Jin^*, Haoyu Wang, Xin Lu, Yilang Sun, Shiyi Zhang, Haoran Li, Xiaoxiao Ma, Yuming Li, Yijun Liu, Yaofeng Su, Yanwen Ma, Haoyu Wu, Zihan Su, Yue Ma, Lvmin Zhang, Haoyang Huang, Zeyue Xue, Anyi Rao, Nan Duan
[Project page] [Paper]

• An open and enterable omnimodal world model for real-time audio-visual generation, camera-intent control, and interactive world exploration.

[Technical Report] JoyAI-Echo: Pushing the Frontier of Long Audio-Visual Generation
Echo Team @ Joy Future Academy, JD
[Project page] [Paper] [Code]

• This technical report presents a memory-driven audio-visual generation framework for minute-level coherent video, real-time streaming, conversational control, and high-resolution output.

[Arxiv] Echo-Infinity: Learnable Evolving Memory for Real-Time Infinite Video Generation
Yuxuan Bian, Zeyue Xue, Songchun Zhang, Shiyi Zhang, Weiyang Jin, Yaowei Li, Junhao Zhuang, Haoran Li, Jie Huang, Haoyang Huang, Nan Duan, Qiang Xu.
[Project page] [Paper] [Code] [Model]

• This paper presents a learnable evolving memory framework for real-time infinite video generation with constant-cost long-history compression.

[Arxiv] Astrolabe: Steering Forward-Process Reinforcement Learning for Distilled Autoregressive Video Models
Songchun Zhang, Zeyue Xue, Siming Fu, Jie Huang, Xianghao Kong, Yue Ma, Haoyang Huang, Nan Duan, Anyi Rao.
[Project page] [Paper] [Code]

• This paper presents an efficient online RL framework for aligning distilled autoregressive video models with human visual preferences.

[ECCV 2026] FlexComposer: Unified Video Compositing from Images to Dynamic Footage with Flexible Trajectory Control
Songchun Zhang, Sitong Guo, Xianghao Kong, Pengwei Liu, Yuwei Guo, Lvmin Zhang, Anyi Rao.
[Project page] [Paper]

• This paper presents a unified trajectory-guided video compositing framework for seamlessly integrating static images and dynamic footage with flexible motion control.

[TOG 2026] LiveLight: Real-time Streaming Video Relighting with Interactive Control
Yue Ma, Jiangming Wang, Yucheng Wang, Xilai Wang, Zhiyuan Li, Xinyu Wang, Hongyu Liu, Ruofan Liang, Songchun Zhang, Yuxuan Xue, Qifeng Chen.
[Project page] [Paper] [Code]

• This paper presents a real-time streaming video relighting framework with interactive 3D point-light control, supporting long videos while preserving appearance and temporal coherence.

[SIGGRAPH Asia 2026] ShotVerse: Advancing Cinematic Camera Control for Text-Driven Multi-Shot Video Creation
Songlin Yang, Zhe Wang, Xuyi Yang, Songchun Zhang, Xianghao Kong, Taiyi Wu, Xiaotong Zhao, Ran Zhang, Alan Zhao, Anyi Rao.
[Project page] [Paper] [Code]

• This paper introduces a plan-then-control framework for cinematic multi-shot video generation, combining a VLM planner with camera-trajectory control for globally consistent shots.

[CVPR 2026 Highlight] Composing Concepts from Images and Videos via Concept-prompt Binding
Xianghao Kong, Zeyu Zhang, Yuwei Guo, Zhuoran Zhao, Songchun Zhang, Anyi Rao.
[Project page] [Paper]

• This paper presents a one-shot method for flexible visual concept composition by binding visual concepts with prompt tokens.

[ICCV 2025] SpatialCrafter: Unleashing the Imagination of Video Diffusion Models for Scene Reconstruction from Limited Observations
Songchun Zhang, Huiyao Xu, Sitong Guo, Zhongwei Xie, Pengwei Liu, Hujun Bao, Weiwei Xu, Changqing Zou.
[Project page] [paper]

• This paper presents a 3D scene reconstruction method from sparse inputs.

[AAAI 2025] Pragmatist: Multiview Conditional Diffusion Models for High-Fidelity 3D Reconstruction from Unposed Sparse Views
Songchun Zhang, Chunhui Zhao.
[Paper] [Code]

• This paper presents a 3D object reconstruction method from sparse and unposed inputs.

[CVPR 2024] 3D-SceneDreamer: Text-Driven 3D-Consistent Scene Generation
Songchun Zhang, Yibo Zhang, Quan Zheng, Rui Ma, Wei Hua, Hujun Bao, Weiwei Xu, Changqing Zou.
[Project page] [Paper] [Code]

• This paper presents a novel text-driven 3D scene generation method that improves visual quality and 3D consistency.

[TCSVT 2023] Cross-Video Contextual Knowledge Exploration and Exploitation for Ambiguity Reduction in Weakly Supervised Temporal Action Localization
Songchun Zhang, Chunhui Zhao.
[Paper] [Code]

• This paper presents a weakly-supervised action localization framework leveraging cross-video information.