Photart is Alamet’s second mobile product. The user uploads or takes a photo, picks an artist’s style or describes what they want, and generative AI turns the photo into a painting in that style.
The foundation we built for Hair AI (sign-in, subscriptions, a generation queue, a multilingual interface) wasn’t specific to hair; it worked for any app that generates images. Rather than writing a new product from scratch, we wanted to carry the same foundation into a completely different experience.
The mobile app is built from a single codebase into two separate store apps; the name, icon, colors and content change based on the app ID. The backend is shared, and each app has its own settings, prompts, storage folder and generation server. In Photart, the photo goes to the model together with a reference image of the chosen artist.
The photo is uploaded to the cloud, and the user’s daily generation allowance is checked.
The chosen artist’s reference image and the prompt are prepared.
The photo and the reference go together to the model on the GPU server.
The composition is preserved; the brushwork and colors take on the painter’s style.
The painting returns to the app, where it can be rated, saved or shared.
Each style is defined by a reference image of an artist. The photo and the reference go to the model together; the composition is preserved while the brushwork changes.
From a cityscape to a portrait, every photo becomes a new painting; the result can be saved or shared, and Pro delivers it in HD.
Hair AI and Photart are built from the same mobile code; the name, icon, color theme and content come from the app ID. A new image-generation app can be built quickly on top of this foundation.
Email sign-up, password changes and in-app account deletion: account management that meets app store requirements.
From the welcome screen to the painting: the user journey in five steps.
The app sends only the photo and the selected painter; the prompt text and the painter’s reference image are kept on the server. The queue passes the photo and the reference together to the GPU server; the result is written to S3 and returned to the app through a signed URL.
External serviceSame codebase as Hair AI
React Native
Expo
TypeScript
Redux Toolkit
i18nextMultilingual interfaceShared API, app-specific settings
Java 17
Spring Boot 3
Spring Security
MySQL
AWS S3From photo to painting
AWSGPU serverSign-in and subscriptions
RevenueCat
Sign in with Apple
