Now, let’s chat about something that’s been on all our minds for close to a year now. Remember when OpenAI dropped Dall-E 2 last April? It had us all speculating when Adobe would step up to the plate with a rival. The idea? A neat integration of a similar tool into design stalwarts like Photoshop, Illustrator, and Premiere.
I’ll admit, the suspense was palpable. Especially given the speed at which these hotshot startups, namely OpenAI, Stability AI, Midjourney, and Runway ML, were churning out these jaw-dropping new gizmos. Then, as if on cue, Adobe answered our prayers last March by releasing a beta version of their own text-to-image model – Firefly. The response? An instantaneous stir in the design and AI communities, nudging our very own Margaux to examine Adobe’s re-emergence as an AI titan this week.
To understand just how sleek and user-friendly Firefly is, take a glance at the images in this piece. All crafted by designer Clark Miller, who used Firefly’s generative fill features in Photoshop. Now, let’s hear what Clark thinks about the beta version of Firefly:
“The true power of Firefly,” Clark tells me, “lies in its seamless integration with Photoshop. It’s a perfect fit for my workflow, unlike Dall-E and Midjourney, which insist on using their own web programs. Plus, it’s free from any resolution constraints – a common limitation with other AI generative programs. Firefly adapts to the quality and resolution of the image I’m working on.”
What else? “The key advantage,” Clark continues, “is that you can start with a base image and build off it. No need to tell Firefly everything at once. It’s a sharp contrast to the burgeoning industries built around Midjourney prompt engineering, which demand you meticulously construct a prompt covering everything from style to composition to lighting, and so on. With Firefly, I can gradually construct an image – a prompt here, another there, maybe a stock element or an original illustration. I can then ask Firefly to blend all these elements together or add to what I’ve created in specific parts of the image.”
Clark’s final verdict? “The best part about using Firefly is that I still feel like an artist, not just someone barking orders at an AI. Firefly feels more like a collaborator, assisting me instead of hijacking the entire process. Maybe I’m fooling myself, but as it’s just another tool in Photoshop, I feel like I’m leveraging it to augment my work, not handing over full creative control to the AI.”
