Adobe Revolutionizes Photography with AI in Project Indigo

Adobe Integrates AI into Project Indigo for Enhanced Photography

Adobe recently unveiled a series of AI-powered features for its experimental camera app on iOS, known as Project Indigo. Launched last year, this app already offered professional tools such as multi-image super resolution and various capture modes. Now, it is enhanced with features that utilize large language models (LLMs) to critique photos and provide editing suggestions.

In addition to these critiques and suggestions, the app introduces additional AI features such as advanced object removal, depth-of-field generation, and the ability to apply different artistic styles to photos.

Innovation Led by Marc Levoy

Marc Levoy, who leads this project at Adobe, is known for previously enhancing the photographic capabilities of Google's Pixel devices. He emphasizes that most current generative AI tools rely on text prompts for editing, which can complicate the process of achieving the desired result. To simplify this process, Adobe's new experimental features are designed as buttons, allowing for more predictable outcomes.

For instance, Google's photo coaching feature for Pixel phones, introduced last year, offered fairly generic framing advice. In contrast, the tools in Project Indigo are more detailed and can teach photography tips, even if the user does not always agree with the AI's assessment.

Capture Critiques and Suggestions

Two main features stand out in this category. The first is photo critique, which provides a "professional" evaluation of the framing, lighting, colors, and emotional impact of an image.

The second feature offers capture and editing suggestions, providing advice on how to retake the photo by adjusting the framing, exposure, or modifying objects present in the frame. For example, the app suggested removing a hexagonal white object from the frame of a photo taken by a user.

Enhancing Photos with Adobe Lightroom

Another section of the app indicates how to enhance an existing photo using Adobe Lightroom controls. While apps like Apple Photos, Google Photos, and Adobe Photoshop have long offered object removal, this feature often requires the user to outline or select an object, which is not always precise.

The new feature in Project Indigo allows users to choose elements to remove from an image, such as people in the background, litter, power lines, vehicles, and other visual clutter. It is also possible to describe a specific object to be removed. The results achieved are often impressive.

Creating Depth of Field and Style Transfer

The app also allows the use of AI to create depth of field, simulating a blurred background for a photo. With this update, Adobe is also experimenting with style transfer functionality, which allows an image to be transformed into various artistic tones such as watercolor, pen and ink, or monochrome. Some styles evoke the early days of the Prisma app. Thanks to advanced models, the results are refined, although style transfer is not a groundbreaking novelty.

Custom Editing Tools

Despite Levoy's critiques of prompt-based editing, Project Indigo offers a feature that allows users to describe the desired editing. This enables users to create editing flows not available in predefined tools, although it may also lead to less polished results.

Adobe uses the Nano Banana model based on Google's Gemini for these features but remains open to integrating other models, including its own Adobe Firefly model. These features, grouped under the AI Playground tab, are still in testing and are accessible only to select users. They may never be widely deployed, but they have the potential to raise user awareness of the subtleties of photography, beyond mere mass content creation.