AIbaseSource publication

Kunlun Wanwei's Tiangong Workbench integrates Kling 4.0: native 30-second generation per run, with up to 15 reference materials

On October 23, Kunlun Wanwei's AI video creation platform SkyProduction (Tiangong Workbench) announced the integration of the Kling 4.0 video generation model, targeting creation scenarios such as AI short dramas, film and television, brand advertising, and e-commerce marketing; specific features and specifications are subject to the platform's announcement.

On October 23, Kunlun Wanwei's AI video creation platform SkyProduction (Tiangong Workbench) announced the integration of the Kling 4.0 video generation model, targeting creation scenarios such as AI short dramas, film and television, brand advertising, and e-commerce marketing; specific features and specifications are subject to the platform's announcement.

Narrative controllability is the focus of this upgrade. Kling 4.0 supports native generation of videos up to 30 seconds per run with input of up to 10 keyframes, allowing creators to control characters, scenes, and plot progression, extending generation from a single frame to complete storyline narration, while also enhancing understanding of prompts and camera scheduling.

A single task supports up to 15 reference materials, which can mix image, video, subject, and voice inputs for comprehensive control over character appearance, actions, style, and sound. In terms of audiovisual performance, the stability of large motions and complex camera movements has improved, supporting high-quality audio, dual-channel stereo sound, and lip-sync matching, and introducing a 21:9 ultra-wide aspect ratio.

The model will also launch 1080p and 4K 10-bit HDR format output; video continuation can extend a video to up to 2 minutes through multiple continuations—this duration is the extended length after multiple continuations rather than a single native generation—and the launch times of the two features have not been determined. Video editing, creative reference, multilingual speech, text generation, and diverse styles, multi-language text, and Emoji and Logo generation have been upgraded in tandem.

In terms of practical application, continuous narration and keyframe control can reduce shot-connection costs for short drama and film teams, while complex camera movements and multimodal references expand paths such as storyboard pre-visualization; advertising and e-commerce can generate multiple content versions based on product materials to support cross-border marketing. As video generation models advance in quality and controllability, the value of content creation platforms is extending from a single generation capability toward more complete workflows.

Original source

AIbase

Content notes

Original publication and rights belong to the source.

Machine translation · Refer to the original