 Hello everyone, welcome back another AI video. In this one, we're in Dolly 3, in particular, we're in the Dolly 3 integration with chat GPT. And I'm going to show you how to reverse engineer a prompt that was used to generate an AI image. And not only that, I'm also going to show you how you can reverse engineer a prompt that maybe is one of your own images. So like for example, this one here is a picture of my dog. And I want to create a cartoon character that looks similar to my Paparrou here. So how would I do it? I'm going to show you both methods. And this is an example here of a bulldog that was created using that method. So let's just get right into it. Okay, so the first step here is I'm going to start a new chat. And under new chat, when I hover over GPT 4, you're probably on Dolly 3 at the moment if this is what you're thinking about. But you actually need to switch to default. When you switch to default at the bottom here in the prompt bar, or in the little toolbar here, you're going to see this attach images option. This is very, very important. I'm going to click on this. And this allows me to upload an image. This is not currently integrated with the Dolly 3 functionality that may change at the future. So if you're watching this in the future, hey, hopefully it is. Anyways, here's the current method. Click on this image here. So I'm going to use this image here of the death of stock photography. And when I do that, before I hit enter, so you'll see here is out of the image, I'm going to go ahead and type this in. I'm just going to copy paste it in. But here's the prompt. It will be in the link in the description below. Provide me with a highly detailed prompt for Dolly 3 that could generate an image that looks just like this. Once I've done that, I'm going to go ahead and click on the little purple doodaddy here. And it is going to give me an example. So here we go. Sure, here's a highly detailed prompt for Dolly 3 to potentially recreate a similar image. And here it is, a Pocalyptic steampunk figure wearing a leather hat. Okay, so far so good. And gloves so far so good. Holding a vintage Canon camera, correct? To their face, correct? This is great, actually. Covering everything except their eyes which peer through a skull mask. Okay, look at that. The figure has a scarf draped around their neck, blowing gently in the wind. Holy smokes is this good. And set against a desolate field with distant mountains and a cloudy sky. Damn, that is amazing. So what am I going to do now? I'm going to go ahead and I'm going to just grab this. I'm going to copy it. And now I'm going to create another new chat. I'm going to go back to GPT-4 here, but from switching from default to Dolly 3. And now when I'm in Dolly 3, all I got to do is copy and paste in that prompt and hit Enter and see what it comes up with. Now, while it goes ahead and generates four candidates for me, I'll show you one last thing here. Here's an example. Actually, no, let me show you this one here, part of me. This is a picture of my dog before he went in to get some tuffy surgery. And God, he's a handsome boy. He's a little old now, but he's just the loveliest dog. But I uploaded this, the same method. And this is an image. This is not AI image generated. And I said, hey, provide me with a detailed prompt for Dolly 3 that could generate an image that looks just like this. And it went ahead and it created something similar to this for me. So it's not perfect, but it's got the white and brown and darkish brindle patch between his eyes here on his head. And it's got the nice look. And I mean, it's pretty close. It's got his nice dark brown eyes, et cetera, et cetera. But anyways, that's just an example. So let's go back to this. It hasn't quite finished yet. When it's finished, I'll come back. We'll wrap it up. All right, so welcome back. So it did a pretty good job. It didn't quite get the camera angle correct, but it still did a great job. And you can go ahead and modify these prompts. So for example, if I like, I mean, number, I guess number one is pretty darn close. I could click on that, take that, and then I can just type in here. I can recopy the prompt. Just go back here, copy it and then go, okay, cloudy sky, wide angle shot, let's say. And then re-roll it and then just add in the little bits and pieces that are necessary. But this gets you about 95% of the way there. And that's the whole technique in Dolly 3. Thanks for watching.