Kling family
# LTX 2.3 Fast High-performance AI video generation with cinematic control. ## Overview LTX 2.3 Fast represents a significant leap in generative video technology, specifically engineered for marketing teams and creative agencies that require high-fidelity output without compromising on speed. By leveraging advanced diffusion architectures, this model excels at transforming text prompts and static imagery into fluid, high-resolution video content. Whether you are generating short-form social media clips or long-form cinematic sequences, LTX 2.3 Fast provides the stability and visual consistency necessary for professional-grade production. What distinguishes this model from its predecessors is its granular control over camera dynamics and temporal consistency. Users can dictate specific camera movements—such as dolly shots or jib pans—directly within the generation pipeline, reducing the need for post-production stabilization or manual keyframing. Furthermore, the model supports synchronized audio generation, ensuring that the auditory experience matches the visual narrative, which is critical for immersive advertising campaigns. ## Use Cases * **Social Media Ad Campaigns:** Generate high-conversion 9:16 vertical video content for TikTok and Instagram Reels with integrated audio. * **Product Visualization:** Animate static product photography into dynamic, high-resolution 4K showcases to highlight features. * **Storyboarding & Pre-visualization:** Rapidly iterate on commercial concepts by converting script prompts into visual sequences to align stakeholders. * **Brand Storytelling:** Create long-form, 20-second cinematic sequences that maintain consistent character and environment aesthetics. * **Dynamic B-Roll Generation:** Produce custom, high-quality background footage for corporate presentations and digital marketing assets. * **Interactive Content:** Utilize image-to-video interpolation to create seamless transitions between two distinct brand assets. ## Parameters LTX 2.3 Fast offers a robust suite of configuration options to fine-tune your output. Please refer to the Input Parameters section in the Pixloop interface to adjust settings such as frame rate (24-50 FPS), resolution (up to 4K), duration (up to 20 seconds), and specific camera motion effects to match your creative vision. ## Tips for Best Results * **Leverage Image Anchors:** For the most consistent results, provide both a starting and ending frame image to guide the model's interpolation path. * **Optimize for Resolution:** When generating content longer than 10 seconds, ensure your resolution is set to 1080p to maintain stability and processing speed. * **Strategic Camera Motion:** Use 'focus_shift' for product close-ups to create a professional depth-of-field effect that draws the viewer's eye to specific details. * **Prompt Engineering:** Be descriptive with your lighting and texture requirements; the model responds exceptionally well to technical photography terms like 'soft box lighting' or 'cinematic 35mm film grain'. * **Audio Integration:** Keep 'generate_audio' enabled for narrative-driven clips, as the model's synchronized audio significantly increases viewer retention in social feeds. ## About LTX 2.3 Fast is provided by Lightricks and is optimized for the Pixloop platform via the Replicate infrastructure (Cog version 0.16.12). Built upon the latest advancements in generative video research (detailed in the paper at https://arxiv.org/abs/2601.03233), this model is designed to bridge the gap between rapid AI prototyping and high-end commercial video production, offering a versatile toolset for modern creative workflows.
Model input reference derived from preset schema.
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
| fps | select | No | 25 | Frame rate in frames per second |
| image | image | No | null | First frame image for image-to-video generation |
| prompt | string | Yes | null | Text prompt describing the video to generate |
| duration | select | No | 6 | Duration of the video in seconds. Durations longer than 10 seconds are only available with 1080p resolution at 24 or 25 FPS. |
| resolution | select | No | "1080p" | Resolution quality of the generated video. Only 1080p is supported for audio_to_video, retake, and extend tasks. |
| aspect_ratio | select | No | "16:9" | Aspect ratio of the generated video |
| camera_motion | select | No | "none" | Camera motion effect to apply to the generated video. Use 'none' for no camera motion. |
| generate_audio | boolean | No | true | Generate audio for the video. Used for text_to_video and image_to_video tasks. |
| last_frame_image | image | No | null | Last frame image for image-to-video generation. When provided, the video interpolates between the first frame and this last frame. |