LTX-2

  • LTX 2.0

byLightricks

Direct cinematic 4K video with synchronized audio and surgical timeline retakes in a single pass

Direct scene
LTX-2

How LTX 2.0 works

Move from written screenplay direction or starting keyframes to complete audio-visual sequences and targeted timeline retakes in three direct steps.

Set your visual prompt or keyframe

Set your visual prompt or keyframe

Enter screenplay-style camera directions or supply a starting image frame at 16:9 aspect ratio.

Choose fidelity and audio mode

Choose fidelity and audio mode

Select Fast mode for rapid ideation or Pro mode for up to 4K resolution at 50 FPS with native audio synthesis.

Refine with surgical retakes

Refine with surgical retakes

Swap specific objects or redub audio tracks by targeting exact timestamps from 0 to 20 seconds.

What LTX 2.0 is good at

Engineered with a 19-billion-parameter Asymmetric Diffusion Transformer, LTX-2 delivers synchronized multi-modal generation, surgical timeline edits, and fluid 4K motion.

Synchronized Dual-Stream Audio

Synchronized Dual-Stream Audio

The 19-billion-parameter Asymmetric DiT synthesizes video frames and matching environmental audio or soundscapes simultaneously in one unified inference pass.

Non-Destructive Scene Retakes

Non-Destructive Scene Retakes

Modify targeted visual elements or redub audio across 6-to-20-second clips using surgical replace modes without re-rendering the entire video.

Up to 4K Resolution at 50 FPS

Up to 4K Resolution at 50 FPS

Render cinematic outputs at 1080p, 1440p, or 2160p with smooth 50 FPS motion and organic film grain that avoids synthetic flatness.

Fast and Pro Inference Tiers

Fast and Pro Inference Tiers

Toggle to Fast mode for high-throughput storyboarding at roughly 30% lower cost, then switch to Pro mode for final high-fidelity master renders.

Made with LTX 2.0

Explore high-resolution motion studies, commercial frames, and atmospheric scenes generated with native audio and organic cinematic grain.

Archival maritime exploration with tactile water turbulence

Archival maritime exploration with tactile water turbulence

Architectural study with atmospheric interior illumination

Architectural study with atmospheric interior illumination

Album art motion visualizer with tactile hardware details

Album art motion visualizer with tactile hardware details

E-commerce product focus with natural fluid dynamics

E-commerce product focus with natural fluid dynamics

Short-form vertical energy captured with dynamic framing

Short-form vertical energy captured with dynamic framing

What people build with LTX 2.0

From film directors blocking complex camera moves to product designers showcasing hardware, see how creators leverage LTX-2 across production workflows.

Cinematic Pre-Visualization and Storyboarding

01

Directors and cinematographers can rapidly block scene beats, camera choreography, and soundscapes from text prompts or keyframe concept art.

Commercial Iteration and Post Retakes

02

Ad agencies and editors can swap featured hero objects, update brand assets, or replace audio tracks on specific timeline timestamps without rebuilding full shots.

High-Fashion Editorial Motion

03

Fashion labels and creative directors can animate editorial lookbooks and runway concepts with fluid textile dynamics and rhythmic background audio.

Industrial Concept Showcases

04

Product designers can present moving vehicle and hardware prototypes at up to 4K resolution at 50 FPS, highlighting mechanical action and materiality.

Documentary and Archival Reenactments

05

Documentarians can produce lifelike historic vignettes with natural film grain, ambient room tone, and period-accurate atmospheric lighting.

Frequently Asked Questions

Yes, LTX-2 and LTX 2.0 refer to the exact same 19-billion-parameter open-weights video generation model developed by Lightricks. Both names describe the dual-stream Asymmetric Diffusion Transformer architecture capable of simultaneous video and audio generation.

The Pro variant delivers maximum visual fidelity, supporting resolutions up to 4K (2160p) and 50 FPS with nuanced textures and lighting, while the Fast variant is optimized for high-velocity concept ideation at roughly 30% lower cost and faster rendering speeds. Use Fast for rapid storyboarding and Pro for final delivery.

Use Retake Video when your clip is mostly successful but contains a localized visual glitch, requires an object swap, or needs audio redubbing. Retake targets specific timestamps between 0 and 20 seconds, modifying only the designated visual or audio layer while preserving continuity across the rest of the clip.

LTX-2 natively synthesizes synchronous ambient audio, environmental noise, and background soundscapes in the same inference pass as the video frames. While it excels at physical Foley effects and atmosphere triggered by descriptive prompt keywords, generating clear, intelligible dialogue remains limited.

LTX-2 accepts text prompts, starting keyframe images, and existing video clips at a 16:9 aspect ratio. It outputs video at 1080p, 1440p, or 2160p (4K) resolutions at either 25 or 50 FPS, with clip durations ranging from 6 to 20 seconds.

Try LTX 2.0 on Fuser

Direct cinematic 4K video with synchronized audio and surgical timeline retakes in a single pass

Direct scene