Adobe’s Project Indigo Uses AI to Critique and Perfect Your Photography
Adobe is pushing the boundaries of mobile photography by integrating Large Language Models (LLMs) into its experimental iOS camera app, Project Indigo. Moving beyond simple filters, the app aims to act as a digital mentor, offering sophisticated critiques and intelligent editing suggestions to help users master the art of photography.
Moving from Prompt-Based Editing to Deterministic AI
One of the most significant shifts in Project Indigo is the move away from the unpredictability of prompt-based generative AI. Marc Levoy, the head of the project and former architect of Google Pixel’s camera capabilities, has noted that finding the "perfect prompt" to achieve a specific visual result is often a frustrating experience for users.
To combat this, Adobe is implementing "deterministic" AI features. Instead of forcing users to type complex instructions, Project Indigo utilizes intuitive buttons and toggles. This approach provides more reliable, predictable outputs, ensuring that the AI assists the creative process rather than complicating it with "AI slop."
The AI Photo Coach: Critiques and Real-Time Suggestions
The core of this update lies in the "AI Playground" tab, which introduces two groundbreaking features driven by Google’s Gemini-based Nano Banana model. Unlike Google’s Pixel "camera coach," which offers generic framing advice, Project Indigo provides highly descriptive, professional-grade feedback.
The Photo Critique feature analyzes images based on technical and aesthetic pillars: framing, lighting, color balance, and even emotional impact. Complementing this is the Capture and Edit Suggestions tool, which provides real-time instructions to improve a shot before it is even taken. For example, the app can detect distracting elements in the viewfinder and suggest adjustments to exposure or framing to ensure a cleaner composition.
Advanced Generative Editing and Style Transfer
Project Indigo is also evolving its post-capture capabilities with sophisticated generative tools:
- Intelligent Object Removal: Unlike traditional tools that require manual circling, Adobe’s new feature offers specific toggles to remove common distractions like people, trash cans, wires, poles, and vehicles. Users can also describe custom objects for removal, producing seamless results without visual artifacts.
- Depth of Field Generation: The app can simulate professional bokeh by intelligently generating depth of field in photos that were originally shot with a flat focus.
- Style Transfer: Drawing inspiration from early AI art apps, users can apply refined styles such as watercolor, pen and ink, or monochromatic tones to their images.
Why This Matters for the AI Landscape
Adobe’s approach signals a maturing phase in consumer AI. While many companies are focused on purely generative "text-to-image" workflows, Adobe is focusing on "AI-augmented" workflows. By using models like Gemini Nano or its own Adobe Firefly to enhance—rather than replace—human intent, Adobe is positioning AI as a sophisticated tool for skill acquisition and precision editing.
Key Takeaways
- Deterministic AI: Project Indigo prioritizes button-based, predictable AI actions over unpredictable prompt-based editing to improve user experience.
- Intelligent Mentorship: The app uses LLMs to provide deep, descriptive critiques on lighting, framing, and composition, acting as a professional photography coach.
- Advanced Object Cleanup: New generative tools allow for one-tap removal of specific clutter like wires and vehicles, significantly lowering the barrier to high-quality photo editing.