Adobe is adding a series of AI features to its experimental iOS camera app, Project Indigo, which uses large language models (LLMs) to analyze photo content and provide users with photography critiques, editing suggestions, and intelligent optimization solutions. The app previously supported professional shooting controls, multi-frame super-resolution, and various photography modes, and this update further enhances AI-assisted creative capabilities.
New features include photo critiques, shooting and editing suggestions, advanced object removal, AI-generated depth of field, and photo style transfer. The photo critique feature provides analysis similar to that of a professional photographer, covering aspects such as composition, lighting, color, and emotional expression; the shooting suggestion feature offers specific guidance on re-composing, exposure adjustments, and optimizing visual elements to help users improve their photography skills.
Marc Levoy, the lead developer of Project Indigo, previously participated in the development of Google Pixel camera technology. He stated that most current generative AI image tools rely on user input prompts, but ordinary users often find it difficult to accurately describe the desired adjustment effects. Therefore, Project Indigo focuses more on button-based and deterministic operations, allowing AI to directly provide executable editing solutions.
In terms of smart editing, Project Indigo's new object removal feature can automatically identify and remove distracting elements in the background, such as people, trash cans, utility poles, and vehicles, while also supporting user-defined descriptions for objects to be removed. Compared to traditional methods requiring manual selection, this feature reduces the operational cost and improves the processing results.
Additionally, the app supports generating depth-of-field effects using AI to simulate background blur for regular photos, and offers style transfer effects such as watercolor, pen drawing, colored ink lines, monochrome, and backlighting. However, style transfer is not a new technology, and its practical application value still needs further verification.
Currently, the new AI capabilities of Project Indigo are supported by the Nano Banana model based on Google Gemini. Adobe also mentioned that it may test other AI models, including its self-developed Adobe Firefly model, in the future. These features are currently available in the "AI Playground" testing area and are only open to a select group of users, and whether they will be fully promoted remains uncertain.
