Adobe used the second day of its MAX creative conference to roll out 10 experimental technologies, a slate weighted toward artificial intelligence tools that reshape how creators edit images, video and sound. The annual Sneaks session, held Nov. 11, has become the venue where the San Jose, California-based software maker previews work still in its research labs, and a number of past demonstrations have later surfaced as shipping features.
The centerpiece demonstrations leaned on making complex workflows feel tactile. Project Deep Brush drew heavy attendee interest by letting users apply color as though painting directly onto walls, columns or car bodies captured in photos and video, with surface contours, highlights and shadows preserved. Adobe said the goal is to let creators stop adjusting individual pixels and instead explore a scene as if it were physical space.
Project Sight Span converts a flat photo, painting or illustration into a navigable three-dimensional environment, generating plausible objects and background outside the original frame. The 3D preview appears rough, but resolution is restored once a viewing angle is locked in. Repositioned facial features and tattoos remain recognizable and faithful to the source.
Project Screen Test pushes character design into conversation. Rather than writing a text prompt, a user assigns baseline traits such as anxious, shy or awkward, picks a voice, then talks with the character to fill in backstory, fears, movement quirks and goals. The tool then renders animated screen tests placing that persona in a described scenario.
Several projects target repetitive drudgery. Project Rinse Repeat copies an edit made in Photoshop and reapplies it automatically to other subjects or objects in the same image, adapting shadows to the shape and position of a differently sized product. Because it rebuilds the changes with native Photoshop tools, the results stay editable.
Others aim at motion and audio. Project On Cue gives users an “event curve” control to dictate how many times an action occurs in AI-generated footage and when, and to sync movement with music. Project Live Set allows on-the-fly changes during video generation. Project Sound Bender turns a text description of a desired noise into generated effects arranged on a square sound canvas, where dragging a cursor shifts the mix in real time; it can also isolate and recombine audio from existing clips.
Rounding out the list, Project Link Up lets Illustrator users drag an object to a new spot and have AI read the surroundings to match size, direction and perspective. Project Scene Sweep alters a performer’s wardrobe, styling or props across an entire video from a one-line prompt, potentially sparing reshoots. Project Hands On marks editable control points on objects so users can drag a perfume bottle into a diamond shape or lengthen a sofa without typing anything.
Adobe design evangelist Kelsey Sleigh, who co-hosted with actor Terry Crews, noted that Relight, the AI lighting tool shown at last year’s Sneaks, is now a standard Photoshop feature – a reminder that these previews are not idle experiments.
“Making these interactions more intuitive is also about giving experienced creatives more precise control,” Gautham Mysore, a senior principal research scientist at Adobe, told The Verge. He pointed to Sound Bender’s sonic exploration and Sight Span’s camera positioning as ways professionals can apply judgment and refine results directly. “That connection between creative intent and control is a big part of what we’re exploring.”
None of the 10 projects carries a release date, and Adobe has not committed to shipping any of them. Most are built on the company’s generative AI research, and with AI now central to its growth strategy, at least some are expected to reach Creative Cloud apps or the Firefly platform.
The stakes extend beyond product roadmaps. Adobe’s creative software business rests on subscriptions to tools that professionals have spent years mastering; features that collapse hours of retouching, compositing or sound design into a prompt or a drag threaten to reset the value of those skills, and to displace the stock libraries and manual workflows that agencies and studios currently bill for. The same capabilities also lower the barrier for smaller teams and solo creators, widening the pool of potential subscribers.
A common thread runs through the slate: turning technical, multi-step processes into something closer to direct manipulation. Some of the demonstrations verge on novelty, but each is aimed at a practical edit – a shadow that needs to match, a sound that needs layering, a camera angle that does not exist in the original shot.