# LTX-2: Production-Ready AI Video & Audio Generation Model

## Overview
LTX-2 is a 19B parameter DiT-based AI foundation model for synchronized audio-video generation. It's the first open-source model of its kind, capable of generating high-quality 4K videos with synchronized audio.

## Key Features
- Text-to-Video Generation
- Image-to-Video Animation
- Video-to-Video Transformation
- Audio-to-Video Generation
- Joint Audio-Visual Synthesis
- 4K Resolution Support
- LoRA Fine-tuning
- Multiple Performance Modes (dev, fp8, fp4, distilled)

## Technical Specifications
- Architecture: Diffusion Transformer (DiT)
- Parameters: 19 Billion
- Resolution: Up to 4K with spatial upscaling
- Frame Rate: Enhanced with temporal upscaling
- License: Apache 2.0 (Open Source)

## System Requirements
- Python: ≥3.12
- CUDA: >12.7
- PyTorch: ~2.7
- GPU: NVIDIA with sufficient VRAM

## Model Variants
- ltx-2-19b-dev: Full precision (bf16)
- ltx-2-19b-dev-fp8: FP8 quantization
- ltx-2-19b-dev-fp4: NVFP4 quantization
- ltx-2-19b-distilled: 8 steps, optimized for speed
- Spatial upscaler: x2 resolution enhancement
- Temporal upscaler: x2 frame rate enhancement

## Use Cases
- Content Creation: Social media videos
- Film Production: Pre-visualization and prototyping
- Marketing & Advertising: Promotional videos
- Education & Training: Educational content
- Research & Development: AI video generation experiments
- Game Development: Cinematic cutscenes

## Resources
- GitHub: https://github.com/Lightricks/LTX-2
- HuggingFace: https://huggingface.co/Lightricks/LTX-2
- Live Demo: https://huggingface.co/spaces/Lightricks/ltx-2-distilled
- Research Paper: https://huggingface.co/papers/2601.03233
- Website: https://ltx-2.run/

## Installation
```bash
git clone https://github.com/Lightricks/LTX-2.git
cd LTX-2
uv sync
source .venv/bin/activate
```

## Key Technical Terms
- DiT (Diffusion Transformer)
- Audio-visual synchronization
- Latent upscaling (spatial/temporal)
- LoRA (Low-Rank Adaptation)
- IC-LoRA (Identity-Consistent LoRA)
- Multi-stage/multiscale pipelines
- Classifier-Free Guidance (CFG)

## Author
Lightricks

## Rating
4.8/5 based on 1,250 reviews