It's triple the power of Sora! Domestic AI-generated video content has arrived, with the potential to lead the way in short videos, marketing, animation, and other fields
Recently, KUAISHOU launched the "Ke Ling" video generation large model, which can transform static images into vivid 5-second videos based on text content, and can be extended with one click and continuously extended multiple times. It is claimed that the model's effect is close to Sora, capable of generating videos up to 3 minutes long, three times that of Sora, indicating that the gap in the AI text-to-video field between domestic and foreign markets is narrowing. Due to high costs, the model will prioritize providing services to enterprises and groups. Debon Securities predicts that the short video, marketing, and animation fields are expected to be the first to land, and recommends paying attention to related companies such as KUAISHOU and Tencent Holdings
Recently, Kuaishou's "Keling" video generation large model announced the launch of the image-to-video function, which supports transforming static images into vivid 5-second videos based on different text content. At the same time, it supports one-click continuation and multiple continuations of the generated videos, with each continuation extending the video by about 5 seconds, and the longest video that can be generated is about 3 minutes.
CITIC Securities mentioned that the effect of "Keling" is close to Sora, and the longest video that can be generated is about 3 minutes, which is 3 times that of Sora. This indicates that the domestic and international gap in the field of AI-generated videos based on text is narrowing, and the future prospects of domestic AI-generated videos are promising! As of June 20th, the number of applicants in the internal test of "Keling" has exceeded 140,000. Due to the high cost of AI computing power, high-quality AI video generation large models with commercial capabilities are difficult to promote to C-end users in the short term. It is expected that they will prioritize providing services to enterprises and groups, and video generation in the C-end field is still in the stage of entertainment applications and observation testing. AI video generation tools enable creators to easily create, edit, and adjust content, or reshape the value composition of the industrial chain, such as the value of IP and creativity relative to shooting, editing, and production, which may be enhanced. Focus on companies related to AI video generation large models, as well as companies in application areas with high-quality IP or IP management capabilities in the film, game, and marketing industries with unique creative capabilities.
Debon Securities analysis stated that high-quality data may be the core, and areas such as short videos/marketing/animations are expected to be implemented first. Recommendations include: 1) Targets related to video large models and tools: Kuaishou, Tencent Holdings, Meitu Inc., Wanxing Technology, Kunlun Wanwei, etc. 2) AI audio and video related applications: Shanghai Film, Bona Film, Huace Film & TV, Alpha Group, Tom Cat, Shengtian Network, Chinese Online, Wan Da Film, Qimu Shares, Kaiying Network, Enlight Media, etc
