Your iPhone Camera, Now With an On‑Device Photo Coach
Adobe’s Project Indigo is an experimental AI photography app for iPhone that combines a computational camera, generative photo editing tools, and built‑in AI photo feedback so users can improve their composition, style, and technical choices while they are still shooting instead of correcting everything later.
The core idea behind Project Indigo is bold: move creative assistance from the editing suite into the camera itself. This iPhone camera AI began as a computational photography app focused on natural, realistic images and desktop‑style processing. Now, with the new AI Playground, it is morphing into a live tutor that critiques your framing, suggests reshoots, and applies generative photo editing in real time.
That shift matters. Instead of hiding AI behind magic “enhance” buttons, Adobe is using Project Indigo to argue that AI belongs at the point of capture, where it can influence how people see and build a scene. It is an opinionated reimagining of what a mobile camera should be: not neutral glass, but a coach that stands over your shoulder.
Inside AI Playground: A Generative Studio in Your Viewfinder
AI Playground is where Project Indigo becomes something more than a smart camera. Arriving in version 1.1, it splits into four sections—Object Editing, Styles, Photo Guidance, and Custom Edit—each powered by generative AI running in the cloud. This is Adobe’s test lab for what an AI‑first camera workflow feels like.
Object Editing removes distractions like people, cars, signs, or fog, and can simulate shallow depth of field by blurring backgrounds. Styles restyles snapshots into artistic looks such as pen‑and‑ink or color‑washed drawings, or swings the other way by transforming paintings and sculptures into photorealistic scenes. Custom Edit opens things further: you type what you want, and the model edits toward that prompt, like adding a lantern on a sunlit ledge.
Technically, this is an AI photography app wrapped around Google’s Nano Banana model, which runs in the cloud, requires an internet connection, and caps edits at 2K resolution. That limitation is not trivial, but it reveals Adobe’s priority: experimenting with creative behavior first, pixel‑peeping later.
From Fixing Mistakes to Changing Habits: AI Photo Feedback in the Moment
The most interesting—and potentially controversial—piece of AI Playground is Photo Guidance, which turns generative AI into an active critic of your photography. After you capture an image, the app can highlight what works and what does not, then recommend both edits and reshoot strategies while you are still on location.
Press one button and a large language model delivers a short critique: good use of light, but cluttered edges, or strong subject, weak leading lines. Press another and it suggests practical reshoots, like shifting your position, changing focal length, or tightening the frame to remove distractions. An AI that tells you how to reshoot would be pointless in a separate editing app; integrated directly into the camera interface, it feels almost like a photo instructor peering through your iPhone camera AI.
This design also nudges users toward better habits. Instead of rescuing sloppy shots later, Indigo teaches you to watch the frame edges, think about background clutter, and consider depth of field as you shoot. If mobile cameras made photography more convenient, AI feedback like this could make it more intentional.
Google’s Model Under the Hood and Why That Matters
Underneath the friendly UI, AI Playground rides on Google’s Nano Banana, an image model that processes Indigo’s edits in the cloud. This choice is notable: Adobe is willing to rely on an outside model for a flagship experiment, instead of defaulting to its own Firefly stack. It signals a future where creative tools are less about single ecosystems and more about whatever model best serves a specific job.
Nano Banana outputs standard dynamic range, so Adobe adds a custom machine‑learning step to push results back toward high dynamic range, while warning that the reconstruction is not always accurate. Edits are capped at 2K pixels and only apply to photos taken inside Project Indigo; imported images cannot enter AI Playground at all. That might frustrate users who want an all‑purpose editor, but again, it underscores the experiment’s focus on shooting behavior rather than bulk post‑processing.
This is not a frictionless magic trick. There are safety filters that can even get testers removed if they trip them repeatedly. The friction is intentional: Adobe wants reliable, constrained experimentation more than viral chaos, at least for now.
A Narrow Beta With Big Implications for Mobile Photography
For now, AI Playground is a small, time‑boxed trial: it is free for only a few percent of Project Indigo users, for a few weeks, with a potential paid version on the horizon if demand is strong. According to Adobe, “our goal is to learn how people use GenAI‑powered editing.”
The team openly frames this as an experiment they may extend, broaden, or follow up with new tests depending on how people respond. They are also exploring device‑specific guidance, recognizing that a dual‑lens phone and a smoothly zooming phone may deserve different advice. That is a subtle but important acknowledgment that AI coaching should respect the hardware you actually carry.
Strategically, Project Indigo fits a wider push to win over phone‑first creators who treat their handset as both camera and studio. If this experiment succeeds, your future iPhone camera AI may not just capture and edit; it may critique, guide, and teach. The question is whether photographers will welcome a generative co‑pilot in their viewfinder—or decide that some mistakes are worth making without an algorithm’s permission.
