Adobe camera app’s new feature will critique your photos using AI

Adobe's experimental iOS camera app Project Indigo just acquired an opinion. The update adds LLM-driven photo critique, a feature that evaluates framing, lighting, color, and emotional impact with the confidence of a contest judge who has seen too many sunset submissions. It also suggests reshoots — change your exposure, reposition that object, step left — and tells you how to salvage the frame inside Lightroom. This is not a filter. It is a backseat driver.

Marc Levoy runs the project. He built Pixel's computational photography before Adobe poached him. His diagnosis: most generative tools drown users in prompt engineering. You type, you wait, you regret, you retype. Levoy's answer is buttons. Deterministic buttons. Tap "remove wires," tap "remove trash cans," tap "remove background people." The app executes. No prompt craft required. The distinction matters. Google's Pixel camera coach offers vague framing nudges. Indigo serves specific, actionable, occasionally brutal feedback. It told one tester to delete a hexagonal white object from the frame. The object existed. The advice was correct. That stings.

Object removal has lived in Photos, Google Photos, and Photoshop for years. The interaction model never evolved: circle the intruder with your finger, pray the mask holds, accept the halo artifacts. Indigo discards that workflow. Category toggles replace the lasso. People in background. Trash and trash cans. Wires and poles. Fences. Vehicles. Custom description box for the rest. A tester watched the app erase a friend and the object he held from a background without inventing a single ghost limb. The seams vanished. This is the first time object removal feels like a professional tool instead of a party trick.

Depth of field simulation arrives beside it. One tap manufactures the blurred background that physics denied you. The effect convinces at Instagram resolution. Whether it survives 100 percent crops remains untested. Style transfer also appears — watercolor, pen and ink, ink line with color wash, monochromatic, backlit subject. The outputs are refined. The feature is useless. Prisma proved this market saturated in 2016. Better models only polish a gimmick. Levoy knows this. He criticizes prompt-based editing, yet Indigo retains a "describe your edit" flow. The contradiction sits there, unresolved, like a lens flare the team forgot to clone out.

The critique engine is the story. Photography software has long offered histograms, highlight warnings, level meters. Those are data. Indigo serves judgment. It assigns professional opinion to emotional impact. That phrase — emotional impact — should make photographers nervous. An LLM trained on internet captions and stock descriptions now rates the feeling of your street portrait. The feedback loop tightens: shoot, receive verdict, reshoot, receive higher score. Gamify the eye. Flatten the idiosyncrasy. The app might teach composition to beginners. It might also teach them that composition is a checklist.

Levoy's button philosophy extends to the critique. No prompt box asks "How should I improve this?" The app volunteers. That design choice respects the shooter's time. It also removes agency. When the machine names the flaw, the human stops looking for flaws. The skill atrophies. Pixel's camera coach avoided this by staying generic. Indigo dives specific. The difference is the difference between a coach and a critic. Critics get clicks. Coaches get results.

Adobe labels Project Indigo experimental. It is. The app ships on iOS only. No Android build. No desktop companion. The features live behind toggles that could vanish next quarter. But the direction is locked. Adobe intends to embed judgment into capture. The camera becomes an editor. The editor becomes a teacher. The teacher becomes a scorer. Photographers who grew up chasing light will distrust the score. Photographers who grew up chasing likes will optimize for it. The middle ground — the shooter who uses the critique to sharpen instinct, then ignores it — shrinks daily.

Levoy built Pixel's camera to make hard shots easy. He built Indigo to make easy shots "correct." The distinction is subtle and dangerous. A tool that removes trash cans perfectly is a gift. A tool that tells you the trash can ruined the emotional impact is a leash. The best photographers break rules. The best LLMs enforce averages. Project Indigo now sits on that fault line. Use the buttons. Distrust the verdict. Shoot anyway.