OpenAI 4o Image Generator Can Spell in Images!

OpenAI has just launched its new image generation tool and it is incredible. Now historically, DALL-E 3 has lagged behind tools like Midjourney and Flux, but now that has completely changed. OpenAI is back on top with their new image generation model. Now DALL-E 3 kind of sucked. It didn't generate images that looked very real and it definitely cannot do text at all. We had to go to Ideogram for that. But now the new OpenAI image generation model basically combined the best of of all of the other models out there and combined it all into one, and you can use it inside of ChatGPT. Best of all, you can use this new image generation model even if you are on the free plan. Now let's dive into it. So to generate a new image, we just have to click down here to View Tools, and we will click on Create Image. And just like all of the other platforms, you just describe with your prompt the exact type of image that you want to generate in any style that you want. This could be hyper-realistic, this could be cartoon, this could a stylized cartoon, this could be anime patterns, you know, the possibilities are endless. There's actually a guide into all of the different styles of images that you can generate with AI. I'll link it below. So let's say in our example, let's do a hyper-realistic image of a dog. And we just press Enter and we wait about a minute or 2 for this image to finish generating. Yes, it does take a lot longer to generate images with the new AI image generation model than historic models, but it is much more accurate. The quality is through the roof. In fact, a friend of mine sent me a WhatsApp message earlier today with an image of his dog, and I literally replied, no way, like, I thought that was an image of your real dog. I couldn't imagine that it was actually AI generated. It's just crazy. So another thing that this new AI generation tool does is that it can actually merge 2 images together. So I have this meme of Julia Roberts being super confused, and I put in this image of myself, and I told it to insert my likeness into the meme and it came back with this. Pretty cool, right? So when you're creating content for social media, uh, now this is a game changer that you can now insert your own likeness into these photos. Another thing that it can do really well now is produce text. So I gave the prompt here to create an image that explains how APIs work and it came back with this. And I told it to create another poster here explaining how photosynthesis works and it came back with this. So again, you can see all of the text are accurate now. And all of the font is consistent as well. So this is a big and major improvement in terms of AI image generation. In addition to this, you can also edit the images right inside of the tool as well. So I had AI generate an image of me in a stylized cartoon holding up a sign that says AI can spell in images now. And you can see it didn't generate the arm quite right. And to edit this arm, I can just click on the Edit tool. Again, it's just right up here and you select that and you just highlight the section it didn't do correctly. And now you explain it to the AI in plain English what it didn't do correctly. The arm is off, for example, needs to be one arm, not one half arms. The arms are distorted. Disjointed. And now I can get the AI to regenerate this image for me. And when— once it has regenerated, you can see the arm is now fixed as well. So this edit button is super, super helpful. Another thing that this image generation model has been getting a lot better at is creating consistent characters. So in the past models, like whenever you put in the same prompt, you get a different character every time, which it makes it hard when you're writing a book or making a movie when you want the characters be consistent. So I put in my prompt here, create a cartoon character Pixar style girl with 2 pigtails, brunette, happy smiling, big large eyes. And it came back with this. You can see she's smiling big. And she's also holding that poster that says AI can spell in images now. And the text is all perfect. And then I asked AI to create a version of her that's crying and sad. And you can see that it has done that. And the characters do look a lot more consistent now. And the last thing that this AI generation tool can do is change the background color. And remove the background as well. I can just tell it in the prompt, remove the background and make it— remove the background and make it transparent. And it can do this as well. Within a minute or 2, I'll be able to get an image back that has the transparent background. So I hope you'll give this tool a try. It's absolutely free to use it on OpenAI. It is now the best AI image generation model available on the marketplace. Now, thank you so much for joining me. For this quick and brief tutorial. If you want to learn more about AI and automations, make sure you sign up for my free course below. I'll walk you through how to build AI agents and use all of the best AI tools available on the marketplace today. And lastly, if you're new here, please make sure you like and subscribe to this channel for weekly AI and automation updates, and I'll see you next time.