Kling family
# Kling V3 Omni Professional cinematic video generation with advanced motion and temporal consistency. ## Overview Kling V3 Omni represents a significant leap in multimodal generative video technology, offering marketing teams and creative professionals an all-in-one solution for high-fidelity content production. By integrating advanced temporal consistency algorithms with precise multi-shot control, the model allows for the creation of complex, narrative-driven video sequences that maintain visual coherence from start to finish. Whether you are generating content from scratch or utilizing reference assets to maintain brand identity, Kling V3 Omni provides the flexibility required for professional-grade ad creative. Unlike standard video models, Kling V3 Omni excels in its ability to handle multi-shot sequences through a unified prompt structure. This allows creators to define specific narrative arcs within a single generation request. The model supports both image-to-video and video-to-video workflows, making it an ideal tool for transforming static brand assets into dynamic, engaging social media content or cinematic advertisements. With native audio generation capabilities and robust reference video support, it bridges the gap between AI-generated imagery and professional video production standards. ## Use Cases * **Social Media Ad Campaigns:** Generate high-conversion 1080p video ads from static product photography using the start_image feature. * **Narrative Storyboarding:** Use the multi-shot mode to create cohesive 15-second storyboards for pre-production planning. * **Brand Asset Animation:** Animate static brand logos or key visual elements while maintaining strict temporal consistency. * **Video Editing & Refinement:** Utilize the 'base' reference video mode to perform style transfers or scene enhancements on existing footage. * **Cinematic B-Roll Production:** Generate high-fidelity cinematic background footage for corporate presentations or website hero sections. * **Audio-Visual Sync:** Create short-form content with native audio generation for immersive storytelling without external sound design. ## Parameters Kling V3 Omni offers a comprehensive suite of configuration options to fine-tune your output. Please refer to the input parameters table in the Pixloop interface to adjust your resolution, duration, multi-shot definitions, and reference asset integration. ## Tips for Best Results * **Leverage Start Images:** For the highest level of brand consistency, always provide a high-quality start_image to anchor the model’s initial frame. * **Optimize Multi-Shot Timing:** When using multi-shot mode, ensure your individual shot durations sum exactly to your total duration to avoid generation errors. * **Balance Reference Assets:** When using reference images, keep the count under 4 if you are also using a reference video to ensure the model maintains focus on the primary motion task. * **Use Pro Mode for Final Assets:** Always select the 'pro' mode for final deliverables to ensure 1080p resolution, reserving 'standard' for rapid prototyping and iteration. * **Refine with Video Editing:** If the initial generation is close but requires specific motion adjustments, use the 'base' video reference type to iterate on the existing output. ## About Kling V3 Omni is powered by the Kling Video 3.0 architecture, provided via Replicate (kwaivgi/kling-v3-omni-video). It represents the latest iteration in unified multimodal generation, designed to streamline the workflow between static creative assets and high-motion video output. Built on the 0.18.0 Cog version, it is optimized for high-performance inference within the Pixloop ecosystem.
Model input reference derived from preset schema.
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
| mode | select | No | "pro" | 'standard' generates 720p, 'pro' generates 1080p. |
| prompt | string | Yes | null | Text prompt for video generation. Supports <<<image_1>>>, <<<video_1>>> template references. Max 2500 characters. |
| duration | integer | No | 5 | Video duration in seconds (3-15). Ignored for video editing (base). |
| end_image | image | No | null | Last frame image. Requires start_image. Supports .jpg/.jpeg/.png, max 10MB, min 300px. |
| start_image | image | No | null | First frame image. Supports .jpg/.jpeg/.png, max 10MB, min 300px, aspect ratio 1:2.5 to 2.5:1. |
| aspect_ratio | select | No | "16:9" | Aspect ratio. Required when not using start frame or video editing. |
| multi_prompt | string | No | null | JSON array of shot definitions for multi-shot mode. Each shot: {"prompt": "...", "duration": N}. Max 6 shots, min duration 1s per shot, total must equal duration. Example: [{"prompt":"A cat jumps","duration":3},{"prompt":"It lands","duration":2}] |
| generate_audio | boolean | No | false | Generate native audio. Mutually exclusive with reference video. |
| reference_video | video | No | null | Reference video (.mp4/.mov). Duration 3-10s, resolution 720-2160px per side, max 200MB. |
| reference_images | image | No | [] | Reference images for elements, scenes, or styles. Supports .jpg/.jpeg/.png. Max 7 without video, 4 with video. |
| keep_original_sound | boolean | No | true | Keep original sound from reference video. |
| video_reference_type | select | No | "feature" | How to use reference video: 'feature' for style/camera reference, 'base' for video editing. |