Model for rich people.
The System Requirements are unacceptable. You haven't made any optimization efforts, after all these months. This team thinks we're in Dubai or something. My pc is sufficient for most AI models, but obviously not for LTX. I can't even test low resolution. Immediate failure.
I run this model (fp8) on my PC: Ryzen 9 7900X3D, 64 GB DDR5, RTX 4070ti. It takes about 180s to create a 5s 720p@24 clip. And ... I'm not in Dubai))
The System Requirements are unacceptable. You haven't made any optimization efforts, after all these months. This team thinks we're in Dubai or something. My pc is sufficient for most AI models, but obviously not for LTX. I can't even test low resolution. Immediate failure.
You need to get better hardware for this sort of model. I recommend a real AI-capable system like an AMD Strix Halo system for example (if you don't want to take 2min to look for quantized models to fit lesser builds).
I run it on a rusty old 3090, and no problems at all, it works great ;-)
And super grateful to LTX team for releasing this model. (PS, I'm also not in Dubai;-)
You can also try Kijais nodes that might run a little easier on low end computer, and even has support for GGUF
https://huggingface.co/Kijai/LTXV2_comfy