And what are we seeing between those two different models? How are they working? Yeah, it's a

great question.

When we introduced cleanup a couple of years ago, it was an on-device model.

It's called a GAN model or a generative adversarial network model.

So that's a smaller model that resides on device.

And it's great at doing specific things and removing kind of smaller blemishes, smaller objects.

But the tradeoff of it being fast and local to the device is that You know, it's not as powerful, so it can't clean up things that are a larger object in the background or a more complex scene.

I mean, this is the fun when just how far you can move it.