Mixup is a New Mad Libs-Style App for Creating AI Images Using Photos, Text, and Doodles

The team of former Google employees behind the 3D design platform Rooms at Things, Inc., has introduced its newest creation: a playful AI photo-editing app called Mixup. Available only on iOS for now, the app lets anyone generate new AI images using customizable “recipes,” a Mad Libs-inspired, fill-in-the-blank prompt format designed for photos, written text, and even quick sketches.

With Mixup, you can turn a messy doodle into an elegant Renaissance-style artwork, dress your pet in a quirky Halloween outfit, or experiment with different hairstyles using a selfie. The app isn’t limited to realistic edits — users can go wild with imaginative ideas, such as transforming a friend into an “Italian brainrot” character. Google’s Nano Banana model powers Mixupl. Still, the introduction of its recipe system offers a fresh, more social way to engage with generative AI — almost like turning the process into a shared digital party game.

Image Credits: Mixup

According to Things, Inc. founder and CEO Jason Toff, the key breakthrough behind Nano Banana is its ability to preserve the essence of an uploaded image “in a convincing way that wasn’t creepy.” Toff previously worked on experimental projects at Google and Meta, and also held a product leadership role at Twitter.

A significant highlight of Mixup is the shareability of its user-created recipes.

Image Credits: Mixup

Toff explained that while generative AI is powerful, many users feel lost when confronted with a blank prompt box. Mixup solves that issue by letting people reuse prompts that already worked well — all they need to do is fill in the missing details.

When someone creates a recipe and generates an image, they can publish both to Mixup’s public feed or save them privately. Other users can then tap “Try recipe” to recreate the effect using their own photo, caption, or drawing. A built-in drawing tool makes it easy to produce sketches directly inside the app.

Image Credits: Mixup

By showing both the original image and the recipe together, Mixup aims to address what the team calls the “slot machine” issue—the unpredictability of generative AI outputs. Users can also choose to view a before-and-after comparison if the creator enables the option.

Just like OpenAI’s Sora, Mixup allows users to upload their own photos to appear in generated results. Anyone you follow in the app can also create images featuring your likeness if you enable the option, a feature the team calls “mixables.” While designed for groups of friends, it could also inspire creator-driven communities — though only those comfortable being remixed in unusual ways should participate. Users who prefer privacy can avoid uploading personal images or choose not to follow others.

Moderation is handled through a combination of OpenAI systems and Google’s built-in guardrails in the image model, which block outputs involving violence, sexual content, and other restricted topics.

Image Credits: Mixup

Mixup launches globally on the App Store at midnight on November 21. It is optimized for iOS 26 but works on devices running iOS 18 or later. A web or Android version may come later if the app gains traction.

Image Credits: Mixup

New users receive 100 free credits (worth roughly $4), with each generated image costing nearly 4 cents. Once those credits run out, subscriptions are available with monthly packages of 100, 250, or 500 credits.

The app is currently available for preorder and will require an invite to join when it officially goes live.

This article has been updated to reflect that Mixup launches on November 21, not November 20 as initially mentioned.