kling-video

What it does

Plans, prompts, generates, references, edits, motion-control, and quality-review processes for Kling AI videos using Kling VIDEO 3.0/Omni tooling. It emphasizes explicit surface selection (direct API or Kling web/app), task/data inputs (scene, reference media, duration, aspect ratio, etc.), and appropriate model choices (kling-v3, kling-v3-omni, kling-3.0-turbo) for coherent production workflows. It includes detailed pricing, input limits, and governance around consent, references, and motion-control requirements.

How it works

• Guides the user to collect delivery surface, geography, task, input media, duration, aspect ratio, resolution, audio/dialogue, shot structure, audience, rights/consent, privacy needs, variant count, and approved spend before producing.
• Recommends selecting the appropriate Kling model based on input needs: kling-v3 for prompt/start-image/start-end frames; kling-v3-omni for reference-driven or multi-input editing; kling-3.0-turbo for faster, scalable tasks.
• Describes asset and reference limits for 3.0 Omni: up to seven images, one reference video (3–10s), and total asset constraints; specifies how Element bindings and motion-control relate to consent and quality. Documents motion-control consent requirements and safety constraints.
• Outlines pricing by resolution and model, with explicit per-second rates and how to compute costs for a given duration, along with pre-billing steps (binding model/task, duration, references, etc.).
• Provides concrete prompt-building guidance for one-shot and multi-shot workflows, including example schemas for multi-shot prompts and instructions to verify live 3.0 task schemas.
• Emphasizes fail-safe API usage: asynchronous direct API, non-durable storage, idempotency notes, and a risk-aware approach to retries and reconciliation.

When to use it

• Use Kling VIDEO 3.0 or Omni when the task requires text-to-video, image-to-video, reference-driven generation, or multi-shot structures with native audio.
• Use kling-v3 when scene control is prompt-driven or start-end frames suffice; use kling-v3-omni when references or editing inputs are central; use kling-3.0-turbo when the task page confirms compatibility and performance is a priority.
• Use Motion Control for transferring performance from driving video to a character with explicit driver/performer consent.

What it can touch

• Direct API endpoint: https://api-singapore.klingai.com for server workflows with API-key authentication.
• Kling web/app for interactive creation, Elements library, and UI features.
• Models: kling-v3, kling-v3-omni, kling-3.0-turbo; Motion Control feature for performance transfer.
• It references the need to use a server-side KLING_API_KEY and to manage input media and references according to the specified limits.

Caveats

• Surface, model availability, pricing, policies, and limits are volatile and should be re-checked against current Kling/Kuaishou sources as of the latest date indicated.
• Some provider claims about consistency or quality are not guarantees; they are described as provider claims.
• Motion Control requires explicit driving-video and performer consent; inappropriate use or consent gaps are disallowed.
• The example code demonstrates an API interaction pattern but is not a guarantee of behavior and should be validated against live endpoints and current schemas.