Keling large model – a large video generation model launched by Kuaishou, which can generate a 2-minute video | AI toolset

Ke Ling large model was launched by Kuaishou teamAI video generationThe large model has powerful video creation capabilities and uses a 3D spatio-temporal joint attention mechanism to generate large-scale motion videos that comply with physical laws and simulate real-world characteristics. KeLing supports the generation of high-definition videos with up to 2 minutes and 1080p resolution, and has the function of freely adjusting the aspect ratio. In addition, this AI video tool also combines 3D face and human body reconstruction technology to achieve full drive of expressions and limbs. You can experience vivid AI singing and dancing functions with just a full-body photo.

Functional features of Keling large model

• large amounts of reasonable movement: Using a 3D spatio-temporal joint attention mechanism, it can model complex spatio-temporal motion and generate large-scale motion videos that conform to motion laws.
• Video generation up to 2 minutes long: Thanks to efficient training infrastructure and inference optimization, video content up to 2 minutes long can be generated.
• Simulate physical world properties: Based on the self-developed model architecture, it can simulate the physical characteristics of the real world and generate videos that conform to physical laws.
• Strong concept combination ability: Utilize a deep understanding of text-video semantics and Diffusion Transformer architecture to transform users’ imagination into concrete pictures.
• Movie-level image quality generation: Based on self-developed 3D VAE technology, it can generate movie-level videos with 1080p resolution.
• Supports free video aspect ratios: Using a variable resolution training strategy, it can output a variety of video aspect ratios during the inference process.
• AI-driven innovative gameplay: Combining 3D face and human body reconstruction technology to achieve full expression and body activation, users can experience vivid AI singing and dancing gameplay by uploading full-body photos.
• tusheng video: The Keling model converts static images into 5-second dynamic videos, and users can generate diverse motion effects through text prompts.
• Video continuation: Supports one-click continuation of existing videos, extending by 4.5 seconds each time, and can be continued multiple times, up to 3 minutes, to realize user creativity.

How to experience the Keling large model

The official website of Keling Large Model (kling.kuaishou.com) has been launched and currently supports experience through Kuaiying App. The AI ​​creation function of Kuaiying App has officially opened the invitation test for the Vincent Video function. Creators can apply and experience the latest Vincent Video function of Keling Model.

1. Download Kuaiying App: First, download and install on your phoneKuaiying App。
2. Register/log in account: After opening Kuaiying App, log in with your Kuaishou account.
3. Enter the AI ​​creation function: Find the AI ​​creation function area in Kuaiying App.
4. Application experience: Ke Ling Large Model is in the invitation beta stage. You can only experience this function by applying to become a test user.
5. Upload material: According to the functions provided by Keling Large Model, upload the required materials, such as full-body or half-body photos, for facial and limb driving.
6. Use AI features: Select the specific functions provided by Keling Large Model, such as AI Dance King or AI Sing and Dance, etc., and follow the prompts.
7. Generate video: According to the guidelines of the App, use the Keling large model to generate video content.
8. Edit and export: After generating the video, you can edit the video in the App and then export or share the work.

Application scenarios of Keling large model

• Artistic video creation: Users can use large models to generate creative artistic videos for personal expression or commercial display.
• social media content: Social media users can create attractive video content through Keling large models, increase fan interaction and improve the attractiveness of the content.
• Advertising production: The advertising industry can use Keling large models to quickly generate high-quality advertising videos, improve production efficiency and reduce costs.
• Film and television special effects preview: Film and television production teams can use large models to preview special effects scenes and speed up the pre-production and decision-making process.
• Personalized entertainment: Users can upload their own photos and generate personalized entertainment videos, such as AI singing and dancing videos, through large models.
• news and reports: News organizations can use Keling large models to generate news report videos to quickly convey information.