Adobe Project Indigo AI Playground critiques photos and suggests edits using generative AI

Reviewed byNidhi Govil

5 Sources

Share

Adobe is testing AI Playground inside its experimental iOS camera app Project Indigo, bringing photo critique tools that analyze framing, lighting, and composition. The update uses Google's Nano Banana model to power distraction removal, depth of field effects, and style transfer features. Only a small percentage of users will access the free test over the next few weeks.

Adobe Project Indigo Introduces AI Playground for Mobile Photography

Adobe is expanding its experimental iOS camera app Project Indigo with a suite of AI-powered features bundled under a new section called AI Playground

1

. The update marks a significant shift for an app that launched last year with an emphasis on delivering natural, SLR-like images through computational photography and full manual controls

2

. Marc Levoy, Adobe Fellow and head of the project who previously developed Pixel's camera capabilities, is leading the initiative to explore how generative AI can assist mobile photographers at the point of capture

4

.

Source: 9to5Mac

Source: 9to5Mac

The AI Playground features are being tested with only a few percent of Project Indigo users over the next few weeks, with free access requiring no sign-on

3

. Adobe's goal is to learn how people use generative AI-powered editing in mobile photography workflows, and depending on the experiment's success, the company may extend the test, expand the user pool, or eventually offer a paid version

2

.

Photo Critique and AI-Driven Photo Suggestions Transform Learning

The most distinctive addition is the photo critique feature, which provides professional analysis of framing, lighting, colors, and emotional impact

1

. Unlike Google's camera coach feature for Pixel phones that offers generic framing suggestions, Project Indigo's feedback is descriptive and aims to help users learn photography techniques, even if they don't agree with the AI assessment

1

.

The Photo Guidance section delivers two types of suggestions. First, capture and edit suggestions provide tips about reshooting photos, including how to change framing, exposure, and objects in the viewfinder

1

. The app can identify distracting elements near the edge of the frame and recommend better composition strategies

4

. A second section explains how to improve existing photos using Adobe Lightroom controls

1

. The Nextcam Team behind Project Indigo is particularly interested in tailoring feedback to specific smartphone capabilities, recognizing that different phone models with varying lens configurations should receive different suggestions

4

.

Advanced Distraction Removal and Depth of Field Generation

The Object Editing section introduces sophisticated distraction removal tools that surpass traditional object removal features found in Apple Photos, Google Photos, and Adobe Photoshop

1

. Instead of requiring users to circle or select objects by drawing on the screen, Project Indigo provides toggles for removing specific categories like people in the background, trash and trash cans, wires and poles, fences, vehicles, and other clutter

1

. Users can also describe custom objects to remove, with the app using generative AI to fill in the blank spaces left behind, similar to Google's Magic Eraser feature

2

.

The results are impressive, with testing showing the app successfully removed people and objects from backgrounds without creating strange artifacts

1

. Additionally, the app allows users to create depth of field effects to simulate a blurred background, adding professional-looking bokeh to mobile photos

1

.

Style Transfer and Natural Language Prompts Expand Creative Options

The Styles section offers preset AI-generated filter effects that can transform photographs into illustrations, adjust lighting effects, and apply various artistic techniques

2

. Available styles include watercolor, pen and ink, ink line with color wash, monochromatic, and backlit subject effects

1

. While style transfer isn't new technology and recalls the early days of the Prisma app, the outputs look refined with advanced models

1

.

Source: TechCrunch

Source: TechCrunch

The Custom Edit feature unlocks broader editing capabilities through natural language prompts, allowing users to describe desired edits in detail

2

. This enables editing workflows not available in preset tools, though Levoy acknowledges this approach carries risks

1

. An intriguing use case involves converting artwork into photorealistic images—users can photograph a sculpture or painting at a museum, and AI Playground attempts to recreate it as a realistic photo

4

.

Google's Nano Banana Model Powers Initial Release

Adobe is using Google's Nano Banana model to power these AI-powered features, though the company remains open to swapping in other models, including its own Adobe Firefly

1

. The Nano Banana model is cloud-based and requires an active internet connection, with edited photos capped at 2K pixels to maintain processing speed

5

.

One technical limitation is that Nano Banana only outputs Standard Dynamic Range images, even when users capture shots in High Dynamic Range

3

. To address this, Adobe built a custom machine learning model to post-process the SDR images returned by Nano Banana, boosting them back to HDR, though the company acknowledges this solution isn't perfect

3

.

Privacy and Transparency Considerations

Adobe is taking a measured approach to user data during this experimental phase. The company will only collect button presses while using the AI Playground and won't examine prompts, images, or upload them to servers for AI model training

3

. Since there's no sign-on requirement, the analytics gathered remain anonymous

3

.

However, the experiment lacks the same level of AI transparency that Adobe typically maintains. The Nextcam Team is working toward adding C2PA Content Credentials—a metadata standard for identifying AI content widely used by Adobe Firefly models—to images manipulated by Project Indigo's generative AI tools

2

. Levoy notes that Nano Banana already applies Google's invisible SynthID watermarking, and emphasizes that whether edits are fair or misleading depends on user behavior and responsible presentation

2

.

The AI Playground features only work on images captured within Project Indigo, meaning the sparkle button that opens the AI Playground will be grayed out for images imported from elsewhere

3

. Adobe warns that repeatedly triggering Nano Banana's safety filters can result in users being removed from the test

5

. Large language models power the critique and suggestion features, marking a significant integration of AI into user learning and creative workflows

1

.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved