Unverified50% confidenceFactExact time
WOVEN is a data source and benchmark that annotates visual transition supervision along three dimensions—scene, action, and reasoning type—with data coming from rollouts of video pretrained generative models.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
Unverified原生视觉能力指模型在预训练阶段就将图像、视频等多模态数据与文本数据联合训练，而非通过后期插件式适配器拼接视觉编码器73% similarUnverified文生视频模型如Sora、Runway Gen-3、Kling本质上是在像素空间中进行概率性采样，基于训练数据统计规律生成画面73% similarUnverified工作流为每个视频概念触发3次独立的图像生成，而非一次，以提供风格基准测试的选样空间70% similarUnverifiedPractice mode uses the canvas to generate simulated video frames and pairs them with a custom event emitter to simulate Director API responses, achieving a fully offline end-to-end testing environment.70% similarUnverified该工具的技术基础是多模态大模型与视频理解技术的协同，包括计算机视觉分析画面构图色彩镜头切换、NLP理解字幕配音文案、时序建模捕捉帧间运动和剪辑节奏70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1013382API
curl https://kongchang.com/api/v1/knowledge/claims/1013382MCP
get_claim(id=1013382)