oxide-lab/LTX-Video-0.9.5-diffusers

1---2base_model:3- Lightricks/LTX-Video-0.9.54library_name: diffusers5tags:6- ltx-video7- text-to-video8- candle9- rust10- gguf11- oxide-lab12- safetensors13language:14- en15license: other16pipeline_tag: text-to-video17---18 19# LTX-Video in Rust (Candle)20 21This repository provides a high-performance, native Rust implementation of [LTX-Video](https://huggingface.co/Lightricks/LTX-Video) using the [Candle](https://github.com/huggingface/candle) ML framework.22 23## Features24 25- 🦀 **Native Rust**: No Python dependency required for inference.26- 🚀 **Performance**: Optimized for NVIDIA GPUs with **Flash Attention v2** and **cuDNN**.27- 💾 **Memory Efficient**: Supports **GGUF quantization** for T5-XXL text encoder and **VAE tiling/slicing** for generating HD videos on consumer GPUs.28- 🛠 **Flexible**: Easy to use CLI for video generation and library for custom integration.29 30## Quick Start31 32### Installation33 34Ensure you have Rust and the CUDA Toolkit installed, then:35 36```bash37git clone https://github.com/FerrisMind/candle-video38cd candle-video39cargo build --release --features flash-attn,cudnn40```41 42### Video Generation43 44```bash45cargo run --example ltx-video --release -- \46    --local-weights ./models/ltx-video \47    --prompt "A serene mountain lake at sunset, photorealistic, 4k" \48    --width 768 --height 512 --num-frames 97 \49    --steps 3050```51 52## Performance & Memory53 54| Resolution | Frames | VRAM (BF16) | VRAM (VAE Tiling) |55|------------|--------|-------------|-------------------|56| 512x768    | 97     | ~8-13 GB      | ~8-9 GB            |57 58*Note: Using GGUF T5 encoder saves an additional ~8-12GB of VRAM.*59 60## Credits61 62- **Original Model**: [Lightricks/LTX-Video](https://huggingface.co/Lightricks/LTX-Video)63- **Framework**: [HuggingFace Candle](https://github.com/huggingface/candle)64- **T5 v1_1 XXl GGUF and Safetensors**: [city96/LTX-Video-gguf](https://huggingface.co/city96/LTX-Video-gguf) (for GGUF support patterns, T5 XXl GGUF and Safetensors)65 66---67For more details, visit the main [GitHub Repository](https://github.com/FerrisMind/candle-video).