Text to Image

I'm curious because so many people are focused on text to image or text to whatever kind of generation. It seems like you all have experimented with some other kind of inputs, like image input, etc. Do you think it'll be primarily a kind of text input or do you see it kind of morphing into something else? That's a great question. I think text is a very large language models in particular. We're exploring and just scratching the surface of a very powerful input mechanism.