Example inputhover

fps: 25
prompt: "a luminous butterfly flying slowly through a misty enchanted garden at sunset, glowing particles trailing behind their wings, dreamy purple and gold lighting, cinematic motion, soft volumetric light rays, magical natural atmosphere, ultra aesthetic fantasy realism, smooth camera movement, premium AI video thumbnail style"
duration: 8
resolution: "1080p"
aspect_ratio: "16:9"
generate_audio: true

Ltx v2.3 · Text to Video

Video·ltx-v2.3·by LTX

LTX 2.3 Text-to-Video generates 4K AI video clips up to 20 seconds from text prompts with synced audio, vertical framing, and selectable 24 or 48 FPS.

Try it now →

API reference

Runtime (p50): 2m
Estimated price: From $0.08

Call the API

prediction.sh

curl -X POST \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "ltx-2-3-text-to-video",
    "version": "0.0.1",
    "input": {
        "fps": 25,
        "prompt": "a luminous butterfly flying slowly through a misty enchanted garden at sunset, glowing particles trailing behind their wings, dreamy purple and gold lighting, cinematic motion, soft volumetric light rays, magical natural atmosphere, ultra aesthetic fantasy realism, smooth camera movement, premium AI video thumbnail style",
        "duration": 8,
        "resolution": "1080p",
        "aspect_ratio": "16:9",
        "generate_audio": true
    },
    "webhook_url": ""
}' \
  https://api.eachlabs.ai/v1/prediction/

Documentation8 sections

Overview
Ltx v2.3 | Text to Video Overview

The Ltx v2.3 | Text to Video model from LTX transforms text prompts into high-quality video clips, enabling users to generate dynamic visuals from simple descriptions. Part of the ltx-v2.3 family, it leverages an OpenAPI schema optimized for queue-based processing via the LTX provider, making it ideal for scalable video creation workflows. This model stands out for its efficient handling of text-to-video generation through fal-ai/ltx-2.3/text-to-video queue, delivering consistent results for creative and professional applications. Available on each::labs (eachlabs.ai), it simplifies access to advanced AI video synthesis without complex setups. Whether prototyping animations or producing marketing content, Ltx v2.3 | Text to Video streamlines the process from idea to output.
Capabilities
Capabilities
- Generates realistic or stylized videos from natural language text prompts.
- Supports multiple aspect ratios for social media and web optimization.
- Handles motion dynamics like panning, zooming, and object animation.
- Integrates queue-based processing for high-volume LTX text-to-video tasks.
- Produces MP4 outputs ready for editing in standard software.
- Allows seed-based reproducibility for iterative refinements.
- Excels in cinematic effects, lighting, and environmental simulations.
- Compatible with OpenAPI for seamless Ltx v2.3 | Text to Video API calls.
Use cases
Use Cases for Ltx v2.3 | Text to Video

For Content Creators: Produce engaging social media reels. Example: "Vibrant street food market in Tokyo at night, bustling crowd, handheld camera style" – leverages motion dynamics for immersive shorts.

For Marketers: Create product demo videos quickly. Example: "Sleek smartphone rotating 360 degrees on reflective surface, soft spotlight, high detail" – uses precise object animation for promotional assets.

For Developers: Prototype app interfaces with video mockups. Example: "User scrolling through a mobile news feed, smooth transitions, modern UI" – benefits from queue processing for batch testing.

For Designers: Visualize concepts in motion. Example: "Fabric patterns flowing in wind, close-up texture, pastel tones" – highlights environmental simulation for mood boards.

These scenarios showcase Ltx v2.3 | Text to Video strengths on each::labs.
Tips & tricks
Tips and Tricks

Optimize prompts for Ltx v2.3 | Text to Video by specifying style, motion, and camera angles explicitly, such as "a serene mountain landscape at sunset, slow pan right, cinematic lighting." Use negative prompts to avoid artifacts, like "no blur, no distortion." Set a fixed seed value in the API payload for consistent iterations. For Ltx v2.3 | Text to Video API, experiment with duration parameters up to 10 seconds for smoother loops. Combine with image inputs if extending from text-to-image workflows.

Example prompts:
- "A futuristic cityscape with flying cars, neon lights, dynamic zoom in, 4K quality."
- "Ocean waves crashing on rocks, slow motion, realistic physics, golden hour."
- "Abstract geometric shapes dancing, vibrant colors, upbeat rhythm."
These techniques enhance results on each::labs' LTX integration.
Technical spec
Technical Specifications
- Resolution Support: Up to 1024x576 pixels, suitable for standard web and social media videos.
- Max Duration: 5-10 seconds per generation, optimized for short-form content.
- Aspect Ratios: 16:9 (widescreen), 9:16 (vertical), and 1:1 (square).
- Input Formats: Text prompts via JSON payload; supports optional seed for reproducibility.
- Output Formats: MP4 video files with H.264 encoding.
- Processing Time: 30-120 seconds average, depending on queue load and prompt complexity.
- API Access: OpenAPI schema for fal-ai/ltx-2.3/text-to-video queue integration.
These specs make Ltx v2.3 | Text to Video reliable for LTX text-to-video tasks on each::labs.
Things to be aware of
Things to Be Aware Of

Ltx v2.3 | Text to Video may produce minor flickering in complex scenes with rapid motion. Queue delays can extend to minutes during high demand; plan workflows accordingly. Overly abstract prompts often yield inconsistent styles – stick to descriptive language. Common mistake: ignoring aspect ratio, leading to cropped outputs. Requires stable internet for API calls on each::labs. Test with short prompts first to gauge performance.
Key considerations
Key Considerations

Before using Ltx v2.3 | Text to Video, ensure your prompts are detailed yet concise to maximize output quality. It requires an API key from LTX via each::labs, with queue-based processing that may introduce slight delays during peak times. Best for short clips where speed and consistency matter over long-form narratives. Compared to real-time models, it offers better fidelity at the cost of wait time. Monitor credit usage, as video generations consume more resources than static images. Ideal for batch processing in development pipelines.
Limitations
Limitations
Ltx v2.3 | Text to Video caps at 10-second clips, unsuitable for long videos. Struggles with highly detailed human faces or text rendering in motion. No native audio generation; outputs are silent MP4s. Sensitive to prompt ambiguity, potentially causing off-style results. Limited to predefined resolutions and ratios. LTX text-to-video queue may throttle heavy users.
---