With Firefly Image 3, Adobe aims to integrate more AI tools for various apps
By Marty Swant • April 24, 2024 • 5 min read •
One year and 7 billion images later, Adobe’s new version of Firefly is now expanding across its design universe.
Adobe yesterday released a third generation of its Firefly AI image model to power new capabilities for creating and generating images. Firefly Image 3, unveiled at Adobe Max London, includes a range of new features for apps to use in Photoshop and InDesign, but it willl also now exist in a new beta version of a standalone Firefly web app.
As Firefly Image 3 brings major qualitative updates such as photorealistic images, Adobe also is adding tools for both new users and professionals. Along with a new text-to-image tool that lets people make images in seconds, other features for Photoshop include ways to generate backgrounds, replace objects within images and use reference images to inform what Firefly should create.
While marketing experts find the new AI model’s capabilities impressive, Adobe’s critical mass and gravitational pull could also give it an added advantage. Marketings experts also think the updates could help lower the barrier to entry for everyday users but also make professionals’ jobs faster and easier. That’s something Adobe is hoping for as well.
Shortening the learning curve has helped draw in new customers to Adobe’s products, according to Erin Boyce, senior director, Photoshop Product Marketing at Adobe. While Photoshop has had a 30% year-over-year increase in gross new Photoshop subscriptions, she said the “vast majority” of people visiting Firefly’s website are completely new to Adobe.
“What we find for new customers is they open Photoshop, they open a new file and there’s a big blank canvas staring them in the face and they don’t quite know where to start,” Boyce said. “With a simple text prompt, they can get started right away creating. For pros, this is an amazing way to ideate right in Photoshop and get from first idea to iteration to iteration, right into editing where they’re doing all of their work.”
Since it debuted a year ago, Firefly has generated 7 billion images, according to Adobe. While rolling out numerous new features across various apps, Adobe has also sought to build more trust with how the AI models are trained. Along with spearheading new content authenticity standards, the company also claims Firely was trained using only licensed images. However, recent reports have noted some of the images used to train Firefly came from Midjourney, which has faced lawsuits for scraping content on the internet without permission.
Adobe’s news is just one of many recent updates from a range of generative AI image platforms. Startups like Stability AI and Leonardo AI have added new tools and ways to access their platform, with Stability debuting a new API for its Stable Diffusion 3 model, and Leonardo eyeing business customers with its enterprise API. Meanwhile, Canva recently acquired the professional layout software startup Affinity, Google Ads introduced a way to use AI images in demand-generation campaigns, and Meta’s updates for Meta AI include ways to instantly create images and animations.
While Firefly 3 helps improve image quality, advertising execs and analysts say Adobe’s strategy of integrating it across its ecosystem of apps provides an advantage for adoption. Along with having a massive base of enterprise customers, Adobe’s familiar UX helps drive adoption of new tools using generative AI.
“Marketers forget to infuse their strategy with generative AI and they look to it more as this series of tech deployments,” said Gartner analyst Nicole Greene. “That’s the risk of going out just to bring gen AI into a function with a new point or feature or function versus saying, ‘What do we already have?’”
Adobe’s integration of content credentials and its use of licensed images help it appeal to enterprise customers, Greene said, adding that both factors have become more like table stakes. And while many marketers are still averse to using AI to generate entire images, tools like generative fill be appealing.
According to Gartner’s recent survey of 300 marketers, 47% said they’ve generated images in the past six months. But the benefits also bring an existential risk. Although 47% said using AI tools have saved time and made them more creative, 87% also said they worry about how AI could potentially replace their roles.
Even if agency designers are using tools in Adobe and Midjourney to help with concepts, the AI-generated images aren’t always making their way into ads — at least not yet.
One of the main ways generative AI is impacting agencies is through automating and optimizing workflows. Tools like Firefly help with creating high volumes of content for clients, said Ly Nguyen, head of creative at Movement Strategy, adding that it’s especially important for agencies that are still in the process of scaling. One example is using Firefly to remove or change the background for hundreds of images that can be done with the push of a button.
“There’s a threshold, a certain point where you have people doing the work that is sustainable,” Nguyen said. “When you get to like large volumes, it becomes not smart anymore to have someone manually do that type of work.”
Rather than just creating AI images from text, Firefly and other tools are part of a broader set of tools agencies are using. Brian Brown, U.S. svp and creative innovation lead at Razorfish, said many generative AI features from companies overlap, but some are better at certain tasks than others. For example, some might differ in aesthetic, some might help more with automation, some might be better at making certain types of content, while others might help more with personalization.
“When it comes to evaluating new tools, it also comes back to automation,” Brown said. “Is there something that this does in a way that’s going to make us more efficient, or more creatively powerful that something else isn’t doing right now?”
https://digiday.com/?p=542324
