Adobe previews AI tools for 3D photos, characters and sound

Project Screen Test lets users give an AI character personality traits and a voice, then develop its backstory, goals, fears and mannerisms by talking with it in a FaceTime-like format. Users can generate animated screen tests by describing a setting or scenario. Adobe has not said whether the standalone experiment will become part of its existing AI video tools. The demonstration showed characters with more consistent personalities, though the tool remains a creative exercise rather than a finished product.

Project Sight Span turns photos, paintings and illustrations into navigable 3D scenes, generating content beyond the original frame. Users can change the viewpoint and choose a new angle; although the 3D preview is pixelated, the selected view regains resolution. Adobe says the tool can preserve artistic styles, and that repositioned facial features and tattoos remain recognizable. The result allows more substantial changes in viewpoint than Apple’s Spatial Reframing, but can feel less like a conventional photograph.

With Project Sound Bender, users upload a video and describe the sound they want. The tool generates and synchronizes effects, then offers a square canvas for adjusting the mix in real time. Project Rinse Repeat, built around Photoshop, copies an edit—such as retouching blemishes or adding a product shadow—to other people or objects in an image. The resulting changes remain editable using native Photoshop tools.

Adobe’s “sneaks” are technology previews, not promises of future releases. The company has not confirmed whether any of the projects will reach Creative Cloud or Firefly. Adobe research scientist Gautham Mysore told The Verge that tools such as Sound Bender and Sight Span could give professionals more direct control over results. The experiments also point to a broader question raised by the article: whether capable AI tools could eventually replace resources and workflows used by creative professionals.