LTX
Produce native 4K AI video with LTX-2 on Somake. Features include 20-second clips, multimodal audio, and cinematic motion. Start creating today.
4K video with audio in seconds
Enhanced detail and consistent motion
After enabling this feature, every image or video you generate will not be automatically shared to the community feed
After enabling this feature, every image or video you generate will not be automatically shared to the community feed
No history found
What is LTX-2
LTX-2 is the latest multimodal AI video generation model developed by Lightricks, designed to bridge the gap between real-time experimentation and professional cinematic production. Unlike previous generations of video models that required users to generate visuals first and add sound later, LTX-2 understands audio and video as a single, unified data stream.
This allows creators to generate high-fidelity, motion-consistent video clips with synchronized audio in one seamless process. It is built to serve a wide range of needs, from rapid storyboarding to high-resolution broadcast-ready content.
Key Features
Native 4k 50 FPS
LTX-2 generates true 3840x2160 resolution without relying on upscaling artifacts. By operating at 50 frames per second, it delivers fluid, lifelike motion that avoids the jitter and "shimmering" common in standard 24 FPS AI videos.
Audio & Video
The model creates a single, coherent data stream containing both visual and audio information. Whether itās dialogue or ambient street noise, the sound is generated in real-time to match the visual action, eliminating the need for immediate post-production sound design.
Three Versions: Fast, Pro and Ultra
LTX-2 Fast
Optimized for speed and iteration.
This is the only version capable of generating full 20-second clips, making it the ideal "sketchpad" for testing concepts, blocking scenes, and rapid storyboarding.
LTX-2 Pro
Optimized for visual fidelity.
While limited to shorter clips, it maximizes texture quality, lighting accuracy, and dynamic range. It is the best choice for final marketing assets and high-resolution social content.
LTX-2 Ultra ā Coming soon
Targets high-end VFX and cinema workflows.
It will feature advanced physics simulations and uncompromised rendering quality for studio production.
Model Support Matrix
Model | Resolution | FPS | Duration (seconds) |
LTX-2 Fast | 1080p | 25 | 6, 8, 10, 12, 14, 16, 18, 20 |
1080p | 50 | 6, 8, 10 | |
1440p | 25, 50 | 6, 8, 10 | |
2160p | 25, 50 | 6, 8, 10 | |
LTX-2 Pro | 1080p | 25, 50 | 6, 8, 10 |
1440p | 25, 50 | 6, 8, 10 | |
2160p | 25, 50 | 6, 8, 10 |
Prompt Guide
Communicating with LTX-2 requires a structured approach similar to writing a screenplay. The model performs best when instructions are broken down into logical filmmaking components:
Scene Header: Establish the "where" and "when" immediately (e.g., "Ext. Desert Highway - Day, harsh sunlight").
Subject & Action: Be specific about movement. Instead of "a man runs," try "a man sprints desperately toward the camera, arms pumping."
Camera Directives: Dictate the lens behavior. Use terms like "dolly zoom," "pan right," or "handheld shake" to control the viewer's perspective.
Atmosphere: Define the mood. Keywords like "foggy," "volumetric lighting," or "technicolor" help the model understand the aesthetic tone.
Use Cases
Social Media & Content Creation
For influencers and brand managers, LTX-2 offers a way to bypass the logistical challenges of filming. You can generate thumb-stopping 4K visuals for Reels or TikToks instantly. The integrated audio generation is particularly useful here, as it provides immediate background sound, speeding up the edit-to-publish workflow.
Film Pre-visualization
Directors and cinematographers can use LTX-2 Fast to visualize scripts in real-time. By generating 20-second clips of complex scenes, production teams can "block out" camera moves and actor positioning before stepping onto a set. This saves significant time and budget during pre-production by ensuring the entire crew shares a unified visual reference.
Marketing and Advertising
Advertising agencies can leverage LTX-2 Pro to iterate on campaign concepts rapidly. If a client needs to see a product in five different environmentsāfrom a snowy mountain to a luxury penthouseāLTX-2 can render these variations in high resolution without the cost of location scouting or travel, allowing for faster client approvals.
Why Choose Somake
High-Performance Infrastructure
Access enterprise-grade GPU clusters to render heavy native 4K models instantly, eliminating the hardware bottlenecks of running LTX-2 locally.
Seamless Workflow
Our unified dashboard lets you toggle between "Fast" drafting and "Pro" polishing with a single click, streamlining the technical complexity.
Scalable Cost
Pay only for the compute you use. Somakeās dynamic resource allocation ensures cost-efficiency for both freelance creators and large production studios.
FAQ
Structure prompts like a script: use scene headers (place/time), brief atmospheric descriptions, and clear blocking directions. Order actions sequentially to fit the duration, include dialogue cues in brackets, and add soft closing actions to resolve the movement.
Yes, LTX-2 is a multimodal model designed to generate audio by default. However, you can typically toggle audio generation off in the settings if you only require the visual video track for your project.
Use the "Fast" model when you are in the ideation phase or need longer clips (up to 20 seconds) to tell a story. Switch to "Pro" when you have finalized your concept and need the highest possible texture quality and lighting effects for the final export.
Standard AI videos often run at 24 FPS, which can look jerky or unnatural when the AI struggles with motion coherence. 50 FPS provides twice the visual data, resulting in buttery-smooth movement that looks more like a high-end broadcast camera than a typical computer-generated GIF.
