hard to get to but very cool model! Stuff we’ve never been able to do before #ai #chatgpt #google #image #2025 #future

Have you tried the new Google model that can write text inside images and that you can edit images with text? It's quite something. It's also Google, so it is named weirdly, and it is almost impossible to get to, but I will explain. If you go to Google AI Studio, which I think is aistudio.google.com, and if you select from their gigantic drop-down of models, specifically Gemini 2.0 Flash Experimental, and be careful, pick just that one. The other ones don't necessarily do this. Once you have done that, you will get options to tell a story with pictures, options to edit pictures by writing text, and I'm going to give you a few examples that I've actually used. So I uploaded a picture of me, and I said, please put this person in a suit. It put me in a nice gray suit. I looked great. It will also combine images together, so you can take a picture of a croissant, and you can take a picture of a person, and say, please put the croissant in the picture with a person, so the person is holding it, and it will do it. We've never been able to do this before. This is actually new. You can say, draw me a picture of a dragon. I don't want it to be like that. Draw it green, not orange, and it will, instead of redrawing the whole image, actually just recolor it so that it's green. Again, that ability to edit with text, that's new. That's not something we've been able to do with models before, and so I know it's a hassle to get there, but it is super, super interesting when you do. The last thing I want to call out that is really remarkable to me is that it has the ability to formulate short phrases and equations inside the image correctly. You know how AI images tend to have gobbledygook when they actually write text? If you specify the text, it looks pretty dang natural if it's written inside the image. If you don't specify what the text should be and just say, write some text, it will do gobbledygook. But all that being said, the new stuff, like being able to edit with text reliably, telling the dragon to be green instead of orange or whatever, being able to combine images together in a way that's actually useful, like you could have someone hold a product in their hand and just have the picture of the product as a separate thing you bring in. It's an amazing model. It's definitely worth checking out, so check it out. Google Gemini 2.0 Flash Experimental. Please, please, someone fix the names. These are all terrible names.