Apple Image Playground gains photorealistic AI image generation to compete with ChatGPT and Gemini

Apple's Image Playground doesn't suck anymore

Have you ever used the Image Playground app on your iPhone? Probably not -- and you might not even know that your phone has a built-in AI image generation tool. That's because until now, Image Playground has kind of sucked, generating AI imagery that simply isn't as high-quality as what Apple's competitors can make. At WWDC 2026 on Monday, Apple announced some necessary improvements to Image Playground. While we haven't yet gotten to use the app, Apple's presentation makes it seem like the company's overall AI overhaul will make Apple Intelligence-powered apps, like Image Playground, perform a lot better. "From stunning nature scenes to fun images using multiple people from your Photos library, you can also transform your photos into endless styles just by describing what you want in natural language, and with private cloud compute, your photos are never stored or shared, even with Apple," said Apple Senior Director Leslie Ikemoto in a WWDC presentation. As an example, Ikemoto explained that if you're throwing a birthday party for your friend, you can design an invitation by using Image Playground to create an image of your friend holding a cake. Then, you can use natural language prompting to add candles to the cake or change your friend's outfit. "You also have more ways to use the images you create, like choosing the right dimensions for what you're working on, including a landscape image for your small business's website, or a portrait image for your flyer," Ikemoto added. Since Image Playground is integrated across your device, you can also use it to generate lock screens, iMessage backgrounds, contact posters, and more. AI art is generally pretty corny, but at least Apple won't use your private photos for AI training, unlike many of its competitors.

Apple's All-New Image Playground Promises More Than Cartoons

Today at WWDC 2026, Apple unveiled a wide range of new software for its product ecosystem, including macOS 27 Golden Gate, iOS 27, and iPadOS 27. A significant part of all the new software updates is an upgraded Apple Intelligence system, which includes a revamped Image Playground, Apple's generative AI for images. The big new feature in the all-new Image Playground is its improved photorealism. Initial iterations of Image Playground focused on cartoonish, emoji-based generative AI, offering little by way of realism. "Image Playground offers new powerful ways for users to bring their imagination to life," Apple says. "They can create high-quality images in virtually any style, now including photorealistic, thanks to a new generative model that runs on Private Cloud Compute." The goal here is clear, Apple wants its generative AI technology to be much more serious and grown-up. A key feature of the all-new Image Playground is that users can easily and quickly edit their creations without having to start from scratch. Users can describe the changes they want the AI to make, or tap, circle, or brush areas they want to revise. There are also new ways for Apple users to actually take advantage of Image Playground. It still works in the expected apps like Messages, but Image Playground can also be used to generate Lock Screen wallpapers, Contact Posters, and more. When Image Playground was unveiled in June 2024, it arrived with just three basic styles: animation, illustration, and sketch. One thing that hasn't changed with the new Image Playground is Apple's commitment to transparency about AI-generated content. All generated images automatically include a hidden SynthID watermark that identifies them as AI-generated, Apple promises. This is also true of actual photos edited with Apple Intelligence features in the upgraded Photos app. "Image Playground now lets you make high-quality images in pretty much any style you want, including photorealistic, thanks to our new generative model that runs on Private Cloud Compute," says Leslie Ikemoto, Director, Input Experience at Apple. "This is a major upgrade for image generation across our platforms, giving you a more powerful way to bring your imagination to life." The new-and-improved Image Playground will be available in macOS 27, iOS 27, and iPadOS 27.

Apple's Image Playground Just Caught Up to ChatGPT and Gemini

Image Playground's generations are a lot more impressive this year. Credit: David Nield/Lifehacker Among all the other artificial intelligence upgrades Apple is rolling out for us this year, you'll find that there's a significant jump forward in Image Playground's AI image generation abilities. Before now, the app's outputs were rather limited in terms of size, style, and possible prompts. It was very much AI images for beginners, with the results mostly basic and generic. With iOS 27, iPadOS 27, and macOS 27, that's changing. You can try the developer betas now (though you probably shouldn't; the public betas are arriving in July), and Apple says the full releases will appear later in the year (most likely around September). Image generation is something both ChatGPT and Gemini have been steadily improving, to the point where some generations are difficult to distinguish from real photos. So how does Apple's new and improved Image Playground compare? Here's what Image Playground can do now Open up Image Playground on the newest versions of iOS, iPadOS, or macOS, and you'll see there are several new capabilities. First, you can ask for photorealistic images in the prompt box, as well as the sketches and illustrations previously possible: Ask for a photo of an English meadow or a towering temple, and Image Playground will oblige. There's more control over your images, too. You can submit a photo as part of a starting prompt, and you can choose between square, portrait, and landscape orientations for your picture -- these options weren't available before. If you want to transport your pet dog to the jungle, that's possible now. Then there are clever editing tools, similar to those in Google's Nano Banana AI model, that let you change specific parts of a generated image without having to render it entirely again. You can use a follow-up prompt to request changes, and even highlight the part of the picture you want to edit. You can now change the color of objects, remove objects altogether, change the weather of a scene, whatever you can think of, really. Based on my testing, it all works well, and produces results that look impressive and consistent. I managed to put a cartoon-style cat in a street, and then change the cat's color without affecting the background. Apple says that everything created through Image Playground will have the same SynthID watermark as generations from Google Gemini and ChatGPT, and we know because of Private Cloud Compute that no user images will be stored or accessed by Apple, or used to train any of its models. Finally, you can do more with these images too -- setting them as Contact Posters or wallpaper for the lock screen, for example. You can find more AI image tools in the Photos app, where you can apply edits similar to those possible in Image Playground to whatever's in your photo library. Image Playground versus the competition Image Playground is much better -- no doubt thanks to a boost from Gemini -- but even still, the images aren't quite up to the very high bar that Gemini and ChatGPT have set now. You can see below how my request for "a photorealistic image of a small, ancient-looking spaceship floating between the stars, with an Earth-like planet behind it" was interpreted by Image Playground (left), Gemini (center), and ChatGPT (right). Apple does okay, but to my eye, Gemini and ChatGPT generated results that are more immersive and detailed -- like something you'd see in an actual science fiction film. There's more detail and more imagination, although to Apple's credit, Image Playground rendered the fastest. For the next challenge, I tried asking these AI tools to "move my cuddly toy from my floor to a pebbly beach, with the tide lapping at its edges." Again, all these attempts are good, but Gemini (center) and ChatGPT (right) add extra layers of verisimilitude in terms of color, angle, and texture (though Gemini seems to have created two shorelines). I asked all three models to remove the toy and just leave the beach, and they all managed it more or less perfectly. These are Photoshop-level edits that used to take me hours, but can now be completed in seconds. It's truly impressive. Gemini and ChatGPT still have the edge in terms of quality, but Image Playground comes built into billions of Apple devices. It's now good enough that many users will likely choose not to switch to something else when they need to generate an image with AI, which might make all the difference.