AI
Adobe's experimental AI tools let you explore photos as 3D scenes and build characters by talking to them
At its Max conference, Adobe is showing off a suite of AI-powered creative experiments that could reshape how video, photo, and audio editing work.
1/2
The Verge
Adobe has unveiled several experimental AI tools at its annual Max conference that could fundamentally change how creators work with images, video characters, and sound. These "sneaks"—research prototypes that may eventually become full products—span photo editing, character development, and audio design.
Interactive character development through conversation#
Project Screen Test flips the script on AI character creation by turning it into an interactive dialogue. Rather than describing a character purely through text prompts, creators can audition AI-generated personalities in a conversation-like interface. Users start by establishing baseline traits—say, "anxious, shy, awkward"—and selecting or describing a voice, then shape the character's deeper qualities through back-and-forth discussion: backstory, mannerisms, fears, and physical quirks.
The tool then generates animated screen tests based on scenarios creators describe. While still in the experimental phase with no confirmed path to Adobe's shipping products, the results show more personality consistency than typical AI-generated video characters, demonstrating what stronger creative direction can achieve in an otherwise limited medium.
Turning flat images into navigable 3D space#
Project Sight Span transforms 2D photos, paintings, and illustrations into explorable 3D environments. The system intelligently generates content beyond the original frame's borders—including objects and backgrounds you'd see if you moved the camera—while preserving artistic style and maintaining facial features and tattoos in photographs. When you settle on a new viewing angle, resolution sharpens for the final output, though the 3D space itself appears pixelated while navigating. The effect is similar to Apple's Spatial Reframing but with more aggressive repositioning freedom.
Sound design via interactive canvas#
Project Sound Bender makes audio customization tactile. Upload video, describe the sound you want in plain language—"bubbling sci-fi cauldron noises"—and the tool generates synchronized effects. An interactive square canvas assigns AI-generated sounds to each corner (dense gurgling, rumbling rocks, pitched bloops, crackling electricity) and lets creators drag a cursor to blend and prioritize effects in real time.

