Adobe Max 2026: New AI "Sneaks" Let You Interview Characters and Walk Through Photos

At its annual Max conference, Adobe has unveiled a new suite of experimental "sneaks"—cutting-edge AI tools that could redefine how creators interact with digital media. These previews showcase capabilities ranging from "auditioning" AI characters to transforming flat photographs into explorable 3D environments.

Project Screen Test: Directing AI Through Conversation

One of the most distinct experiments is Project Screen Test, a tool designed to give creators more granular control over AI-generated characters. Rather than relying solely on text prompts, users can develop a character’s personality by "interviewing" them in an interface similar to a FaceTime call.

Creators start by defining base traits—such as "anxious, shy, or awkward"—and assigning a voice via presets, prompts, or audio uploads. From there, users can discuss backstories, goals, and fears directly with the character to refine its persona. The tool includes a "Personality" tab for manual adjustments and can generate animated screen tests of the character in specific scenarios. This approach aims to solve the "soulless" nature of standard AI video by providing the intentionality found in traditional animation.

Project Sight Span: Walking Inside a Photograph

Project Sight Span takes static images—including photos, paintings, and illustrations—and converts them into navigable 3D environments. The tool uses generative AI to build out objects and backgrounds that exist outside the original frame, allowing a user to reposition the camera or even turn around to see what is "behind" the photographer.

While the environment may appear pixelated during movement, the resolution is restored once a new angle is selected. Notably, the tool preserves artistic styles and maintains the integrity of features like tattoos or facial structures even when the shooting point is aggressively changed.

Project Sound Bender: Real-Time Audio Mixing

For audio professionals, Project Sound Bender introduces a "sound canvas" that functions like a color wheel for audio effects. Users can upload a video and describe a sound—such as "bubbling sci-fi cauldron noises"—which the AI then generates and syncs to the footage.

The interface features a square canvas where each corner represents a different variation of the sound, such as "dense gurgling" or "crackling electric fire." By dragging a cursor across the canvas, creators can blend these elements in real time to find the perfect sonic texture.

Project Rinse Repeat: Intelligent Batch Editing

Project Rinse Repeat focuses on streamlining repetitive tasks within Photoshop. The tool allows users to perform a complex edit on one subject—such as airbrushing blemishes or adding shadows to a product—and then "copy/paste" that entire process onto other objects or people in the same image.

Unlike standard filters, Rinse Repeat adapts the edits to fit different shapes and sizes. Because it is built on native Photoshop tools, the resulting shadows or touch-ups remain fully editable for the user.

The Bridge Between Intent and Control

While these "sneaks" are currently tech previews with no guaranteed release date, they signal Adobe’s strategy to integrate generative AI into the core of the Creative Cloud.

Gautham Mysore, a senior principle Adobe research scientist, noted that these tools are designed to give experienced professionals more precise control. "That connection between creative intent and control is a big part of what we’re exploring," Mysore told The Verge. By making complex workflows like 3D reframing and audio synthesis more intuitive, Adobe aims to turn high-level AI capabilities into practical, interactive tools for the modern creator.