Adobe
Follow ZDNET: Add us as a preferred source on Google.
ZDNET's key takeaways
Adobe has launched new AI image-generating tools at Adobe Max.
Adobe Firefly Image Model 5 is the company's most capable model.
The Prompt to Edit feature lets you edit images using natural language.
OpenAI's launch of DALL-E 2 in 2022 ignited an AI text-to-image generator craze, with dozens of tools mlaunching during the past few years. Yet, the use cases for these image generators remain fairly narrow, and they lack everyday, realistic applications for the average person. Adobe's latest image model wants to change that issue.
Also: The most popular AI image and video generator might surprise you - here's why people use it
At its annual Adobe Max creativity conference, the company unveiled its new Adobe Firefly Image Model 5, which the company describes as its most advanced image generation and editing model yet. Beyond creating high-quality images, the model can also help you edit your existing pictures using AI. After watching demos of the editing features, I think the model solves a significant issue for users.
Adobe Firefly Image Model 5
The new model can now generate images in native 4MP resolution, which have more context and have almost twice as many pixels as 1080p, resulting in finer details. Adobe boasts that the results are high-quality images with "photorealistic" results.
Also: Why open source may not survive the rise of generative AI
Adobe says this attention to detail means the model can tackle challenging tasks, such as anatomically accurate portraits, natural movement, and multi-layered compositions. The real magic of this approach comes to life in the new photo-editing features.
Photo editing made easier
Sometimes you'll take a shot that has potential but needs a couple of tweaks to be perfect. Whether the tweaks are simple or complex, they often require you to become familiar with editing software and to click the correct buttons to make adjustments, leading to a more complicated and time-consuming process. With Adobe's new tool, you can just ask the AI to make edits, and the work is done.
As the name implies, the new Prompt to Edit tool, powered by Firefly Image Model 5, lets you use a conversational prompt to have an action performed on your photo. I got to see a live demo of the feature in action before the release and was pretty amazed.
In the demo, the person uploaded an image of her dog sitting behind a fence. Then she asked the tool to remove the fence from the picture. Within seconds, the fence was removed from the photo of the dog, with AI filling in the spaces where the item previously stood to produce a realistic-looking image. Google launched a similar feature, called Edit with Ask Photos, with the launch of the Google Pixel 10 earlier this year, and I found that tool just as handy.
Also: Are Sora 2 and other AI video tools risky to use? Here's what a legal scholar says
Building on this feature is Layered Image Editing, which maintains the image's composition accuracy as you make adjustments to its elements. For example, in the sizzle video, numerous items are displayed on a surface, but the user can then select one, drag and drop it, resize it, and even tweak it with a prompt. This task is typically a very complex process in Photoshop that requires multiple tools, steps, and precision.
How to access
The Prompt to Edit feature is now available to Firefly customers, supporting Firefly Image Model 5 and partner models from Black Forest Labs, Google, and OpenAI. Firefly Image Model 5 is available in beta today. Lastly, while shown at Adobe Max, the Layered Image Editing feature is still in development.