Introducing WaveSpeedAI LTX 2 Video Extend on WaveSpeedAI

Extending the Boundaries of Video: LTX-2 Video Extend Arrives on WaveSpeedAI

Great footage doesn’t always last long enough. A breathtaking drone shot ends two seconds before the music cue resolves. A product demo clip falls short of the minimum length for a paid social placement. An establishing shot from a shoot day that can’t be repeated needs to hold for just a few more beats. Until recently, the solutions were either expensive—more shoot time, more crew, more post-production—or unsatisfying, like awkward speed ramps and freeze frames. LTX-2 Video Extend is now available on WaveSpeedAI, giving creators and developers the ability to seamlessly extend existing video clips by up to 20 seconds with prompt-guided AI generation.

What is LTX-2 Video Extend?

LTX-2 Video Extend is built on Lightricks’ acclaimed LTX-2 foundation model, the first open-source AI video model capable of generating synchronized audio and video at native 4K resolution. Originally released by Lightricks and quickly recognized as one of the top-performing video generation models available—ranked 3rd in image-to-video and 4th in text-to-video by Artificial Analysis upon release—the LTX-2 architecture brings its strengths in motion coherence, visual fidelity, and temporal stability to the specific challenge of video extension.

Unlike simple frame interpolation or looping techniques that create visible artifacts and repetitive motion, LTX-2 Video Extend generates genuinely new video frames that continue the motion dynamics, lighting conditions, camera trajectory, and scene composition of your source footage. The result is an extended clip that looks and feels like it was captured in a single, continuous take.

The LTX-2 model is powered by a DiT-based (Diffusion Transformer) architecture with a 14-billion-parameter video stream, giving it exceptional capacity for understanding and reproducing complex visual patterns across extended sequences. This architectural depth is what enables the model to maintain fine details—fabric textures, water motion, facial expressions—through the transition from original to generated footage.

Key Features

Up to 20 seconds of extension — The longest extension range among current video extend models, giving you the flexibility to add brief 1-second transitions or substantial 20-second continuations from a single generation.
Prompt-guided continuation — Describe exactly how the video should evolve. Direct camera movements, character actions, environmental changes, and mood shifts through natural language prompts for precise creative control.
Seamless frame generation — Generated frames continue the motion, lighting, and visual style of your source video without visible seams, jumps, or flickering between original and extended footage.
Bidirectional extension — Extend videos at either the beginning or the end, enabling you to add lead-in footage before a key moment or continue a scene beyond its original capture.
Built-in prompt enhancer — An optional tool that automatically refines your continuation descriptions, helping you get better results even with brief or general prompts.
Simple per-second billing — Transparent pricing at $0.10 per second of extension, so a 5-second extension costs $0.50 and a full 20-second extension costs $2.00. No hidden fees.

Real-World Use Cases

Film and Video Production

Production teams can stretch B-roll footage, establishing shots, and transitional sequences without scheduling additional shoot days. A 3-second aerial shot becomes a 15-second scene-setter that matches your narration timing. Directors can extend reaction shots, hold dramatic pauses, or add breathing room between dialogue beats—all without the cost of reassembling cast and crew.

Every platform has its own ideal video length. TikTok, Instagram Reels, YouTube Shorts, and LinkedIn all perform best at different durations. Rather than shooting multiple versions of the same content, extend a single master clip to hit each platform’s sweet spot. Content creators can transform a punchy 3-second moment into a full 15-second piece that gives audiences more time to engage.

E-Commerce and Product Marketing

Product demonstration videos that are too short leave potential customers without enough information to convert. Extend product reveals, unboxing sequences, and feature demonstrations to give viewers the full picture. Create multiple ad-length variations—6-second bumpers, 15-second pre-rolls, and 30-second showcases—from a single original clip.

Creative and Experimental Projects

Push captured moments in new directions by prompting continuations that go beyond what was originally filmed. Guide a time-lapse sunset into darkness, continue an action sequence into its aftermath, or explore alternative scene progressions. LTX-2 Video Extend turns “what happens next?” from an editorial question into a generative tool.

Digital Signage and Ambient Displays

Extend atmospheric footage—cityscapes, nature scenes, flowing water—into longer loops for retail displays, event backgrounds, or live-streaming overlays. The coherent frame generation ensures loops blend naturally without visible repetition points.

Getting Started on WaveSpeedAI

Using LTX-2 Video Extend on WaveSpeedAI takes just a few steps:

Upload your source video — Provide the clip you want to extend
Write your prompt (optional) — Describe how the video should continue: camera movement, actions, scene evolution
Set duration — Choose how many seconds to extend, from 1 to 20 seconds
Enable prompt enhancer (optional) — Let the built-in tool refine your description
Generate — Submit and download your extended video

For developers, the WaveSpeed API makes integration straightforward:

import wavespeed

output = wavespeed.run(
    "wavespeed-ai/ltx-2/video-extend",
    {
        "video": "https://example.com/your-clip.mp4",
        "prompt": "The camera continues tracking forward through the forest as sunlight filters through the canopy",
        "duration": 10,
    },
)

print(output["outputs"][0])

Pricing

LTX-2 Video Extend uses simple per-second pricing with no minimums or subscriptions:

Duration	Cost
1 second	$0.10
5 seconds	$0.50
10 seconds	$1.00
20 seconds	$2.00

Pro Tips for Best Results

Be specific in your prompts — Describe camera movement direction, subject actions, and environmental changes rather than abstract concepts
Include motion cues — Mention how existing motion should continue: “the bird keeps flying left to right,” “the car accelerates through the turn”
Start short, then go long — Test your prompt with a 1-3 second extension before committing to a full 20-second generation
Use clear source footage — Videos with well-defined motion direction and consistent lighting produce the most coherent extensions

Why WaveSpeedAI?

Running LTX-2 Video Extend on WaveSpeedAI gives you production-ready infrastructure without the overhead of self-hosting a 14-billion-parameter model:

No cold starts — Consistent, fast response times for time-sensitive production workflows
Optimized inference — Purpose-built infrastructure for video generation workloads, delivering results faster than local hardware
Ready-to-use REST API — Integrate into your existing pipeline with minimal effort
Affordable, transparent pricing — Pay only for what you generate at $0.10 per second, with no subscriptions or hidden fees

Start Extending Your Videos Today

LTX-2 Video Extend brings the power of one of the most capable open-source video foundation models to the specific, practical task of making your videos longer—intelligently, seamlessly, and affordably. Whether you need two extra seconds to hit a platform requirement or twenty seconds of new footage to complete a narrative arc, the model delivers coherent, high-quality extensions guided by your creative direction.

Try LTX-2 Video Extend on WaveSpeedAI and turn your short clips into complete stories.