The ex-Googler workforce behind the 3D design app Rooms from Things, Inc. is out with its newest venture: a playful AI picture editor referred to as Mixup. The iOS-only app lets anybody create new AI-generated pictures utilizing “recipes,” that are like Mad Lib-style, fill-in-the-blank prompts on your pictures, texts, or sketches.
For example, you could possibly ask Mixup to show your scribbled sketch into a good looking Renaissance portray, or to reimagine your pet in a humorous Halloween costume. You would use a selfie to see what you’d seem like with a distinct coiffure, or create one thing even sillier — like envisioning a buddy reworked into the type of Italian brainrot.
The app is constructed on prime of Google’s Nano Banana, however the “recipe” format opens up a brand new method to work together with the mannequin — turning a generative instrument into an internet get together recreation.
“The factor that [Nano Banana] did that no mannequin did earlier than was that it might take your picture and preserve it in a convincing manner that wasn’t creepy,” stated Issues, Inc. founder and CEO Jason Toff, whose background contains engaged on experimental apps inside Large Tech corporations like Google and Meta, in addition to time spent managing product at Twitter.
However what makes Mixup significantly enjoyable to make use of is that the app’s “recipes” — the user-generated AI prompts — are shareable.
“Generative AI is so highly effective, and but more often than not you go to those instruments and it’s like, right here’s your textual content field — provide you with one thing inventive. And what do you write?” Toff stated, chatting with the shortcomings his workforce noticed with the present AI panorama.
“And so, moderately than having to be inventive and take into consideration what to create, you see one thing that labored, and you’ll simply fill within the blanks,” he stated.
After customers create a brand new immediate in Mixup, they’ll select to publish it together with the ensuing picture to a public feed or simply obtain it for private use. From the feed, different customers can view the picture and faucet the button subsequent to it to “Attempt recipe.” This permits anybody on the app to reuse the recipe to generate a picture utilizing their very own picture, textual content, or doodles. (The latter may be made by way of a easy, in-app drawing characteristic.)
Seeing a photograph alongside the recipe that created it will probably additionally assist deal with the unpredictable nature of generative AI pictures, the workforce believes.
“One other drawback [with generative AI] is what we had been referring to internally as a ‘slot machine’ drawback, the place you push the button, one thing comes out, you push it once more, one thing completely different comes out, and also you don’t really feel like you will have management over the output,” Toff defined.
In Mixup, nevertheless, customers can see each the picture and the immediate that created it in a single place, giving them an thought of what their output would seem like. They will additionally toggle a button to see each the earlier than and after picture, if the creator opts to depart this setting on.
Plus, just like OpenAI’s AI video app Sora, customers can add their very own pictures to Mixup to make use of of their AI pictures. When you select to take action, then any individual you comply with within the app may create AI pictures along with your likeness — a characteristic it calls “mixables.”
The corporate imagines that teams of associates would comply with one another to reap the benefits of the characteristic, however a creator class might additionally probably emerge on the platform — in the event that they don’t thoughts seeing themselves mashed up in weird methods. (In fact, in case you don’t need your picture on the market, both don’t add it or don’t comply with anybody.)
The app additionally makes use of OpenAI know-how to deal with a number of the anticipated moderation considerations round AI imagery, however Toff admits that Mixup additionally leans closely on Google’s built-in controls inside its picture mannequin to limit issues like sexual content material or violence.
At launch, Mixup is optimized for iOS 26 however is supported on iOS 18 and up. If it takes off, an internet model or Android app could also be added later.
Free customers obtain 100 credit, which equates to $4. In the meantime, pictures price almost 4 cents to supply. When the credit run out, customers can subscribe to completely different tiers providing 100, 250, or 500 credit per 30 days.
The app launches at midnight on November 21 on the App Retailer globally, however would require an invitation to get in. TechCrunch readers can use the code TCHCRH (till it runs out) to get in. Mixup is on the market for preorder forward of its launch.
This submit was up to date after publication to notice the app launches Nov. 21, not Nov. 20 as beforehand acknowledged.
