Adobe tests generative AI photo editing in Project Indigo app
Adobe has updated its experimental Project Indigo camera app with generative AI tools that edit and critique photos, marking a strategic pivot for a company historically built on photographic realism.
Adobe has released a major update for its experimental iPhone camera app, Project Indigo, introducing a suite of generative AI tools. Version 1.1 adds the "AI Playground," a bundle of features that can restyle images, strip out unwanted objects, and critique a user's composition in real time.
The update marks a distinct shift for a software company that originally sold photographers on natural realism and minimal processing. By introducing tools that can invent backgrounds and rebuild scenes entirely, Adobe is actively navigating the blurred boundaries of digital imagery. As one tech publication put it: "Welcome to the ‘what is a photo’ debate, Adobe."
The most novel addition is the automated critique function, designed by Marc Levoy, the engineer behind Google's Pixel computational photography. While the photographer is still standing in front of their subject, a large language model evaluates framing, lighting, colour, and emotional impact. The app then offers specific advice on how to reshoot or edit the image.
Rather than relying on complex text prompts, the new editing tools operate primarily through engineered buttons to reduce user ambiguity. Toggles automatically trigger pre-written prompts to remove background clutter, vehicles, or wires, while style buttons can re-render shots in pen and ink or golden-hour light. A custom text box remains available for edits the engineered buttons miss.
In a notable technical choice, these generative features run on Google’s Nano Banana model rather than Adobe's proprietary Firefly software. Because Nano Banana operates in the cloud, the app requires a constant internet connection and caps image edits at roughly 2K pixels to maintain processing speeds. Adobe noted it may swap in its own models later.
Adobe is moving cautiously to address the ethical risks of AI-generated imagery, utilizing invisible watermarks and tagging edited photos through its Content Authenticity Initiative. Users who trigger the model's safety filters too frequently will be removed from the test entirely. Despite these safeguards, Levoy concedes that "no feature can remove the need to behave responsibly."
For now, the AI Playground is a limited, free experiment restricted to a small fraction of Indigo users for a few weeks. However, the test has clear commercial implications, as the development team has admitted it cannot offer the tools for free forever. How consumers react to an AI that fundamentally alters reality at the point of capture will likely dictate Adobe's future product monetization strategy.