With iOS 18.2, Apple added a new app called Image Playground, allowing users to generate fun AI-generated images from a description or reference photo. The result is a bit of mixed bag: sometimes magical, lots of fun, but seriously limited in scope. Here’s how to try this feature for yourself, along with some tips for getting the best results.
Note: Image Playground is built into iOS 18.2 and available on devices compatible with Apple Intelligence – including the iPhone 16 series, iPhone 15 Pro, and iPads with an M1 chip or higher.
To get started, simply open the Image Playground app and tap the + button to begin creating. There are a few ways to influence the generated images, and you can start with any one of them and add more elements as you please.
Describe an image. Simply type anything you can think of into the text box at the bottom of the screen –such as “polar bear wearing a sombrero in space” – and the app will do its best to create it. Image Playground is pretty good at quirky requests like these, so long as they’re not too niche. It struggles with complex prompts, so avoid anything rambling or overly specific.
Use a reference photo. Optionally, you can supply a portrait of someone you know and Image Playground will insert a caricature of them into the scene, using the photo as “inspiration”. Tap the person icon in the bottom right, then pick somebody from the list of people recognized by the Photos app. (If you need to add someone new, you’ll have to do it from Photos > People & Pets).
Customize their look. After choosing a person, you’ll be prompted to choose a starting point by swiping through photos of them to pick one that works well as a reference. Take a look at the generated previews and choose the one that looks the most accurate. We say “most” accurate because they’re all a bit uncanny, and it often gets hair style and color wrong. But it’s not bad at simulating facial features! You can always use the Edit button to come back and choose a new reference photo later.
Alternative references. Maybe you don’t want to generate a real person. If you’d rather make a more generic person, tap the person icon and then Appearance > Edit. Here you can pick a skin tone and a vague facial style, and any images created will adhere to your choices. Alternatively, to reference something specific that isn’t a person, hit the + button in the bottom right and Choose or Take Photo.
Tweak the output. This is where things get fun. Once you’ve got your base, you can add up to five additional modifiers. Choose from ready-made suggestions from the carousel of themes, costumes, accessories, and places. Or simply type out anything you can think of in the Describe an Image box. Whatever you choose or type will be added as an additional prompt and will affect the generated image, which will try to encompass all of your requests into a single picture.
Choose a style. Image Playground is limited to two distinct visual styles, which you can switch between using the + menu in the corner. The default Animation style generates a 3D-rendered look straight from a kids movie, while Illustration creates a bold, colorful, slightly wonky 2D aesthetic. Unlike other AI tools, you won’t get hyper-detailed or artistic results, so it’s best to keep expectations playful.
Keep rolling until you hit gold. When you’ve finished crafting your prompt, tap the floating preview bubble to see your artwork in full. Here, you can swipe through multiple variations before saving the one that works best. Once you’re satisfied with the result, tap Done to save it to your gallery.
Know your limits. While it’s fun and occasionally magical, Image Playground is deliberately limited. The images tend to follow a set formula best suited to making avatars or quirky cartoons. It doesn’t produce realistic images, with Apple presumably keen to avoid being responsible for the proliferation of deepfake imagery and fake news. There’s a long, hidden list of banned topics and words, too. Good luck making the app generate anything outside its cheerful, family-friendly sandbox.
Generate responsibly. Apple’s approach to generative AI stands apart from rivals. All the training data used for Image Playground was – supposedly – properly licensed, addressing concerns about exploiting artists’ work. Additionally, Apple avoids creating photo-realistic images, with the company presumably keen to avoid being responsible for the proliferation of deepfake imagery and fake news. The result is a tool that feels safer and more conscientious, if a little tame. While other generative AI platforms are pushing the boundaries of realism, Apple’s focus remains on privacy and harmless fun.
Experiment! There’s only so much we can write about this feature without you trying it for yourself. So dive in, mess around with prompts, and see what happens. Enjoy!