yeah oh my gosh absolutely well and it's cool too like we talk a lot about how uh you know there's there's the the open source world of the ltx of of of ai and ltx is right in the middle of all that and then you have the closed source models and it almost feels like the the closed source models are almost like a preview of where the the open source models will be in about like four to five months like it always kind of you see this this jump in quality where all of a sudden like now you're seeing like ltx really hitting this huge quality jump i i i gotta ask like where's the where's the secret sauce is it just the data that like is it the data sets is it are are these new techniques like uh ophir and i were talking briefly about like diffusion fidelity rendering all of these new things? How are you doing these these giant quality jumps? Like a fear? This is probably a question for you, since you're the VP of the applied research? What's the what's the trick?

Yeah.

So, well, it's everything.

But yeah, data is king.

Data is king.

And even I think what people don't understand is thinking about data, you think about videos, right? But it's not just videos.

It's everything around it as well, because these models are conditioned on something, usually text.

in