Adobe showcased a suite of experimental artificial intelligence features at its annual Max conference, highlighting advancements in transforming images into explorable 3D environments, developing AI-generated characters through dialogue, and enabling more intuitive photo editing. These "Sneaks," as Adobe terms its experimental technologies, offer a glimpse into the future of creative workflows.
One notable project, dubbed "Project Sight Span," allows a single image to become the foundation for an entire navigable environment. The AI transforms a 2D photograph into a 3D scene, enabling creators to explore the image beyond its original frame, discover new angles, and compose shots that extend the visual space. This technology builds upon earlier AI work in image extension and out-painting, aiming to create more immersive and expansive visual worlds from existing assets.
Another significant development is "Project Screen Test," which focuses on the creation and interaction with AI-generated characters. This tool allows creators to define a character's appearance, voice, and personality. Crucially, users can then engage in natural language dialogue with these characters to explore their potential responses and develop their personas for use in various scenes. This approach aims to make character development a more interactive and exploratory process.
Adobe also demonstrated "Project Deep Brush," an AI that understands an image's shape, color, and lighting to enable more intuitive painting and editing. Brushstrokes can conform to the contours of a surface, colors can be adjusted while preserving existing shadows, and light can be directly incorporated into a scene. This tool seeks to automate complex manual adjustments, allowing brushwork to follow the natural geometry of objects.
Further innovations include "Project Rinse Repeat," which captures edits made in one part of an image and applies them intelligently to another area, maintaining editability. Additionally, "Project Link Up" explores how creator gestures can be translated into precise actions within the software.
These experimental features are part of Adobe's broader strategy to integrate AI more deeply into its Creative Cloud suite. The company also announced the expansion of AI Assistants across applications like Illustrator, Premiere, and After Effects, designed to perform complex, multi-step tasks based on natural language prompts. Adobe is also planning to integrate its applications with third-party AI assistants through the Model Context Protocol (MCP), enabling these external agents to work directly with native project files while preserving editability. This move signals a shift towards AI agents that can participate more actively in the entire creative process, moving beyond simple content generation to executing complex creative tasks.
The company also announced a multi-year AI partnership with Disney, which will involve building custom AI models trained on Disney's characters and styles using Adobe's Firefly Foundry. This collaboration aims to accelerate artistic vision by leveraging AI as an accelerant for creative endeavors.