About
Generative AI has made significant strides in creating visually impressive content, yet it often falls short in producing coherent and engaging narratives. Storytelling, whether in written or visual form, requires consistency and depth across multiple aspects, such as character development, plot progression, and setting dynamics, which current generative models struggle to achieve. This workshop aims to address these limitations by exploring the integration of advanced visual generative models, large language models, and multi-modal AI techniques. Our focus is on advancing the capabilities of generative AI to produce coherent and dynamic narratives that maintain visual consistency and accurately convey story elements, rather than just cool graphics. We seek to push the boundaries of generative AI to create storytelling experiences that are both visually engaging and narratively cohesive, which further expands the applicability of AI to content creation.
Topics include:
- Cross-Modal Storytelling with Text and Visual Generation
- Character Consistency in Multi-Modal Storytelling
- Visual Editing and Consistency for Stories
- Large Language Models for Coherent Story Generation
- Vision and Language Models for Visual Narrative Generation
- Reader-Interactive Story Elements with Real-Time Visual Adaptation
- Symbolism and Thematic Representation in Generated Visuals
- Temporal Progression in Visual Storytelling
- Synchronizing Narrative Tone and Visual Style
- Using Visual Prompts to Guide AI-Generated Storylines
- AI-Generated Visuals for Mood and Atmosphere in Stories
- Visual Interpretation of Textual Plot
- Generating Emotionally Expressive Character Art for Narratives
- Ethical Implications of AI in Multi-Modal Narrative Creation
Format: In-person, Half-day
Contact: Andrew Shin, shin@inl.ics.keio.ac.jp
Call for Papers
We invite [4-page short papers/ 8-page long papers] (excluding references) on topics listed above or any other relevant topic. Review process will be under double-blind system. Accepted papers will be included in ICCV proceeding.
- Deadline: June 27, 2025
- Submission: OpenReview
- Notification: July 11, 2025
- Camera-Ready Submission: August 11, 2025
- Guidelines: Follow ICCV 2025 Author Kit.
Call for Demos
We invite demo works! Any type of medium (e.g., videos, images, audio, text, etc.) that involve elements of storytelling is welcome.
- Deadline: July 25, 2025
- Submission: Send directly to us! (maximum file size: 2.5GB)
Tentative Schedule
- 08:45-09:00: Welcome words by organizers
- 09:00-09:30: Invited Talk 1
- 09:30-10:00: Invited Talk 2
- 10:00-10:15: Coffee break and Poster Setup
- 10:15-11:15: Poster Session
- 11:15-11:45: Best Paper Talk
- 11:45-12:00: Best Demo Presentation
- 12:00-12:30: Invited Talk 3
- 12:30-13:00: Invited Talk 4
Invited Speakers
- Xiaoqian Shen (KAUST): StoryGPT-V: Large Language Models as Consistent Story Visualizers
- Xianchao Wu (NVIDIA): Novel Post-Training Algorithms for LLMs to Prevent Harmful Contents and Hallucinations
- Weijia Wu (NUS): MovieBench: A Hierarchical Movie Level Dataset for Long Video Generation
- Lin Gu (RIKEN AIP):
Organizers
 Andrew Shin Keio University |
 Yusuke Mori |
 Hiroaki Yamane Fujitsu Laboratories |
 Hana Kopecka |
 Hajime Murai Future University Hakodate |
 Xianchao Wu NVIDIA |
 Lin Gu RIKEN AIP |
 Haitao Yu University of Tsukuba |