Want smarter insights in your inbox? Sign up for our weekly newsletters to get only what matters to enterprise AI, data, and security leaders. Subscribe Now
Adobe has been a creative software pioneer since its founding in 1982, and it aims to continue that storied and acclaimed history in the generative AI era.
Today, the company is announcing a fleet of Gen AI upgrades to Photoshop, including putting the image generation capabilities of its newly unveiled Firefly Image 3 AI model directly into the program. Adobe Firefly Image 3 will also exist as a stand-alone web app, like its predecessors.
But now, instead of users having to generate an image with the stand-alone Firefly web app, download it, and import it into Photoshop, they can simply click the new “Generate Image” pop over menu option directly in Photoshop, type in a text prompt or select from several pre-set starter options and edit them, and viola — within a few seconds, the Firefly AI model will generate the requested image.
Solving the ’empty page’ aka blank page or tabula rasa problem
“What we noticed when we looked at all these new people who were coming into Photoshop is that they were getting stuck with this ’empty page,’ problem” explained Zeke Koch, Adobe VP of Generative AI Product Management, in a video call interview with VentureBeat that took place days before the announcement under embargo.
AI Scaling Hits Its Limits
Power caps, rising token costs, and inference delays are reshaping enterprise AI. Join our exclusive salon to discover how top teams are:
- Turning energy into a strategic advantage
- Architecting efficient inference for real throughput gains
- Unlocking competitive ROI with sustainable AI systems
Secure your spot to stay ahead: https://bit.ly/4mwGngO
“So we gave them the ability to generate images, full text to images…Because once you have your image in Photoshop, all of a sudden you can use all the tools that we have already in Photoshop to make it better,” Koch added.
After all, Photoshop has more than 70 different separate tools and literally hundreds of different settings and options for creating and editing imagery, and its long history of continuing to add and advance tools such as lassos and context-aware fill have made it nearly industry standard among photography, illustration, and graphic design professionals (according to revenue intelligence software firm 6sense, Photoshop is the no. 2 graphic design software globally behind Adobe Illustrator).
By adding Gen AI capabilities directly within Photoshop, including the new Generate Image, Adobe seeks to remain in its leading position and further augment the existing tools designers already offer, while also introducing them to a new audience who may come to the program now because they are interested in experimenting and working with Gen AI.
Building on the success of Generative Fill
The addition of Generate Image to Photoship “really makes it possible for our new-to-Photoshop users to be able to bring their creative visions to life,” said Audrey Sousa, Principal Product Manager of Adobe Photoshop, in the same video interview that VentureBeat conducted with Koch.
Sousa also noted that Photoshop had seen an influx of new users following the debut of its Firefly-powered Generative Fill feature that debuted almost a year ago in May 2023.
That feature allowed users to highlight portions of an existing image and replace or augment what was already there by clicking the Generative Fill button and typing in a text prompt for what they wanted to add, remove, transform, or replace — turning a cowboy’s lasso into spaghetti, for one example (a pun on “Spaghetti western” movies).
But now, with the new Generate Image feature, Adobe is going much further, allowing you to generate full images powered by Firefly right from the get-go.
In addition, Adobe is upgrading Generative Fill, too, allowing users to add a “Reference Image.”
This new tool allows users to upload a selected image which Firefly will use to guide its generations in place of a text prompt: you can upload a hand drawn sketch, for example, or a brand logo or product asset, and the underlying Firefly 3 AI model will attempt to fill in the highlighted space of your image with something resembling or inspired by your reference.
Adobe is further launching Generate Background in Photoshop, which simplifies the creation of seamless backdrops.
“It’s probably one of the top use cases of Photoshop, generating new backgrounds,” Sousa noted. “Generate Background gives the user tons of options to work with and it really harmonizes the user image with the background.”
This feature is especially critical for those enterprises and brands offering or marketing physical products: simply upload a product image, select it with a lasso or other selection tool, remove the background and then click “Generate Background” and type in what you want the new background to be.
Firefly will try to match the background to the text prompt instructions you’ve provided, even adding appropriate coloration, shadows, water effects, and realistic textural effects to make it look like your product (or whatever the foreground image) is really there against a new background.
The utility is immense and obvious: allowing for entire series of new marketing and advertising assets to be created rapidly around the same product.
“You want to create a whole set of ads for it — one for summer, winter, fall, spring…in the old days, you’d have to go bring the product into a light box with props and move it all around,” explained Koch. “We automatically wholesale generate the background behind it, but then pick up the shadows, reflections, and light, so it really feels embedded in the image.”
Another feature, Generate Similar, allows you to create variations of what you just made with Generate Image, and Enhance Detail, which essentially upscales images, improving their sharpness and level of intricacy.
All the new Photoshop features are available starting today in the latest Photoshop (beta) app for desktop. Photoshop costs $22.99 per month when purchased on its own as a monthly subscription, with discounts for bundling it with other Adobe Creative Cloud software.
Ethical and enterprise-safe AI?
As has been its party line for the Gen AI era, Adobe’s press materials and employees told VentureBeat the company was committed to AI Ethics and specifically “accountability, responsibility, and transparency.” It says it does this by having trained Firefly only on “licensed content, such as Adobe Stock.”
In addition, Firefly “was designed to generate content for commercial use that does not infringe on copyright and other intellectual property (IP) right such as trademarks and logos,” and in a big win for users, it “offers customers IP indemnification for Firefly generated content,” as Adobe’s press materials state.
In the case of the new Generate Image feature and any other Firefly-powered features used to generate new portions of images with AI, Adobe will automatically apply its “Content Credentials,” metadata embedded in the image that indicates it was made with AI. Adobe says this is “tamper proof.”
It may not ease all concerns — given a recent report that a small percentage of Adobe Stock images used to train Firefly were likely from rival AI image generator Midjourney, itself trained on data and imagery from human artists scraped without express consent from around the web — but in the fast-moving world of gen AI, it probably sounds pretty damn good.
VentureBeat uses Adobe Firefly, Midjourney and other AI tools and programs to generate content.
