As we told you yesterday, Adobe MAX 2023 is taking place these days, a major annual event that is arguably the most recognized company in the creative software and services ecosystem. It’s because of Photoshop, yes, but also because of Premiere, because of Illustrator, because of the PDF format… in short, the list is long and top-notch, adding all the necessary advantages (and some more) to enjoy the said recognition.

The opening presentation of the event arrived, as you have already seen, loaded with interesting news, but The truth is that Adobe still saved some surprises for later. So, if yesterday we told you the news regarding Firefly in its new three variants, as well as everything that is to come to Adobe Express and Lightroom, today we remind you of another announcement: the now classic Adobe Sneaks, as well as a response to one of the most widespread concerns due to the spread generative artificial intelligence.

Adobe Sneaks

Firefly is an important part of the integration of AI-based features, but the overall design in this regard, both current and future, goes much further. And now we will talk about it, about the future, because the key part of Adobe MAX is always the one that consists of the so-called Adobe Sneaks, which are nothing more than proofs of concept and previews of what they are currently working on with a view to their future integration into enterprise solutions. So everything we will see in this section is identified as projects because they are not yet technologies that will be deployed now.

• Project Stardust (photo): an image normally consists of multiple elements and what this new editing module proposes is their segregation. In this way, we can move, edit and delete elements and change the composition of the originally captured image until it suits our preferences.
• The See Through Project (photography): there are few elements in images more annoying than unwanted reflections, and we can’t always (and especially with smartphones) have a polarizing filter on the lens to eliminate them when shooting. The goal of this tool is to significantly facilitate their presence in the editing phase.
• Fast implementation of the project (audio and video): Firefly’s generative fill technology has already proven to be very effective in photo editing, and Fast Fill proposes to bring the same model to video editing and post-production as well. In this way, it will be possible to add, delete and edit content simply using a text prompt.
• Project Dub Dub (audio and video): A very hot area of ​​AI has to do with automatic audio translation. In this sense, Dub Dub is an Adobe proposal and offers the ability to create audio translated into the desired language from the original and preserve its tone of voice.
• Change the scene of the project (audio and video): it aims to facilitate the creation of a video composition from two different videos, one of which contains the protagonist-element and the other the scene in which it must be placed.
• The Res Up project (audio and video): as you may have already guessed from the name, in this case we are facing an intelligent scaling solution with which it would be possible to take a low-resolution video and increase it without resulting in a quality loss.
• The Poseable Project (3D and Design) – The goal is to allow the sensor to seamlessly interact with large 3D objects, including poses from photographs of real people. It represents a major advance that democratizes a type of design that is currently more complex.
• Project Neo (3D and design): If you haven’t experienced it yourself, try asking a designer about the process of integrating 3D elements into 2D designs. This project aims to greatly facilitate this type of operation, for which it offers tools and controls that allow this process to be carried out in a simplified manner.

Other projects shown today were The Primrose Projectfocused mainly on the world of fashion design, Glyph Ease Projectwhich facilitates resizing using glyphs (graphic symbols) and Draw & Delight projecta set of generative artificial intelligence tools aimed at helping designers throughout the creative process, from the creation of the first sketches to their final completion.

Content Credentials: Digital Watermarking for Artificial Intelligence

As generative imaging models proliferated and their results improved, Numerous concerns have been raised in this regard., among which the potential of using this technology to create malicious content undoubtedly stands out. We live, we can’t forget that, in a golden age of disinformation, so technology like this can play a key role in creating fake news, falsification, etc.

As I recalled yesterday, Adobe introduced VoCo as a proof of concept (yes, Adobe Sneak) in 2016, but the technology was never released as such, something we understand and is exactly in line with Adobe’s conscience and responsibilitywith full knowledge of possible harmful use. Instead, today we saw a Sneak that appears to use such technology (Project Dub Dub), but limits the output voice to the input recording, which is safe to use.

So the company has been concerned with preventing malicious use of its tools and AI as much as possible for quite some time now, and today we’re seeing a new and important advance in that regard, lContent mandate, ambitious proposal and for which it has added some top partners, such as Microsoft, Leica Camera, Nikon and Publicis Groupewith the intention of facilitating the identification of content created with the help of artificial intelligence.

The proposal consists of add credentials to any content, whether human-generated or AI-generated, to prove its origin and subsequent modifications, if any. Therefore, if necessary, this content will display an icon with the letters CR (from CRedential, the abbreviation of content credentials, CC, was not used because it could be confused with Creative Commons license types), which will be used to check a series of image metadata in which we can verify their origin, as said metadata will have specific annotations if artificial intelligence was used in the creation or editing of said image or video.

The fact that Microsoft has joined this proposal means that images created by Bingwhich in turn uses DALL-E 3, can be easily identified as such and that manufacturers such as Leica and Nikon have also subscribed to it, it will make it easier for the Czech Republic’s information to be present in digital assets from their creation.

We all know that when the law is done, the trick is done, and it probably won’t take too long to figure out how to remove this information from images and videos. However, the key is not so much at this point, avoiding unregistered modifications, as that users in general, start requiring the presence of an element such as these credentials to ensure the credibility of certain types of content. If this interest can be widely applied among the general public, we will find ourselves facing a major measure against the misuse of artificial intelligence for disinformation.

More information