Lightricks

It's a fully open audio/video generation model released with open weights and, crucially, full training code.

It's Lightricks' flagship open source video foundation model, and it's built for real workflows.

This is the LTX 2.0. With this release, they're making the entire model family public: full and distilled weights, the entire training framework, benchmarks, and LoRAs.

And this is optimized for NVIDIA's RTX GPUs.

And yes, it supports full multimodal pipelines.

We have text to video, image to video, video to video, and even audio-conditioned generation.

[static] Thank you to Lightricks for sponsoring this video.

I was extremely excited to help this project in whatever way I can.