Atomic Generation
One-shot plausible rendering from prompts or latent codes.
Dynamic Project Page | Roadmap Paper Explorer
Visual Generation in the New Era
An evolution from atomic mapping to agentic world modeling.
This roadmap organizes modern visual generation around capabilities: controllable composition, persistent context, agentic interaction, and causal world simulation. Use the dynamic explorers below to retrieve papers, tasks, categories, and stress-test cases from the full roadmap bibliography.
Roadmap Lens
From prettier images to verifiable visual intelligence
Conditional Generation
Faithful control under layouts, references, and explicit constraints.
In-Context Generation
Multi-reference, multi-condition, and long-context visual coherence.
Agentic Generation
Planning, tool use, verification, rollback, and iterative refinement.
World Modeling
Causal, physical, and action-conditioned simulation of visual worlds.
Figures
Key maps from the roadmap
Paper Explorer
Search the full roadmap bibliography
Retrieval is fully client-side. Try paper titles, authors, BibTeX keys, methods, tasks, or capability words.
Year Distribution
Click a year to focus the paper list.
Task Map
Click a task cluster to retrieve related papers.
No papers found. Try a broader keyword or clear filters.
Stress-Test Explorer
When "looks right" is not enough
Browse the in-the-wild probes from the roadmap: spatial constraints, physical causality, multi-turn drift, human-centric editing, restoration, real-world design, and structured vision tasks.
Dimension
Citation
Cite this roadmap
If you find this roadmap useful, please cite the project.
@misc{wu2026visualgenerationnewera,
title={Visual Generation in the New Era: An Evolution from Atomic Mapping to Agentic World Modeling},
author={Keming Wu and Zuhao Yang and Kaichen Zhang and Shizun Wang and Haowei Zhu and Sicong Leng and Zhongyu Yang and Qijie Wang and Sudong Wang and Ziting Wang and Zili Wang and Hui Zhang and Haonan Wang and Hang Zhou and Yifan Pu and Xingxuan Li and Fangneng Zhan and Bo Li and Lidong Bing and Yuxin Song and Ziwei Liu and Wenhu Chen and Jingdong Wang and Xinchao Wang and Xiaojuan Qi and Shijian Lu and Bin Wang},
year={2026},
eprint={2604.28185},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2604.28185},
}