Adobe previews AI-powered object-aware editing

GaryPageau ,
Getting your Trinity Audio player ready...

At Adobe Max, Adobe Inc. unveiled Project Stardust, a new object-aware editing engine using generative AI to edit images, and its latest early-stage innovations, highlighting exciting new uses of generative AI and 3D technologies across multiple creative domains, including photo, video, audio, 3D, design and fashion applications. The 11 inventions are featured in Adobe’s annual “Sneaks” showcase, where Adobe engineers and research scientists offer first looks or “sneak peeks” at prototype ideas and technologies.

Project Stardust enables users to easily select, edit and even delete complex elements in any image, making image editing more intuitive, accessible and time efficient for any user, the company said. For example, Project Stardust lets users select persons in a photograph, move them to a different place in the composition and fill in the background where they were previously standing. Users can also change elements like the color of a person’s clothing or the position in which they’re standing – treating any flat images like a file with layers.

Among the sneak peaks, in addition to Project Stardust:

Photo

  • Project See Through is an AI-powered tool that makes it simple to remove reflections from photos.

Video & Audio

  • Project Fast Fill brings the power of Firefly generative AI to video for the first time. Leveraging Generative Fill technology – which Adobe Photoshop already uses to ease the addition, removal, or expansion of content in images with simple Firefly-powered text prompts – Project Fast Fill offers an early look at what human-prompted generative AI could enable inside Adobe video editing tools including Premiere Pro and After Effects.
  • Project Dub Dub Dub automates the video dubbing process, making a historically labor- and cost-intensive process as easy as clicking a button. aligned with the original dialogue and ready to publish.
  • Project Scene Change makes it easy for video editors to composite a subject and scene, from two separate videos captured with different camera trajectories, into a scene with synchronized camera motion.
  • Project Res Up is a tool that easily converts video from low- to high-resolution using innovative diffusion-based upsampling technology.