The LTX-2 19B Extend Video model is a specialized continuity engine designed to expand existing footage while maintaining absolute consistency in character identity, lighting, and art style. By utilizing the core 19B architecture, it ensures that new frames flow naturally from your original source video.
1. Core Extension Controls
Unlike standard generation, this model features unique parameters to manage the transition between the original clip and the new content:
Video Strength: This slider (0 to 1) determines how closely the new frames must adhere to the visual data of the original clip. A higher strength ensures a more seamless match with the input video.
Extend Duration: Controls the length of the newly generated segment added to the end of your clip.
Context Duration: Allows you to specify how much of the preceding video the AI should "read" to understand the motion and narrative flow before generating new frames.
2. Technical Specifications
The Extend Video model shares the same high-performance base as the rest of the LTX-2 family, ensuring professional-grade output:
Resolution & Aspect Ratio: Supports native resolutions up to 4K (2160p) and features a "Match Input" option to ensure the extension perfectly fits the dimensions of the original file.
Native Audio Sync: When the Generate Audio toggle is enabled, the AI creates synchronized soundscapes for the extended portion of the video, matching the environment and actions.
Advanced Motion Controls: You can apply specific Camera Motion presets (Dolly, Jib, or Static) to the extended segment to direct the narrative flow.
3. Pro Workflow: Maintaining Continuity
To get the most out of LTX-2 19B Extend Video, follow these strategic steps:
Input Video: Upload your high-quality base clip into the Input Video slot.
Match Settings: Set Aspect Ratio and Resolution to "Match Input" to avoid any technical shifts between segments.
Refine the Prompt: Use the prompt field to describe only the new action you want to occur in the extended duration.
Balance Strength: If the new segment feels too detached from the original, increase the Video Strength to lock the AI closer to your initial source.
Was this helpful?