Adobe Photoshop and Firefly AI Upgrades Bring New Features, Unintended Laughter
Adobe’s ongoing efforts to enhance its image editing suite with AI-driven tools have taken a step forward with the introduction of new features and improved performance. However, while these upgrades promise greater functionality, they also come with the occasional humorous mishap.
The latest addition to Adobe’s AI arsenal is Firefly Image 3, now available in beta form within Photoshop and the Firefly web app. This update aims to refine the performance of Adobe’s generative AI tools and introduce novel methods for image creation and editing.
Initially launched last year, Firefly employs AI algorithms to manipulate images, allowing users to add or remove objects, adjust surfaces, and expand images to different aspect ratios, among other functions. With Firefly Image 3, Adobe has prioritized improvements in style and consistency. Users now have access to a broader range of artistic styles and can instruct the software to generate images that mimic a specific reference style.
Notably, the new version of Firefly boasts enhanced lighting accuracy in realistic images and an improved ability to interpret longer and more complex prompts.
In Photoshop, the generative fill tool, popularized by users creatively expanding iconic album covers, receives several enhancements courtesy of Firefly Image 3. For the first time, users can employ text-to-image generation and create backgrounds for existing images. Additionally, they can refine image details for sharper results and explore similar output variations.
Despite Adobe’s demonstration video showcasing seamless integration, real-world usage occasionally yields unexpected outcomes. Developer Nina Nikolic’s attempt to use the generative fill tool to rectify a cropped shoulder in her photo resulted in the generation of an extra person – a naked man – alongside her image. Another endeavor produced a comically distorted giant face.
In related news, Adobe researchers have unveiled their work on VideoGigaGAN, an AI-based video super-resolution tool aimed at enhancing low-resolution footage. This technology, akin to Nvidia’s RTX Video Super Resolution, promises clearer details and less blur on high-resolution displays, marking another stride in Adobe’s AI-driven innovation journey.
