Adobe's Project Indigo camera app gets generative AI tools, and it's not using Firefly

Adobe's Project Indigo launched last year with a clear pitch: give iPhone photos a more natural, SLR-like look through computational photography. The app offered pro controls, multi-frame super-resolution, and different capture modes that prioritized photographic fidelity over the over-processed look smartphone cameras often produce. That philosophy hasn't changed, but the toolset around it just got a lot more synthetic.

Adobe is now bolting a suite of generative AI features onto Project Indigo, calling it an 'AI Playground.' The company describes the update as an experiment and has included an opt-out button for users who want to stick with the original, unadorned shooting experience. The move signals Adobe testing the waters for how much AI intervention photographers will tolerate in a tool originally marketed on its restraint, not its artificiality.

According to TechCrunch, the new tools include an AI photo critic that uses large language models to analyze your shots and suggest editing improvements. It's a feature that shifts the app from a capture tool into something closer to an always-on photography tutor, giving real-time feedback on composition, exposure, and other variables. Separately, advanced object removal capabilities let users strip out unwanted backgrounds and elements from photos directly within the app.

Adobe's own research blog frames the AI Playground as a sandbox for generative experimentation. The tools let users manipulate images in ways that move beyond traditional editing, leaning on AI generation to fill gaps, replace elements, or reimagine scenes. The Verge reports that the entire suite runs on Google's Nano Banana model, a notable departure from Adobe's usual reliance on its Firefly platform. Adobe hasn't explained the decision, but the choice suggests either a speed-to-market play or a deliberate attempt to test third-party AI integration in a low-stakes experimental app.

Adobe built its reputation in the generative AI space by positioning Firefly as a commercially safe, ethically trained alternative to models like Midjourney and Stable Diffusion. That's why the decision to use Google's Nano Banana model in Project Indigo is so striking. It's not a quiet internal test, it's a public-facing product update that bypasses Adobe's own flagship AI.

This could be a pragmatic move. Nano Banana might be lighter, faster, or better suited to on-device or near-device mobile inference than Firefly's current architecture. It could also reflect a broader Adobe strategy to not lock itself into a single AI vendor, especially as competition between model providers intensifies. Whatever the reasoning, the optics are unusual: an Adobe camera app shipping AI features that don't touch Adobe's AI model. It's the kind of detail that suggests Adobe is more focused on shipping useful tools than on brand consistency around Firefly.

Project Indigo isn't happening in isolation. DPReview notes that generative AI is arriving full force across Adobe's ecosystem, including Lightroom and other creative tools. The company has been steadily weaving AI into its professional applications, from Photoshop's Generative Fill to AI-powered masking in Lightroom. The Indigo update is the consumer-facing, mobile-first expression of the same impulse.

By testing AI features in an experimental app, Adobe gets several advantages. It can gather usage data, learn what photographers actually want from AI tools, and iterate quickly without risking the stability of its flagship products. The opt-out button is a safety valve, letting Adobe gauge adoption rates and user sentiment without forcing anyone into the AI experience. If the features land well, they'll migrate into Lightroom mobile and other Adobe apps. If they don't, Project Indigo goes back to being a computational photography demo.

There's a genuine tension between Project Indigo's original mission and its new AI tools. The app launched as a counterweight to AI-heavy smartphone photography, promising images that looked closer to what a DSLR captures. Adding generative AI features that can fabricate or alter image content pulls in the opposite direction. It's not a contradiction so much as a reflection of where photography is heading: a hybrid of captured reality and AI-assisted creation.

Adobe seems aware of the tension. The opt-out button is a concession to photographers who want the natural look without the AI playground. The company is framing the update as an experiment, not a permanent direction. But the fact that the AI features exist at all in a camera app that was pitched on naturalness says something about where Adobe thinks the market is going. Photographers may want natural images, but they also want the ability to remove a photobomber or get an AI coach's feedback on their composition.

Project Indigo remains an experimental app, and Adobe hasn't committed to bringing its AI features to production products. The company's research blog frames the entire AI Playground as a testbed, and the opt-out mechanism ensures the original app experience stays intact for users who prefer it. The real question is whether Adobe will integrate these features into Lightroom or Photoshop mobile, and if so, whether they'll run on Firefly or continue using third-party models.

The broader signal is that Adobe is accelerating its AI experimentation across every surface it owns. From desktop creative tools to experimental iOS apps, the company is embedding generative AI wherever users are working with images. The fact that it's willing to use a competitor's model to do so suggests that speed and user experience are the priorities, not AI model purity. For photographers, the message is clear: AI editing is coming to every camera app, even the ones that were built to resist it.