LTX 2.3 in ComfyUI - First Look! Workflows, Generation Times, Examples, GGUF's, and more!

Описание: Lightricks recently just released their newest model LTX 2.3 – and ComfyUI already has Day 0 support!

In this video, we’ll provide you two different workflows to run the new LTX 2.3 model inside of ComfyUI, one version using the full 48GB model direct from Lightricks, and the famous Kjai’s FP8 model. Both workflows are exactly the same, the only difference are the models that you are loading into each workflow, we’ve just made the distinction a little easier! We’ll also take a look at a handful of generations generated directly on our hardware (8gb Vram with 128 system ram).

While we only show results from the full model in this video, GGUF versions have been released as of this morning, though we haven’t tested them ourselves. At this time, we’d recommend running these LTX 2.3 workflows inside of Runpod to take advantage of high powered GPU’s for generation.

Overall, we’re satisfied with the results and generation times aren’t horrific for the hardware (between 30 min to 1 hour per 5s clip). This is why we recommend leveraging Runpod, as you’ll be able to generate not only at higher resolutions, but longer durations as well.


🔗 Links:

Patreon Post: https://www.patreon.com/posts/ltx-2-3...

Unsloth LTX 2.3 GGUF's (Brand new): https://huggingface.co/unsloth/LTX-2....

Get started with Runpod here!
https://runpod.io?ref=sf7q5jlu

LinkedIN: https://www.linkedin.com/company/sile...