Adobe Max Sneaks: Walk Into a Photo, Audition an AI Character

Adobe has announced four experimental "sneaks" at its Max conference that turn photographs into navigable 3D scenes, let designers build AI characters by talking to them, mix sound effects on a draggable canvas, and copy retouching work across subjects with one click.

None of these are shipping products yet. They're tech previews of capabilities that might eventually reach Creative Cloud apps or the Firefly platform, and Adobe offers no guarantee any will graduate into official features. But nearly all are built on the company's generative AI stack, and with AI now central to Adobe's growth strategy, at least some seem likely to launch.

How does Project Screen Test work?

The strangest of the four is Project Screen Test, a tool for developing AI-generated characters for video content. Instead of describing a character's voice and behavior in text prompts, you direct the character in an audition-like environment.

You start by naming base personality traits — "anxious, shy, awkward," for example — and assigning a voice: pick a preset, describe it in a prompt, or upload sample audio for reference. From there you expand the persona by conversing with the character as if on a FaceTime call, outlining backstory, temperament, goals, flaws, fears and movement quirks. Once satisfied, you generate animated screen tests by describing an environment or scenario.

It remains a standalone creative exercise, with no clear path into Adobe's existing AI video tools. In demos, the results carried a more identifiable and consistent personality than most animated AI characters — which is precisely the point, since AI video generation offers far less creative control than traditional animation, where every aspect of a character is intentional.

What can Sight Span do to a photo?

Project Sight Span turns any flat image — photo, painting, or illustration — into a 3D environment you can move through and reposition. The generated space invents objects and backgrounds outside the original frame that read as natural parts of the image, even when you turn to look behind you. Navigating the 3D view looks pixelated and ill-defined, but resolution is restored once you select your final angle.

It preserves artistic styles, and repositioned tattoos and facial features remain surprisingly faithful to the reference. Think of Apple's Spatial Reframing on steroids: you can reposition aggressively and change the shooting point entirely — though the result starts to feel less like a photograph.

Can sound design really be a color wheel?

Project Sound Bender treats AI audio mixing like clicking a cursor around a color wheel. Upload a video clip, describe your desired audio in a prompt — "bubbling sci-fi cauldron noises," say — and the tool generates synced sound effects, then presents a square canvas for real-time adjustment.

Each corner of the canvas gets a relevant AI-generated sound: dense gurgling with popping bubbles, rumbling boiling rocks, pitched bloops, crackling electric fire. Dragging the cursor across the canvas selects which noises dominate, and each corner can be customized with different effects.

What about copy-paste photo editing?

Project Rinse Repeat copies edits and applies them automatically to other objects or people in the same photo. Airbrush blemishes on one person, and it adapts those changes to everyone else in a group shot needing similar touchups. It can also add consistent shadows to products of different shapes and sizes in marketing graphics.

Built around Photoshop, it stays non-destructive in spirit: applied edits remain fully editable because the tool recreates them with native Photoshop tools.

Why intuitive controls matter to pros

The common thread is making complex photo, film, and audio workflows interactive enough to feel almost like toys — while still producing practical changes. For experienced users, though, the point isn't simplification alone.

"Making these interactions more intuitive is also about giving experienced creatives more precise control," senior principal Adobe research scientist Gautham Mysore told The Verge. "Sound Bender's sonic exploration capability and Sight Span's camera positioning are examples of how professionals could apply their judgment and refine a result more directly. That connection between creative intent and control is a big part of what we're exploring."

The experiments are also a signal. If tools like these prove useful enough, AI may eventually replace the tools and stock libraries creative professionals rely on today — however complicated their feelings about the technology.