It makes you wonder which company is not going to invest in artificial intelligence in the near future. In any case, Adobe wants its share of the pie. The American company has decided to enter the AI race by presenting “Firefly”, described by the company as a “family of new creative AI generative models”.
This model will first be integrated into Photoshop photo creation and editing software. In particular, it will make it possible to produce images and “text effects”, thanks to simple text requests, just like classic generative AIs like Dall-E or Midjourney.
“Give creators superpowers”
In the presentation video, the result is impressive. A simple field photo in summer can be transformed into a winter landscape, by telling the AI to “change the scene to a winter day”. In another example, it is possible to spawn other buildings instead of the ones in the original artwork, as well as add parallel universes, simply by asking the spawn tool to do so.
Protected by copyright
Adobe also wants to avoid criticism and protect copyright well. Unlike the generative artificial intelligence software now known, the US company will not use copyrighted works. To train Firefly, Adobe used its own royalty-free database, Adobe Stock, and not images retrieved from the Internet.
Artists will also be able to request that their production not be used to train Adobe’s AI. Adobe plans, on the other hand, to remunerate creators who agree to “monetize their talents” by helping the machine train. The compensation model is expected to be released in a few weeks, after Firefly’s beta testing phase.
This beta version is already available on the Adobe site, with registration and queue required. Adobe already aims to extend its AI to other software in its suite, such as Illustrator (for logos and vector images) or Premiere Pro (for video editing), according to Alexandru Costin, deputy director in charge of AI at Adobe, for the American outlet. The Verge.
Source: BFM TV
