Adobe Unveils Major New Generative AI Innovations for Photo, Video, Audio, and 3D, Including an Object-Aware Editing Engine

Adobe has released a preview of Project Stardust, a ground-breaking object-aware editing engine driven by generative AI that allows you to click on an object to magically relocate or delete it.

At the world's largest creativity conference, Adobe MAX, on October 11, 2023, Adobe unveiled Project Stardust, a potent new object-aware editing engine that uses generative AI to revolutionize image editing, as well as its most recent collection of early-stage innovations. These innovations highlight exciting new uses of generative AI and 3D technologies across multiple creative domains, including photo, video, audio, 3D, design, and fashion applications. The 11 innovations are highlighted in Adobe's yearly "Sneaks" showcase, where engineers and scientists from Adobe provide first looks or "sneak peeks" at prototype ideas and technologies. Each of the 11 inventions exhibits the potential to eventually become significant components of Adobe products, which are relied upon by millions of users worldwide.

​Project Stardust makes image editing more simple, accessible, and time-effective for all users by allowing users to quickly choose, edit, and even delete complicated parts in any image. For instance, Project Stardust enables users to pick people in a picture, move them around the frame, and fill in the background where they are standing. Users can even alter details like a person's clothing color or standing position by treating any flat image like a file with layers.

​Adobe Firefly, a group of artistic generative AI models, is now commercially available, expanding the variety and range of new uses beyond image creation. This class of Sneaks provides compelling examples of how generative AI can reinvent workflows for video, audio, animation, and motion graphics, speeding up ideation and discovery processes and reducing post-production time from days to minutes.

Here are the innovations that Adobe announced:

Photo

Project Stardust

A potent new object-aware editor that transforms picture editing and magically relocates or eliminates objects by simply clicking on them, users of Project Stardust can choose people in a photo, move them to a different location in the composition, and fill in the background where they were standing by simply selecting, editing, and deleting complicated elements in any image. Users can even alter details like a person's clothing color and standing position by treating any image like a file with layers.

Project See Through

A tool with artificial intelligence (AI) that makes it simple to eliminate reflections from photographs. Everyone is familiar with the frustration of glass reflections in photographs, which obscure the subject of the picture and frequently render it useless. Although it's difficult, if not impossible, to get rid of reflections with current software, See Through makes it a lot easier.

Video and  Audio

Project Fast Fill

It first applies the Firefly generative AI's power to video. Project Fast Fill offers an early glimpse of what human-prompted generative AI could enable inside Adobe video editing tools like Premiere Pro and After Effects by Leveraging Generative Fill technology, which Adobe Photoshop already uses to simplify the addition, removal, or expansion of content in images with straightforward Firefly-powered text prompts.

Project Dub Dub Dub

It automates the video dubbing procedure, making what was once a time- and money-consuming task as simple as pushing a button. A recording or audio track of a video can be automatically translated into any supported language using Project Dub Dub Dub's AI capabilities while maintaining the original speaker's voice, aligned chronologically with the original speech, and ready for publication.

Project Scene Change

It makes it simple for video editors to sync the camera movements in a scene by combining a subject and a scene from two separate videos shot using different camera trajectories.

Project Res Up

A tool that quickly improves videos from low resolution to high definition utilizing cutting-edge diffusion-based technology.

3D and Design

Project Poseable 

It is an innovation in AI-based image generating models, enabling the image generator to easily interact with big 3D objects, including stances taken from real-world photographs of humans. Project Poseable, which can generate a 3D depiction from text input, take into account depth and perspective, and realign the item, is integrated with the prototype. While using conventional skeletal systems to create 3D scenarios is technically challenging and time-consuming, Project Poseable provides an easy solution that enables everyone to broaden their creative abilities.

Project Neo 

It allows designers to add 3D shapes to their 2D works without having to be proficient in 3D creation software. The intricacy of those operations, which can require years of experience, now limits the capacity of designers of 2D designs, such as infographics, posters, or logos, to incorporate 3D elements. Creators can adopt streamlined 3D design using 2D tools and techniques thanks to Project Neo.

Project Primrose 

By exhibiting the possibility of flexible textile displays and transforming clothing into artistic canvases, it blurs the boundaries between technology and fashion. The content displayed on this interactive garment, which was made with Adobe Firefly, After Effects, Stock, and Illustrator, may be customized in an unlimited number of ways.

Project Glyph Ease 

By simplifying the laborious design process of glyphs—the unique form and shape components of each letter character—it increases accessibility to personalized letters. Project Glyph Ease starts with a hand-drawn letter shape and then uses artificial intelligence to create a full set of glyphs that perfectly fit the input lettering style. The created glyphs are then simple to alter in Illustrator.

Project Draw and Delight 

It provides a set of generative AI tools to assist anyone during the creation process. These tools range from turning raw ideas—often given as doodles or scribbles—into polished and refined sketches to experimenting with color schemes, different stylings, and background types.