kling-v1-6 is a video generation model in Kuaishou's Kling V1 series, suitable for turning text concepts or static images into short shots, then continuing creation from existing clips. It provides text-to-video, image-to-video, and video extension; pro mode can also use first and last frames to constrain the start and end of the visuals. When characters need to speak, photos and existing audio can be combined for animation and lip-syncing.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
api.acedata.cloudkling-v1-6Input parameters and result formats vary by service. Use the public API for this model and follow its guide for generation, task retrieval and editing operations.
The above durations, modes, and operation scope correspond to this platform's kling-v1-6 entry point. Photo lip-syncing is a combined creation feature and is not equivalent to the model's native audio.
When there is no ready-made image, use prompts to describe the subject, environment, and actions to generate a shot; when product images, character images, or scene images already exist, use the first frame to fix the starting point, then describe the desired changes. These two approaches are suitable for concept exploration and asset re-creation respectively, allowing short videos to begin with clear visual goals.
Image-to-video in pro mode can specify both the first and last frames to arrange changes in character poses, before-and-after product display states, or scene transitions. The last frame must work together with the first frame and cannot be used alone. It constrains the starting and ending images of the shot, not a frame-by-frame trajectory, and cannot replace dedicated camera movement control parameters.
Video extension can be used in std and pro modes: retain the generated result's video_id, then submit prompts for the next actions or plot. This lets you establish short shots first and then continue with subsequent content, without having to rebuild the scene from text each time. After the task is complete, retrieve the result through the video link for further editing.
Input a product first-frame image, describe the display actions, background atmosphere, and visual changes, and generate short footage suitable for editing. Portrait format can be used for mobile displays, while landscape format can be used for presentation visuals; if there is already a clear final composition, add a last frame in pro, then supplement the finished video with subtitles, music, and brand information.
Write a storyboard as a prompt with a clear subject, actions, and environment, generate a single shot first, then use video extension to try subsequent plot developments. The deliverables are watchable, editable video clips, suitable for discussing pacing and visual direction before formal filming or production, rather than directly replacing a complete multi-shot editing project.
Prepare a clear front-facing photo of one person and recorded audio, select kling-v1-6 in the talking photo feature, and add action or expression requirements. A single task completes photo animation and audio lip-syncing, returning the final talking-head video and intermediate animation clips, suitable for short explanations, character greetings, or content previews.
If the task is image animation, short-shot generation, or continuation of an existing clip, V1.6 offers more direct operation combinations; choose pro when you need to control the ending frame. Compared with kling-v1, you cannot carry over the older version's camera movement parameters to V1.6: kling-v1 camera movement has a 5-second condition, while V1.6 does not support camera_control, so choose according to actual control needs.
If you need to generate visuals and sound simultaneously, consider kling-v2-6 pro with audio support; if you need flexible durations of 3–15 seconds or 4k mode, consider kling-v3. V1.6 is better suited to short-video workflows that clearly use 5 or 10 seconds, with audio produced separately. External audio lip sync and native audio are different tasks and should not be mixed.
Choose text2video to describe the scene, or choose image2video and provide start_image_url; write requirements for preserving people, products, or backgrounds into the prompt.
Specify model=kling-v1-6 to /kling/videos, start with 5 seconds, and choose std/pro based on visual needs. Only pro image-to-video can use a final frame; std uses only the first frame.
Save the task_id asynchronously and obtain the video through task queries or callbacks; check the subject, final frame, and motion continuity. This model does not generate audio, so handle voiceover and music in post-production when needed; retain the video ID when continuing extensions.
The first frame shows a closed gift box, the final frame shows an open gift box, the product position inside the box is clear, the camera and tabletop remain consistent, and the box opens slowly.
First and final frames work together only under pro; check the geometry of the box lid and product. Do not request native audio for regular generation.
Explicitly specify model=kling-v1-6 in video generation or talking photo requests, rather than relying on the default selection after omission. For standard videos, you must also select text2video, image2video, or extend, and provide a prompt, start-frame image, or existing video ID according to the task.
The video generation interface supports 5-second or 10-second options, and talking photos also offer these two durations. To continue creating, you can use extend to submit an existing video_id and a follow-up prompt; extension continues an existing clip and does not mean generating a video of any length in a single request.
kling-v1-6 std does not support end-frame constraints. When a specific ending image is required, use pro image2video and provide both start_image_url and end_image_url; submitting only an end frame cannot complete this type of task, as a start frame is still required input.
No. V1.6 does not support native audio. Talking photos use the audio you provide: the photo is animated first, then lip-synced to the audio. You need to submit image_url and audio_url; the prompt is mainly for actions or expressions during the animation stage, not for generating speech content.
You can use async=true to obtain a task_id and then query the task status, or configure callback_url to receive the completed result. Applications should distinguish between a submitted task and a completed video, and read video_url after completion; if you need to extend it later, you should also save video_id and cannot use task_id as a substitute.
Choose a development approach or application suited to this model and follow its setup guide.
Python SDK Integration Guide
SDK installation, authentication and examplesTypeScript SDK Integration Guide
SDK installation, authentication and examples