This competition has two tracks, the consistency track and the creativity track
Video generative models have ignited a creative revolution, crafting vivid worlds from mere prompts, yet the quest for seamless, coherent content remains a pivotal challenge. We propose a full-day workshop at AAAI 2026, titled Consistency in Video Generative Models: From Clip to the Wild, to address this critical issue through three key dimensions: (1) Intra-clip world knowledge consistency, ensuring semantic and physical coherence within a single clip; (2) Inter-clip camera consistency, maintaining seamless content transitions across camera movements; (3) Inter-shot element consistency, preserving the identity and attributes of characters, scenes, and styles across narrative fragments.
本赛将采用初试与复试，初试参赛者使用(Link)进行生成视频，按格式压缩提交后由committee进行评估，前十名进入复试，复试需要提交给committee模型权重，评估代码以及评估环境确认初试成绩并利用内部数据进行二次评估，前三名颁奖。
奖项：一 等奖20 wRMB(刀) 二 三 后续更新
Contest Status: The GenAI Video Creator Contest is an independent public track open to all GenAI video creators.
Video Production: Submissions may be created using any generative AI video model.
Theme & Length: The theme and video length are unrestricted (free to decide).
Domestic Submission: Douyin (抖音): Publish the video and @mention the official Douyin account (Handle TBD).
International Submission: X (Twitter), TikTok, etc.: Publish the video and @mention the official contest account (Handle TBD).
Number of Awards: XXX (To be determined)
Prize Money: XXX (To be determined)
Evaluation Standard: Popularity-Driven: Winners will be selected based on the highest total interaction score upon the contest deadline. The contest strictly prohibits malicious boosting, botting, or other unfair means.
Interaction Weighting: Total Interaction Score = (Views × 30%) + (Likes × 20%) + (Reposts/Shares × 30%) + (Comments × 20%)