kling-v2-1-master is Kuaishou Kling V2.1 Master video model, suitable for short-form creation that prioritizes visual quality and visual consistency. It can build shots from text descriptions or start animations from a first-frame image, generating 5-second or 10-second videos. On this platform, you can also use the talking photo feature to turn portrait photos and existing audio into lip-synced videos.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
Input 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 are the available specifications for the corresponding entry point on this platform; talking photos are created through animation and lip synchronization and are not native model audio.
V2.1 Master is positioned with an emphasis on visual quality and consistency, making it suitable for focusing on subject presentation, scene atmosphere, and the overall look of a shot. When creating, use text to clearly define the subject, action, environment, and lighting, then refine descriptions around the same shot, compare different results, and select clips suitable for editing and presentation.
When no ready-made visual material is available, use text-to-video to explore shots; when you already have product images, portraits, or scene images, use first-frame image-to-video to begin from a specified frame. The two methods serve concept exploration and asset animation respectively. Prompts should focus on describing what happens next, rather than merely repeating the image content.
The talking photo feature accepts a portrait photo and recorded audio, first generating photo animation and then completing lip synchronization. Use prompts to add expression or movement requirements, producing the final talking-head video and an intermediate animated video. It is suitable for workflows with existing audio assets that need to quickly create short character-expression clips.
Use a product image as the first frame, describe the desired actions, environmental changes, and display highlights, and create short videos suitable for product introductions or social content. Choose a landscape, portrait, or square aspect ratio based on the placement, and deliver video assets ready for the editing workflow; this model does not use an end-frame image to specify the final composition.
Break a script into individual shots, specify the subject, action, and scene for each segment, and use text-to-video to create visual drafts. 5-second clips can be used for brief action demonstrations, while 10-second clips can accommodate more complete process expression, producing previsualization materials for discussing composition, pacing, and atmosphere rather than completing an entire film in one go.
Prepare a clear front-facing photo of one person and a short audio clip to create videos for introductions, greetings, or character dialogue. The audio is recommended to be no longer than the selected video duration, and prompts can add the desired expressions and actions. The final video is used for talking-head delivery, while intermediate animations can be used to check the photo's dynamic performance.
When choosing between V2 Master and V2.1 Master, consider V2.1 Master as the option more focused on quality and consistency, and compare actual results using the same prompt or first frame. Neither should be understood as supporting all advanced controls simply because of the Master name. If end-frame constraints are needed, consider V2.5 Turbo pro, which supports this capability.
For standard video creation requiring synchronized audio, consider V2.6 pro or V3; for native 4K, consider the corresponding modes in the V3 series. If the task involves multi-image references, reference videos, or editing existing videos, choose O1 or V3 Omni. For short talking-head videos using existing photos and audio, you can continue using this model's talking photo feature.
For text prompts, choose text2video; for photo animation, choose image2video and provide start_image_url. Clearly describe the person's actions, the product's appearance, and the camera intent separately.
Explicitly specify model=kling-v2-1-master for /kling/videos, use this model's single mode, and start with 5 seconds; choose 10 seconds when a complete action is needed. Image-to-video provides only the first frame.
Save the asynchronous task_id and retrieve the video through queries or callbacks; compare it with the original image to check the person, clothing, and background. Standard video generation does not include native audio, so add sound in post-production; photo lip-sync requires a separate workflow using a photo and existing audio.
Use a photo of a person standing by a window as the first frame. The person turns toward the camera and smiles, while the camera slowly moves closer, preserving the clothing and indoor layout.
Check the face, clothing, and motion; this model uses the first frame and does not support workflows extended through multi-image lists, end frames, or native audio.
V2.1 Master is better suited as an option focused on visual quality and consistency. It is recommended to compare the actual results of both using the same prompt or first frame, rather than looking only at the version names. Both are used here for 5- or 10-second short clips, which does not imply additional control capabilities.
This model uses first-frame image-to-video: submit start_image_url and describe the subsequent actions with a prompt; end_image_url last-frame constraints are not supported. If you must specify the ending image, you can choose a model that supports first and last frames, such as V2.5 Turbo pro, and prepare assets according to the corresponding creation workflow.
Regular videos generate dynamic visuals from text or a first-frame image; this model does not natively generate synchronized audio. Talking photos, however, require a photo and existing audio, generating a spoken video through animation and lip synchronization; the audio is an input asset, not a voice automatically created by the model.
Call POST /kling/videos, explicitly set model to kling-v2-1-master, choose text2video or image2video, and submit a prompt with a duration of 5 or 10 seconds. Image-to-video also requires a first-frame image link; after completion, retrieve the generated video through video_url.
You can set async=true, first save the returned task_id, then query task progress; you can also configure callback_url to receive completion notifications. After obtaining results, associate video links and statuses by task ID. Photo narration can also use the asynchronous approach and returns the final video along with intermediate animation links.
Choose a development approach or application suited to this model and follow its setup guide.
↗Python SDK Integration Guide
SDK installation, authentication and examples↗TypeScript SDK Integration Guide
SDK installation, authentication and examples
↗This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs